biomedical research laboratory meaning

Laboratory x-ray nano-computed tomography for biomedical research

High-resolution x-ray tomography is a common technique for biomedical research using synchrotron sources. With advancements in laboratory x-ray sources, an increasing number of experiments can be performed in the lab. In this paper, the design, implementation, and verification of a laboratory setup for x-ray nano-computed tomography is presented using a nano-focus x-ray source and high geometric magnification not requiring any optical elements. Comparing a scintillator-based detector to a photon counting detector shows a clear benefit of using photon counting detectors for these applications, where the flux of the x-ray source is limited and samples have low contrast. Sample contrast is enhanced using propagation-based phase contrast. The resolution of the system is verified using 2D resolution charts and using Fourier Ring Correlation on reconstructed CT slices. Evaluating noise and contrast highlights the benefits of photon counting detectors and the contrast improvement through phase contrast. The implemented setup is capable of reaching sub-micron resolution and satisfying contrast in biological samples, like paraffin embedded tissue.

1 Introduction

High-resolution x-ray tomography has become a widely used tool in biomedical research. While synchrotron imaging is the gold standard allowing for fast scans with high resolution and with excellent contrast through the use of propagation-based phase contrast imaging (PB-PCI) [ Snigirev et al., 1995 ] or x-ray holography [ Cloetens et al., 1999 ] , laboratory setups are becoming viable alternatives. With developments of brighter micro-focus sources [ Hemberg et al., 2003 ] and nano-focus sources [ Nachtrab et al., 2014 , Müller et al., 2017 , Ferstl et al., 2018 ] , an increasing amount of biomedical imaging applications are becoming feasible in the laboratory. Investigations of tissue samples are often performed with resolution of around 1 µm or with sub-micron resolution down to a few hundred nm. The setup presented in this paper can achieve sub-micron resolution in biological samples, such as tissue samples, with sufficient contrast to enable biomedical studies.

There are a variety of methods to achieve high-resolution in x-ray tomography: incorporating visible light optics into a scintillator-based detector [ Stampanoni et al., 2006 ] , x-ray optics [ Niemann et al., 1976 , Withers, 2007 , Robinson and Harder, 2009 ] to high geometric magnifications using small x-ray spots [ Müller et al., 2017 , Fella et al., 2018 , Eckermann et al., 2020 ] , and techniques like ptychography [ Rodenburg et al., 2007 ] . Using visible light optics to achieve high resolution is a common approach at synchrotron sources [ Stampanoni et al., 2006 ] and also in laboratory imaging [ Töpperwien et al., 2016 , Dierks et al., 2022 , Rashidi et al., 2020 ] . Alternatively, x-ray optics have been used to achieve even higher resolutions [ Fella et al., 2017 , Jacobsen, 2019 ] . However, such optics-based approaches typically require monochromatic and highly coherent x-rays [ Niemann et al., 1976 ] . Lens-less approaches like ptychography [ Rodenburg et al., 2007 , Thibault et al., 2008 , Pfeiffer, 2018 ] can provide extremely high resolution not limited by the spot size given coherent illumination. Recent developments are working towards making ptychography available in the laboratory as well [ Batey et al., 2021 ] .

While x-ray optics require quasi-monochromatic illumination and are only widely available for lower x-ray energies [ Müller et al., 2021 ] and magnifying detectors utilise low-efficiency scintillation detectors, high geometric magnification with a small x-ray spot can provide an alternative to achieve very high resolution in a laboratory setup [ Nachtrab et al., 2014 , Fella et al., 2018 ] . Such setups are very similar to established micro-CT systems, but utilise transmission x-ray sources (as opposed to reflection-type x-ray sources), allowing us to place the sample significantly closer to the x-ray source (closer than 1 mm) resulting in very high magnification factors while retaining a compact system [ Fella et al., 2018 ] . This has clear advantages for higher energy applications as shown by Graetz et al [ Graetz et al., 2021 ] and Müller et al [ Müller et al., 2021 , Müller et al., 2022 ] . Studies on biomedical samples have also shown that laboratory nano-tomography can provide sufficient data quality for medical studies [ Ferstl et al., 2018 , Eckermann et al., 2020 , Romell et al., 2021 ] .

Building a laboratory nano-tomography imaging system comes with a variety of challenges and requirements to the equipment. Thermal and mechanical stability are a main concern, which includes the x-ray source, stages, detector, and the environment of the system. Another concern is the scan quality, i.e. the achievable contrast and scan times required for samples consisting of low-Z materials. The following section discusses how the requirements and parameters were taken into account to build the imaging system.

2 Methodology

2.1 requirements for biomedical nano-ct.

In low-density samples, such as soft tissue, high contrast is best achieved with low energy x-ray photons. With samples that are paraffin embedded and small enough to not absorb too much of the lower energy photons, good soft tissue contrast can be achieved using propagation-based phase contrast [ Snigirev et al., 1995 , Wilkins et al., 1996 ] . Thus the sample size should be minimised, as well as the distance to the detector reducing absorption in air. Denser parts of the sample might lead to beam hardening artefacts. Conventional staining agents are typically very dense compared to tissue. Hence, no staining should be performed ideally, or appropriate contrast agents should be selected [ Dreier et al., 2022 ] .

The biological tissue samples used as test objects in this study were fixated and embedded in paraffin. Sample preparation of the atherosclerotic plaque is described in [ Truong et al., 2022 ] , whereas the procedure for the cow lung injected with dye is described in [ Dreier et al., 2022 ] .

subscript 𝑅 1 subscript 𝑅 2 z_{\text{eff}}=R_{2}/M=R_{1}R_{2}/(R_{1}+R_{2}) italic_z start_POSTSUBSCRIPT eff end_POSTSUBSCRIPT = italic_R start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT / italic_M = italic_R start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT italic_R start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT / ( italic_R start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT + italic_R start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT ) . Evaluation and optimisation of phase fringes was performed by Dierks et al. [ Dierks et al., 2023 ] .

Another requirement is a stable x-ray spot from the source, both in size and position. Drift or defocusing of the x-ray spot will lead to a loss of resolution. The whole imaging system is affected by thermal drift leading to slow movements of the equipment during the measurements, which can be corrected using alignment images, as described in Section 2.3 . Vibrations are much more difficult to correct, particularly in laboratory setups with long exposure times, and should be avoided through careful design of the setup. However, vibrations cannot always be excluded completely and can be caused by instabilities of the equipment, i.e. precision of the motion axes, air flow, or other equipment nearby utilising a pump or cooling fan. Vibrations that can be resolved, i.e. when the frequency is sufficiently low while the exposure time of the detector is sufficiently short, can be corrected using algorithms like joint reprojection [ Gürsoy et al., 2017 ] , phase symmetry [ Pande et al., 2022 ] , or distributed optimisation [ Nikitin et al., 2021 ] .

To improve contrast in low density samples, such as soft tissue, propagation-based phase contrast [ Snigirev et al., 1995 , Bravin et al., 2013 ] is often utilised. Propagation-based phase contrast considers illumination of a sample with a wave, which is perturbed by interfaces between materials with different refraction indices. Given a sufficiently large propagation distance, phase shifts are detected as intensity variations on the detector. Considering illumination with a magnifying cone beam, as in any optics-free laboratory x-ray imaging system, the geometric magnification has to be taken into account to find the effective propagation distance z eff = R 2 / M subscript 𝑧 eff subscript 𝑅 2 𝑀 z_{\text{eff}}=R_{2}/M italic_z start_POSTSUBSCRIPT eff end_POSTSUBSCRIPT = italic_R start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT / italic_M . Since reducing M 𝑀 M italic_M up to the point where R 1 = R 2 subscript 𝑅 1 subscript 𝑅 2 R_{1}=R_{2} italic_R start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT = italic_R start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT will increase z eff subscript 𝑧 eff z_{\text{eff}} italic_z start_POSTSUBSCRIPT eff end_POSTSUBSCRIPT while reducing the effective pixel size p eff = P / M subscript 𝑝 eff 𝑃 𝑀 p_{\text{eff}}=P/M italic_p start_POSTSUBSCRIPT eff end_POSTSUBSCRIPT = italic_P / italic_M limiting the achievable resolution. Thus, optimising z eff subscript 𝑧 eff z_{\text{eff}} italic_z start_POSTSUBSCRIPT eff end_POSTSUBSCRIPT and p eff subscript 𝑝 eff p_{\text{eff}} italic_p start_POSTSUBSCRIPT eff end_POSTSUBSCRIPT might require to increase R 1 subscript 𝑅 1 R_{1} italic_R start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT and R 2 subscript 𝑅 2 R_{2} italic_R start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT and thus absorption in air, resulting in longer scan times increasing effects from thermal drifts.

A variety of phase-retrieval algorithms have been discussed in detail by Burvall et al [ Burvall et al., 2011 , Burvall et al., 2013 ] , a general summary for x-ray imaging has been provided by Nugent et al [ Nugent, 2007 ] , and practical considerations have been described by Gureyev et al [ Gureyev et al., 2009 ] . A common approach is single material phase-retrieval [ Paganin et al., 2002 ] . Another algorithm working well for laboratory data is the Bronnikov Aided Correction (BAC) [ De Witte et al., 2007 ] .

Since resolution is achieved through a small x-ray spot combined with high geometric magnification, R 1 subscript 𝑅 1 R_{1} italic_R start_POSTSUBSCRIPT 1 end_POSTSUBSCRIPT and R 2 subscript 𝑅 2 R_{2} italic_R start_POSTSUBSCRIPT 2 end_POSTSUBSCRIPT have to be selected to yield a sufficiently small effective pixel size p eff subscript 𝑝 eff p_{\text{eff}} italic_p start_POSTSUBSCRIPT eff end_POSTSUBSCRIPT . Further, it is beneficial to reduce the total length of the setup to minimise absorption in air while maximising photons per area on the sample. Hence, there are two detector parameters to consider: physical pixel size and detection efficiency.

Common x-ray cameras are sCMOS (scientific Complementary Metal–Oxide–Semiconductor) detectors. These detectors use a scintillator to convert incoming x-rays into visible light, which is then detected using a sCMOS camera sensor. One of their main advantages is that small pixels (down to a few µm) can be achieved. However, the point-spread function (PSF) of the camera sensor is typically a few pixels large, i.e. visible light photons from a single scintillation event are detected in several neighbouring pixels resulting in blurring of the image. Additionally, the scintillator material and thickness also might cause blurring and affect the detection efficiency.

To achieve high resolution at low x-ray energies, the scintillator has to be as thin as possible to minimise blurring while being as efficient as possible for the relevant energies. A material such as Gadolinium Oxysulfide (Gadox) can be produced very thin, down to a few µm, with decent efficiency at around 10 keV allowing high resolution at lower energies. Besides blurring from the scintillator and PSF of the detector, noise has to be considered as well. Readout noise and electronic noise are a significant concern for photon-limited applications.

As an alternative to scintilllation based cameras, photon counting detectors provide a significantly smaller PSF close to a single pixel [ Rossi et al., 2006 ] , lower noise, and higher efficiency at the expense of larger pixels making them a viable alternative for biomedical and high resolution imaging [ Bech et al., 2008 , Flenner et al., 2023 ] . Photon counting detectors detect x-ray photons directly in a semiconductor sensor. Every pixel has its own dedicated electronics allowing to suppress noise and reduce cross talk between pixels yielding an almost box-like PSF covering a single pixel [ Johnson et al., 2014 ] . Using a silicon sensor yields almost 100 % efficiency for 10 keV photons dropping with increasing photon energy [ Donath et al., 2013 , Donath et al., 2023 ] . A comparison between the theoretical and practical efficiency of a Pilatus detector [ Henrich et al., 2009 ] has been shown by Donath et al [ Donath et al., 2013 ] . These properties make photon counting detectors a viable choice for imaging of low density samples especially with limited flux as demonstrated by Scholz et al [ Scholz et al., 2020 ] and Dudak et al [ Dudak et al., 2022 ] or for high resolution as shown by Flenner et al [ Flenner et al., 2023 ] . However, their main limitation for this particular application is the pixel size (75 µm for the Eiger detector implemented in our work), which is significantly larger than those of sCMOS cameras. It should be noted, however, that pixel size cannot be directly compared between scintillator-based and photon-counting detector due to their different PSFs. Using a photon-counting detector, equivalent resolution can be achieved with a larger p eff subscript 𝑝 eff p_{\text{eff}} italic_p start_POSTSUBSCRIPT eff end_POSTSUBSCRIPT , thus reducing the required distance, compared to a sCMOS detector.

2.2 Setup design

In the setup (Figure 1 A), the following coordinate system is used: z represents the horizontal axis along the beam from the source towards the detector, x represents the perpendicular horizontal axis, and y represents the vertical axis. Tilt is represented by u perpendicular to the beam and v along beam direction. Rotation around the y 𝑦 y italic_y axis is referred to as ry . Another coordinate system on top of the rotation axis have axes referred to as fx and fz , coinciding with the x 𝑥 x italic_x and z 𝑧 z italic_z axes respectively with the rotation ry at 0 degrees.

The setup utilises a NanoTube N2 60 kV nano-focus x-ray source (Excillum AB, Sweden), which can achieve spot sizes from 1.2 µm down to 300 nm. The acceleration voltage can be set between 40 and 60 kV. The source utilises a patterned tungsten on diamond transmission target, which is used to calibrate the desired spot size, internally verify the spot size, spot position on the target, and sets the maximum safe power automatically. The main energy emitted is around 8 keV (Tungsten L α subscript 𝐿 𝛼 L_{\alpha} italic_L start_POSTSUBSCRIPT italic_α end_POSTSUBSCRIPT ), matching well to the efficiency of the detectors.

A 9-axis sample holder is constructed from a H-811 hexapod (Physik Intrumente GmbH, Germany) allowing translation in x, y, z as well as tilting along and perpendicular to the beam direction ( u, v ). The hexapod is used for coarse positioning of the sample, i.e. height and centering of the rotation axis, and tilt correction of the rotation axis with a maximum range of ± plus-or-minus \pm ± 16 mm in x,z and ± plus-or-minus \pm ± 6.5 mm in y direction. On top, a RT100S air bearing rotation stage (LAB Motion Systems, Belgium) is mounted providing rotation around the y axis. The stage was chosen for its positional accuracy, stability, and position repeatability. For fine positioning on top of the rotation axis, a U723 dual-axis piezo stage (Physik Instrumente GmbH, Germany) with ±11 mm range in x, z direction and a positional accuracy of 10 nm, allowing centering of the sample or feature on the rotation axis. The distance from the x-ray spot to the rotation axis of the sample (source-object distance) is particularly critical for the achievable resolution. Slight differences can have significant impact on p eff subscript 𝑝 eff p_{\text{eff}} italic_p start_POSTSUBSCRIPT eff end_POSTSUBSCRIPT .

The 3-axis detector positioning system consists of three linear stages (Owis GmbH, Germany) allowing for detector alignment in x, y, z direction. A long-range dz stage with 0.6 m range was chosen to be able to select a wide range of magnifications. Using the dx, dy stages, the detector can be aligned to the optical axis of the system. The two detectors are shown in Figure 1 B–C.

The top of the sample holder is detachable using a KB25 kinematic base (ThorLabs Inc., USA). Different holders were designed to allow a variety of sample types and sizes. As primary mounting approach polyimide tubes were selected, which can be glued into a variety of 3D printed pins. Larger or less fragile sample can be glued directly to carbon fibre tubes (2 mm outer diameter, 1 mm inner diameter) using wax (paraffin wax or beeswax) or superglue. For small polyimide tubes, self-aligning 3D printed pins with a conical base [ Holler et al., 2017 ] are used. Larger polyimide tubes and carbon fibre tubes are glued into 3D printed 6 mm diameter pins, which are attached to the kinematic base on top of the piezo stage using a collet knob.

To avoid collisions of the sample with the source, a Zelux 1.6 MP CMOS high-resolution camera (ThorLabs Inc., USA) with a MVL50M23 lens (ThorLabs Inc., USA) is placed to the side monitoring the tip of the source. For a general overview, a M5055 PTZ (pan, tilt, zoom) camera is placed above the setup (AXIS Communications AB, Sweden).

Two detectors are available in the setup, a photon counting Eiger 2R 500K (Dectris Ltd., Switzerland) and the sCMOS GSense 16M (Photonic Science Ltd., UK) with 9 µm pixel size, a 4096 × \times × 4096 pixel sensor, and a 17.5 µm Gadox scintillator. The Eiger 2R 500K detector has a 450 µm Silicon sensor with a pixel size of 75 µm, and 1028 × \times × 514 pixels.

To achieve sufficient stability, a vibration-free and thermally stable environment is required. To reduce vibrations, e.g. from surrounding equipment, the setup was built on an optical table with air-cushioned feet. The experimental environment needs to be able to achieve thermal equilibrium to avoid drifts of the equipment itself through thermal expansion or contraction. Airflow needs to be minimised to avoid instabilities of, particularly, the sample. Aluminium has a fairly high thermal expansion coefficient compared to other materials, i.e. it causes larger drifts when the temperature changes. It should also be noted that a lot of equipment is built from Aluminium, thus thermal stability needs to be taken into account when selecting components.

2.3 Data acquisition

To align a sample for a scan, first the Centre-of-Rotation (CoR) is centred on the detector. Secondly, the sample or feature of interest is centred on the detector using the precise axes on top of the rotation axis. At 0 degrees sample rotation, the fx axis is used, at 90 degrees sample rotation the fz axis is used.

Before starting the scan, the source is calibrated to the desired spot size and acceleration voltage. Then, the selected acquisition parameters are entered into a Python script, which will execute the scan. A measurement typically consists of flat-fields, a pre-alignment scan, the actual CT scan, and a post-alignment scan. With the Eiger detector, it is sufficient to acquire flat-fields before the scan, both, the silicon sensor and emission from the source, are sufficiently stable. For the GSense detector, dark-field and flat-field images are acquired before the scan using the detector software. To later correct for drifts during the scan, pre- and post-alignment images are acquired, i.e. the sample is rotated in e.g. 10 or 20 degree steps, where at each step a projection is acquired.

2.4 Data processing

The acquired projections, flat-field images, and alignment images are loaded and broken, noisy, or outlier pixels are removed. The Eiger detector has an internal mask, setting bad pixels to a fixed value. Other outliers, or outliers in images acquired with the GSense detector, can be identified by, e.g. thresholding pixels above or below a certain value to consider them noisy or dead, or by identifying pixels that deviate a certain number of standard deviations from the image mean. To replace the identified bad pixels, values from a median filtered version of the same image are inserted.

The Centre-of-Rotation (CoR) is found by performing multiple reconstructions of the centre slice, using a fan beam reconstruction, with varying shifts along the x 𝑥 x italic_x axis, which are then evaluated visually. This approach has proven to be the most reliable to find the CoR. With a first estimate for the CoR, ring filtering can be performed. A slice is reconstructed with different ring filters and parameters. Several filters are implemented as described in [ Vo et al., 2018 , Münch et al., 2009 ] . After selecting a ring filter and fine tuning its parameters, the filter is applied to all projections.

Following the ring filter, drift correction can be performed. It is necessary to first perform ring filtering since these filters work by identifying and removing stripes from the sinograms, which may be deformed by applying the drift correction. A cross correlation [ Guizar-Sicairos et al., 2008 ] between the alignment images and the projections is performed yielding shifts in x 𝑥 x italic_x and y 𝑦 y italic_y direction. To obtain shifts for each projection, linear interpolation is performed. Corrections corresponding to these shifts are then applied to all projections.

1 𝛿 𝑖 𝛽 n=1-\delta+i\beta italic_n = 1 - italic_δ + italic_i italic_β . The effect on a single projection is estimated, which is inspected visually for removal of phase fringes at material borders and blurring caused by the algorithm. Following, an unsharp mask [ Sheppard et al., 2004 , Paganin et al., 2020 ] can be applied to compensate some of the blurring caused by phase retrieval. With parameters selected, all projections are processed.

Reconstructions are performed using a GPU implementation of the Feldkamp-Davis-Kress (FDK) algorithm [ Feldkamp et al., 1984 ] , an efficient implementation of filtered back-projection for cone beam geometry, via the ASTRA toolbox [ van Aarle et al., 2015 , van Aarle et al., 2016 ] . Fine tuning of shifts and tilts utilises single slice FDK reconstructions. To correct tilts of the rotation axis perpendicular ( u ) and along the beam ( v ), single slices for a range of values are reconstructed using the FDK algorithm and inspected visually. If a drift correction was performed or if a tilt correction is necessary, the CoR needs to be fine-tuned using the same approach. With all shifts and tilts resolved, the full volume is reconstructed.

2.5 Evaluation of resolution and image quality

Measuring the performance of an imaging system in 2D using resolution charts is a standard approach. These charts consist of differently sized and shaped metal structures (often Tungsten or Gold) on a thin membrane. Thus, their contrast is comparably high particularly when compared to biomedical samples. Commonly, line patterns are used to determine the achievable resolution. The charts used in this paper were placed approximately 200 µm from the x-ray source.

To evaluate the resolution of reconstructed slices, Fourier Ring Correlation (FRC) is used [ van Heel, 1987 ] . Compared to obtaining an Edge Spread Function (ESF) and Modulation Transfer Function (MTF) by fitting a sharp edge as described by slanted-edge method [ Estribeau and Magnan, 2004 ] and previously used in [ Dreier et al., 2021 ] , the FRC takes a full slice into account resulting in a more general resolution estimation. To calculate the FRC, the acquired projections are split and two identical reconstructions are performed. Before correlation, shifts, rotation, and scaling of the selected slices are aligned with sub-pixel precision. The parameters are found by matching features in both slices using an Oriented FAST and Rotated BRIEF (ORB) classifier [ Rublee et al., 2011 ] and removing outliers with an iterative Random Sample Consensus (RANSAC) algorithm [ Fischler and Bolles, 1981 ] . The resulting FRC curve is compared to a threshold criterion [ van Heel and Schatz, 2005 ] . Since the input data is split, the half-bit criterion is an appropriate measure. The last intersection of the FRC curve with the criterion yields the resolution limit of the reconstructed slice.

The resolution was evaluated using three different samples: a piece of cork (roughly 0.5 mm in size), a cylindrical punch fragment from paraffin embedded bovine lung tissue (0.5 mm in diameter), and a cylindrical punch from paraffin embedded human atherosclerotic plaque tissue (0.5 mm in diameter). Scans were performed with both detectors and the scan parameters are summarised in Table 1 .

To assess the image quality, the Signal-to-Noise ratio (SNR) and Contrast-to-Noise ratio (CNR) are calculated. Both methods work without a reference image and instead utilise a region on the sample and on the background area of the image. The SNR is defined as:

where s 𝑠 s italic_s is the mean value of a region on the sample and n RMS subscript 𝑛 RMS n_{\text{RMS}} italic_n start_POSTSUBSCRIPT RMS end_POSTSUBSCRIPT is the root mean square of a region containing only background. The CNR is defined as:

where s 𝑠 s italic_s is the mean value of a region containing the sample, n 𝑛 n italic_n is the mean value of a region containing only background and σ n subscript 𝜎 𝑛 \sigma_{n} italic_σ start_POSTSUBSCRIPT italic_n end_POSTSUBSCRIPT is the standard deviation of the region containing only background as previously described [ Bech et al., 2008 , Dreier et al., 2020 ] .

3 Results and discussion

3.1 mechanical stability.

The temperature in the experimental hutch is monitored with a simple temperature logger and one data point per 5 s is acquired. Figure 2 A shows a temperature measurement over 160 h with an average temperature of 21.5  ∘ C and standard deviation of ± plus-or-minus \pm ± 0.12  ∘ C. Accessing the hutch will cause a change in temperature. Hence, access should be limited in time and frequency to limit the time required to reach thermal equilibrium again. Figure 2 B shows the measured drift during a scan measured using alignment images acquired before and after a scan. Slow drifts in the range of a few µm could be measured, which can typically be corrected without issues.

3.2 Resolution

The 2D performance of the system was evaluated using line pattern on a JIMA RT RC-02 chart and a resolution chart from XRnanotech (Villigen, Switzerland). The smallest available features on the JIMA chart are 0.5 µm and 0.4 µm, which could be resolved by both detectors, as shown for the Eiger detector in Figure 3 A,B,D and for the GSense detector in Figure 3 F,G,J. On the XRnanotech chart, the 0.25 µm feature could be resolved, as shown for the Eiger detector in Figure 3 C,E and the GSense detector in Figure 3 H,K. Profile plots in horizontal and vertical direction over multiple features are shown in Supplementary Figure S2 .

Performance of the system in 3D was evaluated using FRC on three test samples: (i) a 0.5 mm piece of cork, (ii) a 0.5 mm punch fragment from a paraffin embedded cow lung (previously scanned with our micro-CT [ Dreier et al., 2022 ] ), and (iii) a 0.5 mm fragment punched from a human atherosclerotic plaque (previously scanned with synchrotron micro-CT [ Truong et al., 2022 ] ). The resolution is only measured on slices without phase-retrieval applied since phase-retrieval may corrupt the higher frequencies [ Riedel et al., 2023 ] . FRC of the lung punch fragment (Figure 4 F) shows that the achieved resolution is close to p eff subscript 𝑝 eff p_{\text{eff}} italic_p start_POSTSUBSCRIPT eff end_POSTSUBSCRIPT at 1.26 × p eff absent subscript 𝑝 eff \times p_{\text{eff}} × italic_p start_POSTSUBSCRIPT eff end_POSTSUBSCRIPT , while for the plaque punch fragment (Figure 4 I) a resolution of 1.72 × p eff absent subscript 𝑝 eff \times p_{\text{eff}} × italic_p start_POSTSUBSCRIPT eff end_POSTSUBSCRIPT could be achieved. The cork sample shows a resolutions of 1.01 × p eff absent subscript 𝑝 eff \times p_{\text{eff}} × italic_p start_POSTSUBSCRIPT eff end_POSTSUBSCRIPT . Variations may be caused by the precision of the geometry alignment (CoR, tilt) and precision of the drift correction.

When using the GSense detector, the resolution of the three scans (Figure 5 ) was found to be around 2 – 3 × p eff absent subscript 𝑝 eff \times p_{\text{eff}} × italic_p start_POSTSUBSCRIPT eff end_POSTSUBSCRIPT corresponding to the expected PSF stated by the manufacturer. Since the results of the two scans appear consistent with each other, one could assume the cork scan with the Eiger detector shows in fact uncorrected tilts or drifts affecting the resolution.

3.3 Contrast and noise

Evaluating the SNR (Equation 1 ) of the two detectors (Table 2 ) shows that concerning reconstructions without phase-retrieval, the SNR cannot be accurately measured with all measurements yielding around 49 dB. When applying phase-retrieval, the SNR yields a more reasonable number, possibly due to the smoothing, which is caused by phase retrieval. From a visual comparison, the GSense reconstructions look considerably more noisy, as seen in Figure 5 A, D, and G, as compared to the Eiger reconstructions shown in Figure 4 B, E, and H. The used regions for SNR and CNR calculations are shown in Supplementary Figure S1 . Similar regions of interest (ROI) were selected for both detectors. The ROI used on the cork sample was placed on a denser part of the sample. Two ROIs were used on the lung sample, the first ROI was set on a dye filled vessel and the second on the tissue close to the airway in the upper part of the slice. Also, for the plaque sample, two regions were selected, focusing in calcified and non-calcified tissue respectively.

In contrast to the SNR, the CNR (Equation 2 ) yields a more robust measure on noisy images. Two things can be observed: (i) applying phase-retrieval significantly improves the CNR as evident from the reconstructions (Panels B, E in Figures 4 and 5 ) and (ii) the Eiger detector performs significantly better, as expected. Hence, the CNR appears to be an appropriate measure to judge the improvement through phase contrast when dealing with low absorbing samples and generally noisy images.

4 Conclusions

In this paper, we have presented the design and implementation of an x-ray nano-CT setup for biological samples capable of reaching sub-micron resolution and satisfying contrast. Data processing, including alignment, drift correction, and tilt correction, were described. Further, we have shown that the contrast can be increased by utilising propagation-based phase contrast while retaining the resolution in 3D through incorporation of unsharp masking into the phase-retrieval workflow.

The presented experiments show that, for biological applications with low contrast samples and low flux sources, photon counting detectors are the preferred choice. Resolutions close to the ideal single-pixel PSF of the detector could be achieved. Due to their high efficiency around 10 keV, the main energy in the x-ray spectrum, the photon counting detector performed significantly better than the scintillator-based detector despite the much larger distances required due to its larger pixels.

It has been shown that the setup is capable of producing CT scans that are somewhat comparable to synchrotron scans, all though at significantly longer exposure times and requiring the sample to be very close to the source, i.e. typically a biopsy punch. Typically, tissue samples are embedded into larger paraffin blocks on which ROI scans are performed at a synchrotron. A downside of using an x-ray CT setup, as described in this paper, is the required sample size. To achieve high resolution, the sample has to be as close as possible, while keeping the whole system as compact as possible. Many features in e.g. tissue samples do not require the theoretically maximum achievable resolution of the x-ray source at 150 nm, but rather a larger FOV at relaxed resolution. The main limiting factor in the current system is the detector size, a wider detector would allow to scan larger samples at higher resolution.

Ethical Approval

The local Swedish Ethical Review Authority approved the use of the human atherosclerotic plaque samples (diary number 472/2005). The bovine lung samples were obtained from a local slaughterhouse and approved for research use by the local Swedish authorities (Jordbruksverket) under diary number 6.7.18-16885/2021.

Acknowledgements

This project has been supported by the Swedish Foundation for Strategic Research (SSF grant ID17-0097) and by the Swedish research council (VR grant 2022-04192). The project has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (Grant agreement No. 101089334). Further, the authors acknowledge infrastructure funding from the Faculty of Science at Lund University. IG acknowledges support from the Swedish Research Council (2019-01260; 2023-02368), Swedish Heart and Lung foundation, and the Swedish Research Council, Strategic Research Area Exodiab, Dnr 2009-1039 (as EXODIAB member). Jesper Wallentin and Hanna Dierks (Synchrotron Radiation Research, Lund University) are acknowledged for input and discussions during the design of the setup. Daniel Nilsson, Daniel Larsson, and Tomi Tuohimaa (Excillum AB) are acknowledged for their help and guidance using the source.

• [Batey et al., 2021] Batey, D. J., Van Assche, F., Vanheule, S., Boone, M. N., Parnell, A. J., Mykhaylyk, O. O., Rau, C., and Cipiccia, S. (2021). X-Ray Ptychography with a Laboratory Source. Physical Review Letters , 126(19):193902.
• [Bech et al., 2008] Bech, M., Bunk, O., David, C., Kraft, P., Brönnimann, C., Eikenberry, E. F., and Pfeiffer, F. (2008). X-ray imaging with the PILATUS 100k detector. Applied Radiation and Isotopes , 66(4):474–478.
• [Bravin et al., 2013] Bravin, A., Coan, P., and Suortti, P. (2013). X-ray phase-contrast imaging: from pre-clinical applications towards clinics. Physics in Medicine and Biology , 58(1):R1–R35.
• [Burvall et al., 2013] Burvall, A., Larsson, D. H., Lundström, U., Stig, F., Hallström, S., and Hertz, H. M. (2013). Phase-retrieval methods with applications in composite-material tomography. Journal of Physics: Conference Series , 463(1):012015.
• [Burvall et al., 2011] Burvall, A., Lundström, U., Takman, P. A. C., Larsson, D. H., and Hertz, H. M. (2011). Phase retrieval in X-ray phase-contrast imaging suitable for tomography. Optics Express , 19(11):10359.
• [Cloetens et al., 1999] Cloetens, P., Ludwig, W., Baruchel, J., Van Dyck, D., Van Landuyt, J., Guigay, J. P., and Schlenker, M. (1999). Holotomography: Quantitative phase tomography with micrometer resolution using hard synchrotron radiation x rays. Applied Physics Letters , 75(19):2912–2914.
• [De Witte et al., 2007] De Witte, Y., Boone, M., Vlassenbroeck, J., and Dierick, M. (2007). The Bronnikov Aided Correction for X-ray Computed Tomography. 710:239–244.
• [Dierks et al., 2023] Dierks, H., Dreier, T., Krüger, R., Bech, M., and Wallentin, J. (2023). Optimization of phase contrast imaging with a nano-focus x-ray tube. Applied Optics , 62:5502.
• [Dierks et al., 2022] Dierks, H., Stjärneblad, P., and Wallentin, J. (2022). A versatile laboratory setup for high resolution X-ray phase contrast tomography and scintillator characterization. Journal of X-Ray Science and Technology , 54(3):1–12.
• [Donath et al., 2013] Donath, T., Brandstetter, S., Cibik, L., Commichau, S., Hofer, P., Krumrey, M., Lüthi, B., Marggraf, S., Müller, P., Schneebeli, M., Schulze-Briese, C., and Wernecke, J. (2013). Characterization of the PILATUS photon-counting pixel detector for X-ray energies from 1.75 keV to 60 keV. Journal of Physics: Conference Series , 425(6):062001.
• [Donath et al., 2023] Donath, T., Šišak Jung, D., Burian, M., Radicci, V., Zambon, P., Fitch, A. N., Dejoie, C., Zhang, B., Ruat, M., Hanfland, M., Kewish, C. M., Riessen, G. A. V., Naumenko, D., Amenitsch, H., Bourenkov, G., Bricogne, G., Chari, A., and Schulze-Briese, C. (2023). Eiger2 hybrid-photon-counting x-ray detectors for advanced synchrotron diffraction experiments. Journal of Synchrotron Radiation , 30:723–738.
• [Dreier et al., 2022] Dreier, T., Bernström, G., Ganji, S., Norvik, C., Tran-Lundmark, K., and Bech, M. (2022). Radiopaque dyes allow vessel imaging in lung tissue using laboratory phase contrast micro-CT. In Müller, B. and Wang, G., editors, Developments in X-Ray Tomography XIV , volume 12242, pages 1224203–1 – 1224203–7, San Diego, California. SPIE.
• [Dreier et al., 2020] Dreier, T., Lundström, U., and Bech, M. (2020). Super-resolution X-ray imaging with hybrid pixel detectors using electromagnetic source stepping. Journal of Instrumentation , 15(3):C03002.
• [Dreier et al., 2021] Dreier, T., Peruzzi, N., Lundström, U., and Bech, M. (2021). Improved resolution in x-ray tomography by super-resolution. Applied Optics , 60(20):5783–5794.
• [Dudak et al., 2022] Dudak, J., Zemlicka, J., Mrzilkova, J., Zach, P., and Holcova, K. (2022). Applicability of Large-Area Single-Photon Counting Detectors Timepix for High-Resolution and High-Contrast X-Ray Imaging of Biological Samples. IEEE Transactions on Nuclear Science , 69(4):753–760.
• [Eckermann et al., 2020] Eckermann, M., Töpperwien, M., Robisch, A.-L., van der Meer, F., Stadelmann, C., and Salditt, T. (2020). Phase-contrast x-ray tomography of neuronal tissue at laboratory sources with submicron resolution. Journal of Medical Imaging , 7(01):1.
• [Estribeau and Magnan, 2004] Estribeau, M. and Magnan, P. (2004). Fast MTF measurement of CMOS imagers using ISO 12333 slanted-edge methodology. In Chatard, J.-P. and Dennis, P. N. J., editors, Detectors and Associated Signal Processing , volume 5251, pages 243–252.
• [Feldkamp et al., 1984] Feldkamp, L., Davis, L. C., and Kress, J. W. (1984). Practical cone-beam algorithm. Journal of the Optical Society of America A , 1(6):612.
• [Fella et al., 2017] Fella, C., Balles, A., Hanke, R., Last, A., and Zabler, S. (2017). Hybrid setup for micro- and nano-computed tomography in the hard X-ray range. Review of Scientific Instruments , 88(12):123702.
• [Fella et al., 2018] Fella, C., Dittmann, J., Muller, D., Donath, T., Murer, D., Tuohimaa, T., Sofiienko, A., Zabler, S., and Hanke, R. (2018). Implementation of a Computed Tomography System based on a laboratory-based nanofocus X-ray source. Microscopy and Microanalysis , 24(S2):236–237.
• [Ferstl et al., 2018] Ferstl, S., Metscher, B., Müller, M., Allner, S., Dierolf, M., Busse, M., Achterhold, K., Gleich, B., and Pfeiffer, F. (2018). Laboratory-based X-ray NanoCT Explores Morphology of a Zebrafish Embryo. Microscopy and Microanalysis , 24(S2):184–185.
• [Fischler and Bolles, 1981] Fischler, M. A. and Bolles, R. C. (1981). Random Sample Consensus: A Paradigm for Model Fitting with Applications to Image Analysis and Automated Cartography. Communications of the ACM , 24(6):381–395.
• [Flenner et al., 2023] Flenner, S., Hagemann, J., Wittwer, F., Longo, E., Kubec, A., Rothkirch, A., David, C., Müller, M., and Greving, I. (2023). Hard x-ray full-field nanoimaging using a direct photon-counting detector. Journal of Synchrotron Radiation , 30:1–10. Nano-CT with a photon counting detector.
• [Graetz et al., 2021] Graetz, J., Müller, D., Balles, A., and Fella, C. (2021). Lenseless X-ray nano-tomography down to 150 nm resolution: on the quantification of modulation transfer and focal spot of the lab-based ntCT system. Journal of Instrumentation , 16(01):P01034–P01034.
• [Guizar-Sicairos et al., 2008] Guizar-Sicairos, M., Thurman, S. T., and Fienup, J. R. (2008). Efficient subpixel image registration algorithms. Optics Letters , 33(2):156.
• [Gureyev et al., 2009] Gureyev, T. E., Mayo, S. C., Myers, D. E., Nesterets, Y., Paganin, D. M., Pogany, A., Stevenson, A. W., and Wilkins, S. W. (2009). Refracting Röntgen’s rays: Propagation-based x-ray phase contrast for biomedical imaging. Journal of Applied Physics , 105(10):102005.
• [Gürsoy et al., 2017] Gürsoy, D., Hong, Y. P., He, K., Hujsak, K., Yoo, S., Chen, S., Li, Y., Ge, M., Miller, L. M., Chu, Y. S., De Andrade, V., He, K., Cossairt, O., Katsaggelos, A. K., and Jacobsen, C. (2017). Rapid alignment of nanotomography data using joint iterative reconstruction and reprojection. Scientific Reports , 7(1):11818.
• [Hemberg et al., 2003] Hemberg, O., Otendal, M., and Hertz, H. M. (2003). Liquid-metal-jet anode electron-impact x-ray source. Applied Physics Letters , 83(7):1483–1485.
• [Henrich et al., 2009] Henrich, B., Bergamaschi, A., Broennimann, C., Dinapoli, R., Eikenberry, E., Johnson, I., Kobas, M., Kraft, P., Mozzanica, A., and Schmitt, B. (2009). PILATUS: A single photon counting pixel detector for X-ray applications. Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment , 607(1):247–249.
• [Holler et al., 2017] Holler, M., Raabe, J., Wepf, R., Shahmoradian, S. H., Diaz, A., Sarafimov, B., Lachat, T., Walther, H., and Vitins, M. (2017). OMNY PIN—A versatile sample holder for tomographic measurements at room and cryogenic temperatures. Review of Scientific Instruments , 88(11):113701.
• [Jacobsen, 2019] Jacobsen, C. (2019). X-ray Microscopy . Cambridge University Press.
• [Johnson et al., 2014] Johnson, I., Bergamaschi, A., Billich, H., Cartier, S., Dinapoli, R., Greiffenberg, D., Guizar-Sicairos, M., Henrich, B., Jungmann, J., Mezza, D., Mozzanica, A., Schmitt, B., Shi, X., and Tinti, G. (2014). Eiger: a single-photon counting x-ray detector. Journal of Instrumentation , 9(05):C05032–C05032.
• [Müller et al., 2022] Müller, D., Fella, C., Altmann, F., Graetz, J., Balles, A., Ring, M., and Gambino, J. (2022). Characterization of electrically stressed power device metallization using nano-CT imaging. Microelectronics Reliability , 135(February):114589.
• [Müller et al., 2021] Müller, D., Graetz, J., Balles, A., Stier, S., Hanke, R., and Fella, C. (2021). Laboratory-Based Nano-Computed Tomography and Examples of Its Application in the Field of Materials Research. Crystals , 11(6):677.
• [Müller et al., 2017] Müller, M., Oliveira, I. d. S., Allner, S., Ferstl, S., Bidola, P., Mechlem, K., Fehringer, A., Hehn, L., Dierolf, M., Achterhold, K., Gleich, B., Hammel, J. U., Jahn, H., Mayer, G., and Pfeiffer, F. (2017). Myoanatomy of the velvet worm leg revealed by laboratory-based nanofocus X-ray source tomography. Proceedings of the National Academy of Sciences of the United States of America , 114(47):12378–12383.
• [Münch et al., 2009] Münch, B., Trtik, P., Marone, F., and Stampanoni, M. (2009). Stripe and ring artifact removal with combined wavelet—Fourier filtering. Optics Express , 17(10):8567.
• [Nachtrab et al., 2014] Nachtrab, F., Firsching, M., Uhlmann, N., Speier, C., Takman, P., Tuohimaa, T., Heinzl, C., Kastner, J., Larsson, D. H., Holmberg, A., Berti, G., Krumm, M., and Sauerwein, C. (2014). NanoXCT: development of a laboratory nano-CT system. In Stock, S. R., editor, Developments in X-Ray Tomography IX , page 92120L, San Diego, California. SPIE.
• [Niemann et al., 1976] Niemann, B., Rudolph, D., and Schmahl, G. (1976). X-ray microscopy with synchrotron radiation. Applied Optics , 15(8):1883.
• [Nikitin et al., 2021] Nikitin, V., De Andrade, V., Slyamov, A., Gould, B., Zhang, Y., Sampathkumar, V., Kasthuri, N., Gursoy, D., and De Carlo, F. (2021). Distributed Optimization for Nonrigid Nano-Tomography. IEEE Transactions on Computational Imaging , 7:272–287.
• [Nugent, 2007] Nugent, K. A. (2007). X-ray noninterferometric phase imaging: a unified picture. J. Opt. Soc. Am. A , 24(2):536–547.
• [Paganin et al., 2002] Paganin, D., Mayo, S. C., Gureyev, T. E., Miller, P. R., and Wilkins, S. W. (2002). Simultaneous phase and amplitude extraction from a single defocused image of a homogeneous object. Journal of Microscopy , 206(1):33–40.
• [Paganin et al., 2020] Paganin, D. M., Favre-Nicolin, V., Mirone, A., Rack, A., Villanova, J., Olbinado, M. P., Fernandez, V., da Silva, J. C., and Pelliccia, D. (2020). Boosting spatial resolution by incorporating periodic boundary conditions into single-distance hard-x-ray phase retrieval. Journal of Optics , 22(11):115607.
• [Pande et al., 2022] Pande, K., Donatelli, J. J., Parkinson, D. Y., Yan, H., and Sethian, J. A. (2022). Joint iterative reconstruction and 3D rigid alignment for X-ray tomography. Optics Express , 30(6):8898.
• [Pfeiffer, 2018] Pfeiffer, F. (2018). X-ray ptychography. Nature Photonics , 12(1):9–17.
• [Rashidi et al., 2020] Rashidi, A., Olfatbakhsh, T., Crawford, B., and Milani, A. S. (2020). A Review of Current Challenges and Case Study toward Optimizing Micro-Computed X-Ray Tomography of Carbon Fabric Composites. Materials , 13(16):3606.
• [Riedel et al., 2023] Riedel, M., Taphorn, K., Gustschin, A., Busse, M., Hammel, J. U., Moosmann, J., Beckmann, F., Fischer, F., Thibault, P., and Herzen, J. (2023). Comparing x-ray phase-contrast imaging using a talbot array illuminator to propagation-based imaging for non-homogeneous biomedical samples. Scientific Reports , 13:6996.
• [Robinson and Harder, 2009] Robinson, I. and Harder, R. (2009). Coherent X-ray diffraction imaging of strain at the nanoscale. Nature Materials , 8(4):291–298.
• [Rodenburg et al., 2007] Rodenburg, J. M., Hurst, A. C., Cullis, A. G., Dobson, B. R., Pfeiffer, F., Bunk, O., David, C., Jefimovs, K., and Johnson, I. (2007). Hard-X-Ray Lensless Imaging of Extended Objects. Physical Review Letters , 98(3):034801.
• [Romell et al., 2021] Romell, J., Jie, V. W., Miettinen, A., Baird, E., and Hertz, H. M. (2021). Laboratory phase‐contrast nanotomography of unstained Bombus terrestris compound eyes. Journal of Microscopy , 283(1):29–40.
• [Rossi et al., 2006] Rossi, L., Fischer, P., Rohe, T., and Wermes, N. (2006). Pixel Detectors . Particle Acceleration and Detection. Springer Berlin Heidelberg, Berlin, Heidelberg.
• [Rublee et al., 2011] Rublee, E., Rabaud, V., Konolige, K., and Bradski, G. (2011). ORB: An efficient alternative to SIFT or SURF. In 2011 International Conference on Computer Vision , pages 2564–2571. IEEE.
• [Scholz et al., 2020] Scholz, J., Birnbacher, L., Petrich, C., Riedel, M., Heck, L., Gkoumas, S., Sellerer, T., Achterhold, K., and Herzen, J. (2020). Biomedical x-ray imaging with a GaAs photon-counting detector: A comparative study. APL Photonics , 5(10):106108.
• [Sheppard et al., 2004] Sheppard, A. P., Sok, R. M., and Averdunk, H. (2004). Techniques for image enhancement and segmentation of tomographic images of porous materials. Physica A: Statistical Mechanics and its Applications , 339(1-2):145–151.
• [Snigirev et al., 1995] Snigirev, A., Snigireva, I., Kohn, V., Kuznetsov, S., and Schelokov, I. (1995). On the possibilities of x‐ray phase contrast microimaging by coherent high‐energy synchrotron radiation. Review of Scientific Instruments , 66(12):5486–5492.
• [Stampanoni et al., 2006] Stampanoni, M., Groso, A., Isenegger, A., Mikuljan, G., Chen, Q., Bertrand, A., Henein, S., Betemps, R., Frommherz, U., Böhler, P., Meister, D., Lange, M., and Abela, R. (2006). Trends in synchrotron-based tomographic imaging: the SLS experience. In Bonse, U., editor, Developments in X-Ray Tomography V , volume 6318, page 63180M.
• [Thibault et al., 2008] Thibault, P., Dierolf, M., Menzel, A., Bunk, O., David, C., and Pfeiffer, F. (2008). High-Resolution Scanning X-ray Diffraction Microscopy. Science , 321(5887):379–382.
• [Töpperwien et al., 2016] Töpperwien, M., Krenkel, M., Quade, F., and Salditt, T. (2016). Laboratory-based x-ray phase-contrast tomography enables 3D virtual histology. In Khounsary, A. M. and van Dorssen, G. E., editors, Advances in Laboratory-based X-Ray Sources, Optics, and Applications V , volume 9964, page 99640I.
• [Truong et al., 2022] Truong, M., Dreier, T., Wassélius, J., Sundius, L., Persson, A., Lovric, G., Bonnin, A., Goncalves, I., and Bech, M. (2022). Sub-micrometer morphology of human atherosclerotic plaque revealed by synchrotron radiation-based μ 𝜇 \mu italic_μ CT—A comparison with histology. PLOS ONE , 17(4):e0265598.
• [van Aarle et al., 2016] van Aarle, W., Palenstijn, W. J., Cant, J., Janssens, E., Bleichrodt, F., Dabravolski, A., De Beenhouwer, J., Joost Batenburg, K., and Sijbers, J. (2016). Fast and flexible X-ray tomography using the ASTRA toolbox. Optics Express , 24(22):25129–25148.
• [van Aarle et al., 2015] van Aarle, W., Palenstijn, W. J., De Beenhouwer, J., Altantzis, T., Bals, S., Batenburg, K. J., and Sijbers, J. (2015). The ASTRA Toolbox: A platform for advanced algorithm development in electron tomography. Ultramicroscopy , 157(2015):35–47.
• [van Heel, 1987] van Heel, M. (1987). Similarity measures between images. Ultramicroscopy , 21(1):95–100.
• [van Heel and Schatz, 2005] van Heel, M. and Schatz, M. (2005). Fourier shell correlation threshold criteria. Journal of Structural Biology , 151(3):250–262.
• [Vo et al., 2018] Vo, N. T., Atwood, R. C., and Drakopoulos, M. (2018). Superior techniques for eliminating ring artifacts in X-ray micro-tomography. Optics Express , 26(22):28396.
• [Wilkins et al., 1996] Wilkins, S. W., Gureyev, T. E., Gao, D., Pogany, A., and Stevenson, A. W. (1996). Phase-contrast imaging using polychromatic hard X-rays. Nature , 384(6607):335–338.
• [Withers, 2007] Withers, P. J. (2007). X-ray nanotomography. Materials Today , 10(12):26–34.

Supplementary

The selected regions for SNR and CNR calculations from the different reconstructed slices are shown in Figure S1 . Background regions are marked in blue and are placed on regions containing air. Regions-of-interest (ROIs) are placed on the sample, indicated in red and orange. On the lung sample (Figure S1 B), two regions are set: one on a dye filled vessel, and one on the tissue around an airway in the upper part of the image, respectively. On the plaque sample (Figure S1 C), two regions are set on tissue of high and low density, repectively.

The line profiles over the 500 – 200 nm features on the XRnanotech chart, presented in Figure  S2 , show that the 250 nm features can be resolved with both detectors in both directions with effective pixel sizes of approximately 50 nm and 150 nm for the GSense and Eiger detector respectively.

Warning: The NCBI web site requires JavaScript to function. more...

An official website of the United States government

The .gov means it's official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you're on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings
• Browse Titles

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

National Academy of Medicine; Yamamoto K, Woolley M, Higginbotham E, et al., editors. Transforming Human Health: Celebrating 50 Years of Discovery and Progress. Washington (DC): National Academies Press (US); 2023 Feb 13.

Transforming Human Health: Celebrating 50 Years of Discovery and Progress.

• Hardcopy Version at National Academies Press

1 BIOMEDICAL SCIENCE AND TECHNOLOGY

Depiction of gene-editing technology (iStock ® ).

“What started as a curiosity-driven, fundamental discovery project has now become the breakthrough strategy used by countless researchers working to help improve the human condition.” —Jennifer Doudna, Winner of the 2020 Nobel Prize in Chemistry and NAM/NAS Member

Digitally enhanced 3D magnetic resonance imaging (MRI) scan of a normal human brain (Science Photo Library ® ).

• Neuroscience and Neurological Diseases

Treatment for Parkinson's Disease

A “gps” in the brain, role of the brain in hormone regulation, discovery of opioid receptors in the brain, understanding alzheimer's disease, mapping the olfactory system, navigating neuroplasticity, diagnosing huntington's disease, gene therapy for infants, three-dimensional mapping of the brain.

• Genomics and Gene Regulation

An Unexpected Origin of Cancer

“seeker” cells, reading biological blueprints, the beginnings of biotechnology, tapping the potential of dna replication, sequencing the human genome, the first cloned animal, a revolutionary advance in gene editing, direct-to-consumer genetic testing, patenting dna.

• Infection, Immunity, and Inflammation

Differentiating Self from Non-Self

Understanding the vast diversity of antibodies, infectious proteins, two types of immune help, recognizing microbial invaders, triggering the death of cancer cells, a cell that kills other cells, engineering the immune system to defeat cancer, adoptive cell transfer to treat cancer.

• Metabolism and Bioenergetics

Understanding Photosynthesis

A multifaceted regulatory mechanism, the formation of adenosine triphosphate, rapid-effect steroid hormones, expanding the reach of newborn screening, cancer-causing metabolites, the progression of multifactorial disorders.

• Diagnostics, Therapeutics, and Imaging

3D Imaging Through X-Ray Slices

Advanced drug delivery systems, the origins of magnetic resonance imaging, increasingly specific and sensitive imaging, the first baby conceived in a laboratory, the first artificial heart, a bridge to heart transplantation, robotic assistance in surgery, watching the brain function, reprogramming human body cells, rna vaccines.

• Scientific and Medical Ethics

Research Misconduct Exposed in the Tuskegee Syphilis Study

After tuskegee: putting human subjects research guidelines in place, the birth of clinical medical ethics, bolstering regulation of research, fabrication, falsification, or plagiarism of research, preventing data suppression in clinical trials, added oversight for potentially harmful research, the origin of hela cells, addressing structural racism and health inequities, #metoo and #timesup in science and medicine, genetically modified babies in china.

• Neuroscience and Neurological Diseases: Unlocking the Mysteries of the Brain

Illustration of neural networks in the brain (iStock ® ).

As the nexus among health, behavior, and medicine, the brain is vital in extending human life and enhancing well-being. Intensive study of the cellular and molecular processes that control our thoughts and actions, combined with the use of animal models in neurological research, have produced key insights into how the brain works and the development and treatment of neurological disorders.

Advanced 3D MRI scan of a human brain (Science Photo Library ® ).

Although Parkinson's disease was first described in medical literature in 1817, an effective treatment wasn't available to most patients until the 1970s. Levodopa works by addressing dopamine deficits in the brain thought to be responsible for symptoms such as tremors and stiffness. In 1986, an effective surgical technique known as deep brain stimulation was introduced. In 1997, scientists discovered that a neuronal protein called alpha-synuclein plays both a genetic and neuropathological role in Parkinson's disease, opening a path to new therapies.

Illustration of a deep brain stimulation treatment for Parkinson's disease: A brain pacemaker sends electrical impulses to specific areas of the brain through an electrode implanted below the scalp. Continuous electrical stimulation blocks the signals (more...)

Humans and other mammals can orient themselves in space and remember the route from one place to another thanks to a positioning system in the brain. Nerve cells known as “place cells,” discovered in 1971, form a mental map of a person's surroundings, while “grid cells,” not discovered until 2005, act as a coordinate system that enables precise positioning and pathfinding.

iStock ® .

In 1971, scientists synthesized luteinizing hormone-releasing hormone, a compound produced by the hypothalamus to regulate the pituitary gland's secretion of certain reproductive hormones. Culminating decades of research to understand the relationship between the nervous and endocrine systems, the achievement led to new treatments for infertility and hormone-sensitive cancers.

Illustration of the pituitary gland (upper red) and the thyroid gland (bottom red), part of the endocrine system (iStock ® ).

Opiates have been used to dull pain for thousands of years, but scientists didn't understand how drugs like morphine and heroin affected the brain until 1973. This breakthrough led to the discovery of endorphins and other naturally occurring opiate-like molecules, which dramatically advanced understanding and treatment of pain and substance use disorders.

Illustration of an enkephalin molecule, which is an endorphin and one of the opioid peptides that occurs naturally in the human brain (Science Photo Library ® ).

Within the brains of individuals with Alzheimer's disease, proteins responsible for the formation of plaques were discovered in the 1980s, followed shortly by the identification of genes associated with inherited and idiopathic forms of the disease. Cognex, the first drug to treat the memory loss and dementia associated with Alzheimer's, reached the market in 1993—a year before President Ronald Reagan announced his own Alzheimer's diagnosis. By 2010, Alzheimer's had become the sixth leading cause of death in the United States, and prevention and treatment remain urgent areas of research.

Illustration of amyloid plaques among neurons: The plaques lead to degeneration of the affected neurons. (iStock ® ).

The human brain can recognize about 10,000 different odors. In 1991, scientists discovered that about 3 percent of the human genome is dedicated to maintaining this complex sense of smell. About 1,000 genes determine the makeup of highly specialized olfactory receptors located on cells in the nasal cavity, which in turn send signals to the olfactory center of the brain.

Illustration of olfactory nerves (Science Photo Library ® ).

For many years, scientists believed that brain plasticity—the ability to change and adapt—was limited to early childhood. However, beginning in the 1990s, research began to demonstrate that the brain is capable of remodeling itself at any age, although at a more limited scale and pace. Principles of neuroplasticity have informed treatment for brain injury and a range of neurological disorders. More recently, scientists have learned about processes that inhibit plasticity, including the role of glial cells in “synaptic pruning” and the accumulation of proteins that affect the brain's receptors.

In 1993, scientists succeeded in isolating and sequencing HTT, the gene that causes Huntington's disease—an autosomal-dominant, fatal neurodegenerative disorder. A simple blood test can now determine the presence of HTT, allowing those with the gene to make informed decisions about family planning.

Illustration of pyramidal neurons of human brain temporal cortex (Science Photo Library ® ).

The most common cause of inherited infant mortality, spinal muscular atrophy usually results from a missing or mutated form of the survival motor neuron 1 gene. Children with this condition suffer from debilitating and often fatal muscle weakness and have problems holding their head up, swallowing, and breathing. In 2019, the U.S. Food and Drug Administration approved a gene replacement therapy involving a one-time intravenous injection that replaces the defective or missing gene with a working copy that increases motor neuron function, decreases the need for respiratory support, and increases the likelihood of survival.

Light micrograph of dendrites (black) from motor neurons in muscle tissue (pink-purple strands) (Science Photo Library ® ).

Woman with spinal muscular atrophy (Science Photo Library ® ).

In 2019, a map of neural connections in the brain of the microscopic worm C. elegans was published—the first complete “connectome” of a multicellular animal. Partial connectomes of a mouse and fruit fly followed months later, marking a significant step forward in visualization of brain processing. These and future connectomes may yield important insights into brain diseases like Alzheimer's and schizophrenia.

Colored 3D diffusion spectral imaging scan of the bundles of white matter nerve fibers in the brain (Science Photo Library ® ).

To reduce the ravages of neurological diseases, much still needs to be learned about how disease and age alter the structure and function of the brain. New knowledge could inaugurate an era of “neurotherapeutics” in which gene therapies, pharmaceuticals, and brain–computer interfaces can address a wide variety of neurological conditions.

• Genomics and Gene Regulation: Deciphering the Instruction Books of Biology

Circular genetic map, showing human chromosome 17. Scientists have isolated chromosome 17 to be the site of a defective gene responsible for many cases of inherited breast cancer. (Science Photo Library ® ).

Following Rosalind Franklin's groundbreaking X-ray diffraction image of DNA and identification by James Watson and Francis Crick of the double helix form of the DNA molecule in the 1950s, the development of genetic engineering in the 1970s opened the door to a wealth of exciting scientific discoveries. Since then, genetic and genomic techniques have expanded far beyond the laboratory, propelling advances in health care, forensic science, evolutionary biology, and biotechnology.

Gene research into breast cancer using a grid of DNA fragments making up human chromosome 17 (Science Photo Library ® ).

The 1970 discovery of cancer-causing genes, known as oncogenes, radically changed the future of cancer research. Building on decades of study of viruses that cause tumors in chickens, scientists discovered that oncogenes originate from normal genes, called proto-oncogenes, that can cause cancer when activated. Identification of oncogenes has led to cancer drugs that target the proteins they make in the body.

Molecular model of H-Ras p21 oncogene protein (Science Photo Library ® ).

For decades, scientists searched for a “magic bullet” that could target specific molecules within the body. In the 1970s, they created one. They combined short-lived cells that produce antibodies with an immortal myeloma cell to produce hybridomas, immortal cell lines that can generate an endless supply of identical antibodies that selectively bind to particular substances. Researchers use these antibodies to detect, isolate, and classify proteins, viruses, cells, tissues, and organs. Monoclonal antibodies also have many therapeutic applications in the prevention, diagnosis, and treatment of disease.

No method existed to sequence DNA molecules until the mid-1970s, when two techniques appeared almost simultaneously: the Sanger chain-termination method, and the Maxam-Gilbert sequencing method. In the Sanger method, enzymes create DNA fragments of varying lengths; in the Maxam-Gilbert method, chemical reactions produce DNA fragments. Both methods then sort these fragments by size to read out the sequence of the original DNA molecule. Though the Maxam-Gilbert method quickly fell from favor, the Sanger method remains widely used today.

NAM member Frederick Sanger (1918–2013), a British biochemist and double Nobel Laureate who pioneered a method of establishing base sequences of DNA (Science Photo Library ® ).

In 1972, scientists discovered how to cut and paste individual genes from one organism into another, creating the first genetically modified organisms and paving the way for the biotechnology industry. This revolutionary process uses molecules known as restriction enzymes as “scissors” to cut molecules of DNA from both organisms. Enzymes known as ligases then act as “glue” to combine the DNA pieces. The resulting DNA molecules have genes from both organisms and are known as recombinant DNA.

DNA profile from a human sample (Science Photo Library ® ).

Since its invention in 1985, the polymerase chain reaction (PCR) technique has become one of the most widely used and transformative technologies in science and medicine. Within a matter of hours, PCR allows for millions of identical copies of DNA to be created from even the smallest sample. In medicine, PCR is used to diagnose genetic defects and detect viruses. In law, PCR generates DNA “fingerprints” that are used to help solve crimes. Evolutionary studies use PCR to reproduce DNA from fossils. Perhaps most importantly, PCR has enabled genomic sequencing on a massive scale that has transformed biological and biomedical research.

Nasal swab sample collection for PCR testing to detect antigens of SARS-CoV-2, responsible for COVID-19 (iStock ® ).

Molecular model of Taq polymerase replicating DNA: The Taq polymerase is blue, the two strands of DNA are green. (Science Photo Library ® ).

In 1988, the U.S. Congress launched the Human Genome Project, an international research collaboration to determine all 3 billion bases encompassed within the DNA sequence of a representative human being. Recognized as the biggest, costliest, and most controversial biomedical project in history, the project's objectives were achieved in 2003. Completion of the project created a new research infrastructure that has revolutionized human genetics and biomedicine. Since then, many thousands of people and animals have had their entire genetic code sequenced.

Computer screen display of a human DNA sequence for the Human Genome Project (Science Photo Library ® ).

Technician loading samples into automated DNA sequencers used for the Human Genome Project. Both images photographed at the Sanger Centre in Cambridge, United Kingdom (Science Photo Library ® ).

On July 5, 1996, Dolly the sheep was born at The Roslin Institute in Scotland, becoming the first mammal successfully cloned from an adult cell. This feat was accomplished through a process called somatic cell nuclear transfer in which a mammary gland cell from an adult sheep was inserted into an unfertilized egg cell without a nucleus. Dolly's birth proved that a cell from part of the body could be used to create an entire individual. This line of research led to advances in stem cell research and the cloning of many other mammals.

British embryologist and NAS member Professor Ian Wilmut and Dolly, the cloned sheep (Science Photo Library ® ).

When viruses infect certain single-celled organisms, they leave behind fragments of their DNA. The organisms use these fragments to form families of repeated DNA sequences, known as clustered regularly interspaced short palindromic repeats, or CRISPRs, which identify and help to combat infections. In 2012, scientists reengineered a CRISPR system that relies on a protein called Cas9 and created a “cut-and-paste” tool that can selectively alter almost any DNA sequence. This powerful gene-editing technology inaugurated a new era in biomedical research and biotechnology.

Computer artwork depicting CRISPR-Cas9 gene editing: The multicolor piece of DNA is a new part replacing an existing portion of the DNA with a new one. (Science Photo Library ® ).

The National Academy of Sciences and the National Academy of Medicine launched an initiative on Human Genome Editing in 2015 to inform decision-making related to ongoing advances in human genome editing research. Subsequent consensus studies and summits have explored the scientific underpinnings of these technologies, their potential use in biomedical research and medicine, and implications of their use.

Since 2008, when Time magazine named “The Retail DNA Test” the Invention of the Year, direct-to-consumer genetic testing (now called consumer genomics) has become increasingly available and affordable. Most direct-to-consumer tests do not analyze the whole genome. Instead, they check for the presence or absence of specific genetic variants known as single nucleotide polymorphisms (SNPs) in a person's genetic code. These markers provide DNA-based information on health, traits, and ancestry. Also in 2008, Americans became protected from discrimination based on their genetic information in both health insurance and employment through the Genetic Information Nondiscrimination Act.

A home saliva collection gene testing kit for ancestry and health-related genes (Science Photo Library ® ).

Beginning in 1994, Myriad Genetics discovered, isolated, and patented two genes—BRCA1 and BRCA2—that can contain genetic variants that greatly increase the risk of breast and ovarian cancer. In 2013, the U.S. Supreme Court overturned these patents, ruling that naturally occurring DNA cannot be patented because it is not a product of human invention. However, synthetically created composite DNA is “patent eligible” because it doesn't occur naturally. This ruling could improve medical innovation and patient care by making it harder for diagnostics companies to gain exclusive control over a person's genetic information.

The previously large gap between how genetic information is used in biomedical research and how it is used in health care is narrowing. Genetic testing, genome editing, the growth of tissues and organs in the lab, and other techniques based on genomics and gene regulation will find a steadily expanding range of clinical uses, but ethical, legal, and social implications will need to be addressed.

• Infection, Immunity, and Inflammation: Unraveling and Applying the Capacities of the Immune System

Tremendous advances in understanding of the immune system and inflammation have led to powerful new treatments against infectious and autoimmune diseases. This golden age of immunology research continues today, offering promise that the body's immune system could be harnessed to fight a wide range of diseases.

Illustration of CAR (chimeric antigen receptor) T cell immunotherapy, a process that is being developed to treat cancer: T cells (one at center), are taken from the patient and have their DNA modified by viruses (spiky spheres) so that they produce CAR (more...)

Receptors on cells known as T cells are responsible for recognizing foreign molecules in the body and triggering an immune system response. In 1974, scientists discovered that these receptors can only recognize foreign molecules that are bound to proteins generated by what is known as the major histocompatibility complex (MHC), a set of genes that varies from person to person and population to population. This discovery heightened understanding of the immune system by revealing the link between people's immune responses and their genetic profiles.

NAS member George Davis Snell shared the 1980 Nobel Prize for Physiology or Medicine for his discovery of the MHC. (Science Photo Library ® ).

B cells help regulate the immune system by producing antibodies and more. Our bodies can generate many millions of structurally different antibodies that defend us against a host of foreign molecules, organisms, or substances. For many years, scientists were unsure whether the genetic diversity required to produce millions of antibodies was generated during evolution (and therefore is carried in sperm and egg cells) or occurs during development (and is generated in other body cells). In 1976, researchers answered this question by showing that the rearrangement of genes in cells other than sperm and egg cells allows the generation of a great diversity of antibodies from a finite number of genes, solving the mystery and opening the way to further discoveries involving gene rearrangements.

Illustration of plasma cells (B-cells, orange) secreting antibodies (white) against viruses (blue). (Note that B cells would be larger than any known viruses.) (Shutterstock ® ).

In 1982, the discovery that variants of normal proteins can act as infectious agents expanded understanding of disease mechanisms and transmission. These proteins, named “prions” (short for proteinaceous infectious particles), are misfolded forms of proteins that act as templates to to promote misfolding among other proteins. Once thought to cause only relatively rare brain diseases, similar aggregates of misfolded proteins are found in Alzheimer's disease, some cancers, and other disorders, suggesting that protein misfolding may play a larger role in human diseases than previously suspected.

Illustration of prion spread: A normal form of prion protein (tiny green balls, lower left) is made with instructions from cell nuclei (brown, one at bottom right). The prions (pink balls, upper right) are a different shape of this normal prion protein, (more...)

A balanced immune response requires the proper regulation of immune cells known as helper T cells. In 1986, scientists identified two types of helper T cells that differ in how they respond to foreign molecules. One type orchestrates an inflammatory response that leads to direct killing of invading microbes, while the other helps stimulate the production of antibodies that ultimately destroy the invader. Recently, the discovery of new subsets of helper T cells has further illuminated the remarkable complexity of the human immune system.

Illustration of a mycobacteria (upper right) infecting a dendritic cell (upper left), and triggering the release of heterodimeric cytokine (lower left), which is an important part of the inflammatory response against infection (Science Photo Library ® (more...)

Roughly a decade after the 1985 discovery of a protein known as Toll in the fruit fly Drosophila melanogaster , many laboratories discovered Toll-like receptors (TLRs) on human cells that play a vital role in activating the immune system in response to danger signals. TLRs are a class of proteins that are primarily produced by white blood cells and recognize white blood cells and recognize molecules broadly shared by pathogenic microbes. The specificity of TLRs and an elaborate regulatory system associated with these receptors ensure that they only recognize molecules associated with these microbial threats, do not activate an inappropriate response to benign microbes, and are highly specific in their response to the different types of antagonistic microbes.

Illustration of a plasma cell (left) secreting antibodies (white) against influenza viruses (right) (Shutterstock ® ).

In 2001, scientists discovered the FOXO3 gene, which plays a pivotal role in causing cell death. Mounting evidence suggests that FOXO3 normally functions as a tumor suppressor by detecting cancerous cells and triggering a variety of responses to these abnormalities, including cell death, but is dysregulated in several types of cancer. Better understanding of the protein encoded by this gene may provide new possibilities for cancer treatment and may have implications for human longevity.

Breast cancer scan (left) with gene mapping visualization (center) and cell cultures (right) (Science Photo Library ® ).

Natural killer (NK) cells are a type of white blood cell that provides a rapid response to cells that have been invaded by a virus or have become cancerous. For many years, NK cells were thought to lack specific cell surface receptors that, if present, would enable them to recognize and kill infected or abnormal cells that they have previously encountered. However, in 2006, a subset of NK cells was found to possess memories of past exposure to a threat, demonstrating a new immunological capacity independent of B and T cells.

Illustration of white blood cells attacking a cancer cell (Shutterstock ® ).

T cells were first used in 2011 to treat advanced cancers. By modifying the cells to express particular receptors, they could be targeted at malignancies that display the protein they were engineered to recognize. Treatment using these engineered immune cells delayed cancer progression and launched a wave of research into modulating the immune system to fight disease.

Illustration of a genetically engineered CAR T cell (top right) recognizing cancerous cells (left) (Science Photo Library ® ).

In 2017, the U.S. Food and Drug Administration approved a therapy for certain kinds of leukemia and lymphoma that uses genetically engineered T cells. In an approach known as adoptive cell transfer, T cells from a person with cancer are removed, genetically altered to target tumor cells, and then transferred back into the person. The technique heralds a new era of individualized cancer therapy.

Young girl with acute lymphocytic leukemia receiving chemotherapy (Science Photo Library ® ).

Continued research into the immune system will drive continual improvements in the diagnosis, prevention, and treatment of disease. At the same time, better understanding of inflammation will provide key insights into widespread diseases, such as dementia and obesity, where inflammation plays a major role.

• Metabolism and Bioenergetics: Linking Energy Use to Health

Light micrograph showing a section through a leaf: Each cell contains several round, green vesicles that are known as chloroplasts. (Science Photo Library ® ).

Building on research conducted during the first seven decades of the 20th century, scientists have made dramatic advances in understanding metabolism and bioenergetics since 1970. Defined as the chemical processes that maintain life, metabolism is highly individual, with both genetic and environmental influences. But research has identified larger patterns and factors that affect metabolism, providing new insights into body weight, metabolic disorders, and the development and progression of disease.

The conversion of light from the sun into energy to drive biological processes, known as photosynthesis, involves the transport of electrons among proteins bound in specialized membranes. Before the early 1980s, the structure of these membrane-bound proteins remained unknown. Then scientists determined the three-dimensional structure of a protein complex that performs the primary energy conversion reaction in a purple bacterium. Though photosynthesis in bacteria is simpler than in algae and green plants, the commonalities among organisms have resulted in greatly increased understanding of how living things capture energy from light.

Molecular model of purple bacterium photosynthesis center (Science Photo Library ® ).

Thousands of structurally different proteins within the body regulate biochemical processes. A critical mechanism in this regulation is phosphorylation, a process probed in the late 1980s, in which enzymes attach a phosphate group to targeted proteins. This chemical alteration is responsible for such vital mechanisms as the regulation of blood sugar, the battle against infection, and the development of cancers. The purification and characterization of proteins that carry out phosphorylation or become phosphorylated themselves launched a wave of research into regulatory processes within the body.

Researcher analyzing protein kinase inhibitors to determine their selectivity as part of research into drug targeting: This technique is known as “kinase profiling.” (Science Photo Library ® ).

The molecule adenosine triphosphate (ATP) functions as the carrier of energy in all living organisms. An enzyme known as ATP synthase, characterized in the 1980s, uses the energy derived from nutrients to add a phosphate group to the molecule adenosine diphosphate. The resulting ATP molecule then transfers this energy to other biochemical molecules, driving the fundamental processes of life.

Illustration of the enzyme complex that drives the synthesis of the energy-carrying molecule ATP (red): The enzyme complex is embedded in the mitochondrial inner membrane (orange). The lower part is a channel through which protons (yellow dots) move. (more...)

For a long time, scientists believed that steroid hormones, which regulate many physiological and developmental processes, act by binding to specific receptors in target cells. But by the mid-1990s they had learned that steroid hormones can act much more quickly through such mechanisms as biochemically altering cell membranes. These fast-acting mechanisms are involved in many cell functions and in the development of hormone-responsive cancers, offering new opportunities for chemotherapeutics.

Researcher with mass spectrometer at the National Physical Laboratory in Teddington, United Kingdom (Science Photo Library ® ).

By 1970 it was possible to diagnose a handful of treatable diseases in newborn babies who appeared healthy by examining their blood a day or two after birth. Technological advances in mass spectrometry in the 1990s made it possible to diagnose dozens of these diseases at once by examining metabolite levels in a drop of blood. These expanded newborn screening assays are now performed in millions of babies each year, allowing physicians to begin lifesaving therapies before the babies even appear sick.

A blood sample is taken on a purpose-designed form from the heel of a newborn infant for a PKU (Phenylketonuria) test at a California hospital. (Science Photo Library ® ).

Tumors frequently display altered metabolism, but until recently it was unknown whether these changes could initiate cancer. Starting in 2008, two related metabolic enzymes were discovered to be frequently mutated in cancers of the brain, bone marrow, and other organs. The mutant enzymes produce large quantities of a metabolite that prevents cells from activating genes needed for cells to mature properly. Some leukemias can now be treated with drugs that block these mutant enzymes and cause cells to lose their malignant properties.

Research on the genetic material of families suffering from diabesity (diabetes and obesity) by the UMR 8090 unit of the French National Centre for Scientific Research, which specializes in the genetics of multifactorial diseases (Science Photo Library (more...)

With multifactorial diseases, heterogeneous combinations of genetic and environmental factors account for the origins of the disease, but how this happens and how it varies from one individual to another have been difficult to unravel. Nevertheless, progress is being made. In 2010, for example, scientists discovered that people have inherited differences in the ability of cells to oxidize nutrients and release chemical energy to be stored in adenosine triphosphate. These differences in cellular energetics may contribute to diseases subject to environmental factors such as diabetes, cancer, and neurodegenerative diseases, and better understanding of these linkages could open new avenues for treatment.

Diabetes patient workshop (Science Photo Library ® ).

Improved knowledge of metabolic processes will enable personalized medicine, in which therapies are tailored to individual patients. Rising rates of obesity and diabetes make this approach especially critical to prevent an onslaught of diseases related to metabolic dysfunction. >

• Diagnostics, Therapeutics, and Imaging: Technologies to Understand, Treat, and Repair the Body

One of the first computed tomography (CT or CAT) scans ever taken of the brain (Science Photo Library ® ).

Technological advances over the past 50 years have transformed diagnostics, therapeutics, and imaging. New devices and techniques have produced sharper images, more reliable diagnoses, and better treatments. Medicine has become more effective, safe, and accurate as a result.

Colored 3D CT scan of a cross-section of healthy lungs from a young adult seen from below. The color differences represent differences in aeration. (Science Photo Library ® ).

A woman with a suspected brain tumor received the first clinical computed tomography (CT) scan at Atkinson Morley Hospital in London on October 1, 1971. This non-invasive body imaging procedure relies on multiple X-ray transmissions, or “slices,” that are then reassembled by a computer to create 3D images of the body's tissues and organs. Though early scanners took several minutes to create a single slice and days to create the reassembled image, modern CT machines take only a few seconds to generate images of the body.

Colored CT scans of the human head and brain (Science Photo Library ® ).

A drug delivery system is a formulation or device that facilitates the introduction of a therapeutic substance into the body and improves its effectiveness and safety by controlling the rate, time, and place of release of drugs in the body. The first drug delivery systems with internal control of their rate of delivery of a therapeutic agent were developed in 1971. By the 1980s, advances in biotechnology and molecular biology had led to systems that enabled ever greater control over the delivery of drugs to the body.

Colored scanning electron micrograph of an open drug delivery capsule. The outer layer enteric coating resists being digested by the stomach and breaks down to release the drug particles inside a specific area of the small intestine. (Science Photo Library (more...)

The quest to develop clinically useful applications of what would become known as magnetic resonance imaging (MRI) started after the first image was taken of a mouse using this technique in 1974. MRI scanners measure the speeds at which atoms in the body return to equilibrium after exposure to magnetic fields. The data then are reconstructed using computed tomography to create a 3D image of the body's internal tissues, including tumors or tissue damage.

Positron emission tomography (PET) generates images by detecting the energy given off by decaying radioactive isotopes following their injection into the body. First developed in 1974, PET scanning is the most specific and sensitive method to image molecular interactions and pathways within the human body, using biomarkers to reveal information about disease, pharmaceutical effects, and many other biological processes. Hybrid PET/CT and PET/MRI imaging systems are now common in clinical settings.

The world's first “test tube baby” conceived via in vitro fertilization (IVF) was born on July 25, 1978, in Aberdeen, Scotland. In this procedure, mature eggs retrieved from the mother's ovaries are fertilized with the father's sperm in a laboratory dish and incubated briefly before the embryo is implanted into the mother's womb. Since 1978, millions of children have been born using IVF, which is now viewed as a mainstream medical treatment for infertility. However, the procedure remains far from perfect, with fewer than one-third of attempts ending in a live birth.

Light microscope image of intracytoplasmic sperm injection for IVF (Science Photo Library ® ).

In 1982, a dentist received the first totally artificial heart to permanently replace his failing natural heart. This device, called the Jarvik-7, featured a pump that replicated the lower two chambers of the heart, providing blood flow to the rest of the body. Though the operation was a success and the patient lived on the Jarvik-7 for 112 days, his quality of life was poor, tethered to a large air compressor and suffering from convulsions, kidney failure, and memory lapses before his death. Subsequent patients with artificial hearts fared better, but work on the “holy grail” of modern medicine, a permanent synthetic heart, continues today.

Jarvik-7 artificial heart (Science Photo Library ® ).

In 1984, the left ventricular assist device (LVAD) made history when it began keeping heart patients alive until donor hearts became available for transplantation. A shortage of donors has led to improvements to these devices that have enabled them to be used as long-term alternatives to heart transplants.

Surgeon (left, at console) performing minimally invasive surgery on a patient's heart using a remotely-controlled surgical robot (center right) (Science Photo Library ® ).

In 1985, surgeons used a robotic arm for the first time to perform a delicate brain biopsy. Since then, robotic systems have become increasingly adept, generally in combination with flexible fiber optic cameras. With their steadily improving dexterity, accuracy, and visualization, robotic systems can now perform minimally invasive procedures that result in shorter hospital stays and quicker recovery times.

MRI is so powerful that it can detect changes in the oxygenation level of the blood. In 1992, several groups began using this technique to map activity in the human brain. Functional MRI (fMRI) imaging can be used to study any motor, sensory, or cognitive task that a patient can perform while in a scanner. Applied clinically to preserve critical functions in patients needing neurosurgery, it has produced groundbreaking insights into how the brain functions in health and disease.

fMRI obtained during presentation of a visual stimulus to the subject in the MRI unit (Science Photo Library ® ).

Almost all of the cells in the human body have developed into a specific type of cell and cannot change—they are skin cells, heart cells, brain cells, and so on. But in 2008 scientists learned how to reprogram human body cells through a process called induction, converting them into cells with the potential to develop into many different cell types. These induced pluripotent stem cells have become vital tools for research and drug development and may someday be used to generate tissues and organs for transplantation.

Colored scanning electron micrograph of a clump of pluripotent stem cells (Science Photo Library ® ).

A classical vaccine works by artificially introducing a weakened or inactivated infectious agent (called an antigen) into the body, stimulating the immune system to produce antibodies that will fight against that infectious agent in the future. In contrast, RNA-based vaccines contain genetic instructions that the body uses to produce antigens, which trigger the immune response to create disease-specific antibodies. The development and use in 2020 of several RNA-based vaccines that effectively protect against COVID-19 will likely drive huge growth of this technology. In the future, RNA vaccines may allow for a single vaccination to protect against multiple diseases.

COVID-19 RNA vaccine, illustration: The vaccine consists of strands of mRNA (messenger ribonucleic acid) encased in a lipid nanoparticle sphere (red) surrounded by a polyethylene glycol coat (violet). (Science Photo Library ® ).

As the quest for better medical imaging devices, diagnostic measures, and therapeutic tools advances, health care will continue to improve. Diagnoses will become more accurate, procedures less invasive, and treatments more effective.

• Scientific and Medical Ethics: Protecting People While Enhancing Autonomy and Justice

Philosopher and physician Hippocrates (c. 460–c. 370 BC), also known as the Father of Medicine. His code of ethical conduct forms the basis of the modern-day Hippocratic oath taken by doctors. (iStock ® ).

Ethical lapses and emerging quandaries in biomedical research and practice have provided continual reminders of the need to emphasize scientific and medical ethics in the training of students and the oversight of researchers and health care providers. Today, hospitals and research institutions maintain ethics committees to provide researchers and physicians with consultation and guidance, and bioethics education has become a strong component of training and research programs.

For 40 years, a group of 600 African American men from Tuskegee, Alabama—about 400 with syphilis and 200 who did not have the disease—were unknowing subjects in a study of the effects of syphilis sponsored by the U.S. Department of Health, Education, and Welfare (today the U.S. Department of Health and Human Services). Administrators misinformed study subjects about the purpose of the research and withheld penicillin, a widely available and effective treatment. After a news story alerted the public and Congress about this misconduct in 1972, the study was found to be “ethically unjustified” and was ended. Although victims of the Tuskegee Syphilis Study eventually received a settlement and health benefits from the U.S. government, the incident led to long-lasting mistrust of public health officials among many African American communities.

Tuskegee Syphilis Study participant being X-rayed (Science Photo Library ® ).

In 1973, Congress passed the National Research Act, which called for the development of regulations on research with human subjects, required institutions to form Institutional Review Boards to oversee these regulations, and formed a new commission to shape bioethics policy in the United States. In 1978, this commission was replaced with the President's Commission for the Study of Ethical Problems in Medicine and Biomedical and Behavioral Research, which was charged with creating recommendations for policymakers, practitioners, and the public on such issues as health care access, the definition of death, patient consent, human research subjects, genetic engineering, and the terminally ill.

Nurse discussing paperwork with two brothers who are taking part in research on siblings (Science Photo Library ® ).

Prior to 1970, little research and guidance existed to help physicians deal with the ethical dilemmas they faced in daily clinical practice. In response to this need, the field of clinical medical ethics emerged in the 1970s to provide a platform for research, education, and evaluation around ethical decision-making in clinical care. In contrast to the broader focus of bioethics, clinical medical ethics seeks to help physicians and other health professionals identify and respond to ethical challenges that arise in the ordinary care of patients, including truth telling, informed consent, confidentiality, surrogate decision making, and end-of-life care.

Reports of research misconduct in the 1980s led to the creation of the Office of Research Integrity within the U.S. Department of Health and Human Services, which works to investigate scientific misconduct and provide support to universities conducting research using human subjects. In 1991, a new regulatory framework dubbed the “Common Rule” added protection of human subjects in almost all government agencies. Among other requirements, the Common Rule provides special protections for pregnant women, children, and incarcerated people.

The White House Office of Science and Technology Policy finalized a federal definition of research misconduct in 2000 as “fabrication, falsification, or plagiarism” in proposing, performing, or reviewing research, or in reporting research results. The definition does not include “honest error or differences of opinion.” Misconduct must be proved to have been committed knowingly, intentionally, or recklessly, or with a significant departure from accepted practices of the relevant research community. Overseen by a federal Office of Research Integrity, this process built on a 1992 report on responsible research from the National Academy of Sciences, the National Academy of Engineering, and the Institute of Medicine.

Science Photo Library ® .

For decades, drug companies were reported to have suppressed data about dangerous side effects of popular medications, both in publications and in seeking approvals by the U.S. Food and Drug Administration (FDA) for new drugs or new uses of drugs. In 2001, reports of propensity for heart attack and stroke associated with the pain reliever Vioxx and increased suicidality among children and young adults taking certain antidepressants prompted action. The FDA and academic journals began to require that the initiation of clinical trials and the availability of results from clinical trials be registered on a publicly available website and also implemented the Sentinel System for real-time monitoring of new therapeutics.

Vioxx was used mainly to relieve pain due to osteoarthritis, but was withdrawn due to fears that it increased the risk of heart attack. (Science Photo Library ® ).

In response to recommendations from the 2004 National Research Council report Biotechnology Research in an Age of Terrorism , the U.S. Department of Health and Human Services established the National Science Advisory Board for Biosecurity in 2005. The goal was to provide advice and guidance to federal agencies, scientists, and journals concerning oversight and public availability of research in biotechnology or biomedicine that has the potential to be a threat to public health, agriculture, the economy, or national security. A high-profile example involved the decision by Science magazine to publish the molecular methods for recreating and studying the strain of influenza virus that caused the 1918 pandemic that killed tens of millions of people.

Technician working in a high-security isolation cabinet, used for work with the most dangerous of microorganisms: Photographed at the World Influenza Center at the National Institute for Medical Research, Mill Hill, London. (Science Photo Library ® (more...)

In 2010, a widely acclaimed book recounted the story of Henrietta Lacks, an African American woman who provided the tissue from which HeLa cells, an immortal cell line widely used in biomedical research, were derived in 1951. In an echo of the Tuskegee Syphilis Study, the public learned that researchers had used Lacks's cells without her consent and without providing the family any compensation. In 2013, criticisms concerning privacy and informed consent intensified after the online publication of the whole genome sequence of one strain of HeLa cells. Though the sequence was quickly removed from the public domain, controversy continues over whether consent should be required for the use of biospecimens in research.

Colored scanning electron micrograph of HeLa cells that have just replicated (Science Photo Library ® ).

Henrietta Lacks (Science Photo Library ® ).

In addition to gender and other inequities, addressing ongoing racial disparities must be intentionally integrated into the movement toward health equity and high-quality care for all. The Black Lives Matter movement, increased large-scale demonstrations and national dialogue around the material impacts of structural racism, and disproportionately high levels of COVID-19 infection and mortality in communities of color have demonstrated the critical need to address inequities and structural racism in the health and medicine fields today and into the future. Ensuring racial equity will be a focus of scientific and medical ethics in the coming years.

Starting in 2017, the #MeToo and #TimesUp movements brought the scope and severity of sexual harassment and gender inequity to the forefront of public consciousness—extending to the experiences of women in the sciences, engineering, and medicine. In 2018, a report from the National Academies of Sciences, Engineering, and Medicine identified sexual harassment of women as an enduring problem in academia, and the following year each of the National Academies adopted a code of conduct establishing standards for personal and professional conduct.

Sign photographed at the Women's March, January 20, 2018, San Francisco, CA (Shutterstock ® ).

In defiance of an unofficial international moratorium on editing human embryos intended for a pregnancy, a Chinese scientist announced in 2018 that he had edited the genomes of twin girls in an attempt to make them immune to human immunodeficiency virus. His actions sparked international outrage and caused many scientists and policymakers to call for an official ban on human germline genome editing. A 2020 National Academy of Medicine/National Academy of Sciences report, Heritable Human Genome Editing , detailed the scientific, medical, ethical, moral, and societal issues that need to be addressed before heritable genome editing could be permitted.

Scientific and medical ethics will become increasingly complex as the world confronts new issues, such as the effects of climate change on human health or the role of government in shaping individual behavior. Continual attention to medical and research ethics will be essential to ensure individual and collective responsibility and human well-being.

• Cite this Page National Academy of Medicine; Yamamoto K, Woolley M, Higginbotham E, et al., editors. Transforming Human Health: Celebrating 50 Years of Discovery and Progress. Washington (DC): National Academies Press (US); 2023 Feb 13. 1, BIOMEDICAL SCIENCE AND TECHNOLOGY.
• PDF version of this title (154M)

In this Page

Recent activity.

• BIOMEDICAL SCIENCE AND TECHNOLOGY - Transforming Human Health BIOMEDICAL SCIENCE AND TECHNOLOGY - Transforming Human Health

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

Connect with NLM

National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894

Web Policies FOIA HHS Vulnerability Disclosure

Help Accessibility Careers