The quantification of human organ morphology, in both health and disease, is a complex task that can be tackled by multimodal spatial imaging modalities, capable of spanning across dimensional scales. Complete tissue morphological characterization requires the detection of interactions between and across scales; however, most imaging techniques are limited by either resolution or field of view, making it difficult to bridge macroscopic with microscopic observations and data. Conventional histology1,2,3 or electron microscopy4,5,6 approaches permit the visualization of the tissue’s microstructural organization and composition through serial sections, and the data can be suitably quantified; however, these approaches normally require sampling and sectioning of the tissue and are extremely labor intensive and time consuming. Optical clearing combined with light sheet microscopy can provide a large field of view with high resolution; however, tissue clearing also requires large timescales and is often expensive; in addition, the depth of imaging for a light sheet microscope is limited by the objective lens working distance7,8. Even where whole adult human organs8 or whole animals9 have been cleared over a period of several months, imaging them remains challenging. Similar drawbacks apply to optical coherence tomography10,11, multiphoton microscopy12 or confocal microscopy13,14, which can capture the local three-dimensional (3D) microstructure of the tissue at the cellular scale, but have limited tissue penetration, hindering deep tissue imaging15. Recently, high-resolution magnetic resonance imaging (MRI) achieved an isotropic voxel size of 100 µm in a whole ex vivo human brain16. Although MRI is nondestructive and has a large field of view17, the resolution is still not sufficient to examine tissue microstructure. Hierarchical imaging techniques are capable of overcoming the trade-off between resolution and field of view. In a hierarchical approach, multiple images of the same sample are acquired at different resolutions to bridge the different scales. Microcomputed tomography (µCT) has been used to image entire lungs with a resolution of 150 μm voxels, followed by subsequent extraction of biopsy cores in the lungs; these small cores were then scanned with µCT to achieve 10 μm voxels18.

Considerable progress has been made in the field of X-ray imaging over the past decade, especially for the visualization of soft tissue19. Synchrotron X-ray computed tomography (sCT) has proved to be one of the most powerful X-ray-based imaging techniques due to its high brightness20, enabling the observation of soft tissue at high resolution, with enough contrast to detect microstructural components21, such as individual neurons22 or elastin fibers23. In particular, phase-contrast-based sCT24, combined with optimized sample preparation and mounting procedures has provided extensive information of heart fiber orientation25,26 and, separately, of brain cellular maps27. Phase-contrast imaging refers to the detection of phase shifts of an X-ray beam imparted by a sample28. This technique enables the visualization of soft tissues that would be undetectable with conventional X-ray tomography in the absence of a contrast agent29. Nevertheless, the studies using sCT are mostly limited to fetal organs30, small subsamples of adult human organs31 or organs from small animal models, e.g., mouse32 and rabbit33, due to the restricted field of view. Some studies have demonstrated the imaging of larger samples, such as coelacanths with a diameter reaching 10 cm and a height of 30 cm34,35; however, the voxel size was limited to 30 µm and the structure of interest were hard tissues or cartilages.

Fourth-generation synchrotron sources, such as the European Synchrotron Radiation Facility (ESRF)’s Extremely Brilliant Source, provide enough beam spatial coherence and flux to visualize intact human organs from the macro- to the microscale. Using the ESRF’s Extremely Brilliant Source, we have recently developed a technique termed hierarchical phase-contrast tomography36 (HiP-CT) that allows the scanning of large intact human organs with ~20 µm isotropic voxels, with subsequent zooming (without sectioning), achieving up to 1 μm isotropic voxels locally. Although often overlooked, sample preparation and mounting are crucial to achieve the highest resolutions with this and other techniques, especially when visualizing large soft tissue samples, such as human organs, where the ratio of voxel size to organ diameter is 1:150,000 (1 µm in 150 mm). The maximum imageable diameter is limited by the equipment and setup parameters of the beamline, such as the width of the X-ray beam, the size of the detector, the computing power available to reconstruct the data and the size of the data. Currently, the maximum organ diameter we have imaged is 150 mm, but the present setup would be compatible up to 250 mm diameter by 500 mm vertically.

The success of all these experimental techniques depends on the careful preparation of the sample to avoid imaging artifacts. Soft tissue imaging using µCT or sCT presents a number of challenges compared with hard tissues. One of these issues is the lack of contrast when imaged with X-rays due to the similar densities of the sample components. Furthermore, the back-projection algorithms typically used to reconstruct the 3D volume from X-ray projections assume no movement of the sample during scanning. Most sample preparation methods are designed to prevent drifting or deformation of the sample, to avoid movement artifacts that would reduce the quality of the images37. Some reconstruction algorithms, such as motion compensation38,39 or machine learning based40,41, have been developed to overcome this issue42 but they remain complex and computationally costly to operate19.

Although the protocol we describe is applicable to a wide range of biological samples of varying sizes and using a range of imaging modalities, here we focus on samples to be imaged using synchrotron phase contrast imaging. The method is also applicable to other imaging modalities; however, the procedure may need optimizing according to sample size and modality chosen (see protocol steps for possible optimization recommendations).

Development and overview of the protocol

The sample preparation and mounting protocol described herein was developed from the need to image large biological volumes, such as intact human lung, brain, heart, kidney and spleen. The technique enables scanning of whole organs with 25 µm isotropic voxel size. Areas are then selected for further high-resolution scanning without requiring biopsies. To image such large structures, a high-energy X-ray beam is required to penetrate the samples. We used a polychromatic beam with energy ranging from 64 to 120 keV for organ imaging, but higher energies (up to 140 keV) would be optimal, if available.

Preparation and stabilization of the organ for this technique is essential as any density inhomogeneity, gas bubbles or movement during scanning would greatly reduce the image quality. During the development of the method, several challenges arose, as described in the Extended Data Fig. 1. Bubble entrapment (during mounting) and bubble formation (during scanning), both create motion blurring and phase contrast artifacts, and reduce the scan quality. This was resolved by limiting the dose absorbed by the sample and including multiple degassing steps during the organ preparation and mounting procedure. Sample movement during scanning is a further challenge, particularly as the samples are large, and hence often require a long time (>5 h) to image. This challenge was solved by carefully packing a mixture of crushed agar gel and liquid (in our case ethanol 70%) as a mounting media around the organ. In addition, the dehydration of the organ with ethanol increased the contrast of the images43 and diminished the bubble formation.

Here, we present a procedure to prepare whole human organs for imaging with sCT, µCT, medical CT and MRI that is compatible with a final stage of classical paraffin embedded histology. In this protocol, we mainly describe the sample preparation and mounting with ethanol-agar; the X-ray imaging protocol using HiP-CT and its application to a medical application (quantifying the damage coronavirus disease 2019 (COVID-19) does to lung vasculature)44,45. In brief, after fixation of the body (Step 1A(i)), the organ(s) are extracted (Step 1A(ii–iii)), immersion fixed (Step 2), dehydrated and degassed with vacuum (Step 4A) or thermal cycles (Step 4B) depending on its fragility, mounted with crushed agar gel mixture (Steps 5–33) and imaged using sCT (Steps 34–54), µCT (Step 55), clinical CT (Step 55) and MRI (Step 55), and finally, histological analysis is performed (Step 56). For an overview of the procedure, see Fig. 1.

Fig. 1: Overview of the sample preparation, stabilization and scanning of a large biological sample.

The protocol contains three major steps (organ acquisition and preparation, organ mounting, and imaging) indicated in color-coded boxes. The organ can be retrieved either from a biobank, a surgery, a dismissed transplantation or it can be extracted from a donated body. We provide two protocols for degassing depending on the fragility of the organ (vacuum degassing and thermal cycles degassing). Once the sample is mounted with the agar gel, different imaging techniques can be performed (µCT, sCT, clinical CT and MRI). After imaging, histology can be carried out on the sample.

Advantages and limitations

This protocol was developed and optimized for the preparation of whole human organs to be imaged at high-resolution with HiP-CT, as shown in Fig. 2, and has been applied to various organs including brain, heart, lung, kidney and spleen. Nevertheless, the organ preparation procedure is flexible and can be used with soft tissues of animal origin or even with intact small animals. A list of human organs and biological samples imaged with this method and the preparation time for each step is provided in Extended Data Fig. 3. Most common sample preparation methods for X-ray imaging include fixation, alcoholic dehydration, paraffin embedding or critical-point drying. One of the main drawback of wet embedding such as alcoholic immersion is sample drift, which creates movement artifacts37. Paraffin embedding and critical-point drying modify the specimen structure to increase its rigidity, enabling the specimen to be immobile for a long period of time.

Fig. 2: HiP-CT images of an intact human heart.

a, 3D view of the human heart imaged using the beamline BM05 at ESRF. b–d, Cross-sections of the heart with an isotropic voxel size of 25 µm (b), 6.5 µm (c) and 2.5 µm (d). e, Magnification of the 2.5 voxel image with annotation on the principal structures observed (ca, coronary artery; mc, myocyte cells; ad, adipose tissue). All the basic information on the patient from whom this organ originates is provided in Extended Data Fig. 2. All experiments followed the relevant governmental and institutional ethics regulations for human experiments.

Prevention of sample movement and shrinkage

In this protocol, sample movement is prevented by the use of small blocks of agar gel and crushed agar gel in density equilibrium with the mounting liquid, holding the sample in place. Scans taken several months apart on the same sample prepared with our organ-stabilization protocol can simply be registered using manual rigid transformation. This demonstrates the stability of this method over long time periods. An issue common to all these sample preparation methods is tissue shrinkage. This effect can hinder morphological quantification and lead to erroneous analyses. However, compared with paraffin embedding46,47 and critical point drying37,48, the sample shrinkage in this protocol can be mitigated using multiple ethanol baths in ascending ethanol concentrations43 up to 70%, thus preserving the morphology of the tissue. Furthermore, the dehydration of the sample with ethanol ensures a higher contrast with µCT43,49 or sCT50. This specimen preparation protocol is compatible with many MRI, clinical CT and histology (after 3D imaging and dismounting) techniques. Finally, whilst the specific imaging equipment may be specialized, the preparation protocol can be implemented in any biological laboratory with an adequate fume hood. The equipment and materials are readily available from standard scientific suppliers, with the most specialized equipment being a vacuum pump and desiccator chamber.

Although this procedure presents several advantages compared with other sample preparation protocols, some limitations should be noted.

Fixation and dehydration of the sample

The timing of the different steps of the protocol involving fixation and degassing are heavily dependent on the composition and size of the tissue. In the present procedure, we give examples of timing for different human organs; whereas timing for other types of tissues would need to be tested and verified. Some guidelines for optimization of timings are provided in this protocol.

Whilst fixation of the organ is critical for long-term tissue preservation, it alters tissue mechanical51,52 and diffusion properties, which could confound other measurements. Despite increasing the image contrast, ethanol dehydration also affects the mechanical properties of the tissue43,53. This limits the use of this preparation method for in situ testing; however, by not fixing the tissue and replacing ethanol with water, it is possible to use in situ imaging13,54. Thus, the protocol could be expanded in the future to cover dynamic experiments and quantification of mechanical properties over a short timeframe (as biological degradation would occur). If the contrast provided by ethanol is not high enough, or not adapted to the experimental needs, various contrast agents, such as iodine-based55 or tungsten-based contrast agents56 could be used to increase the overall contrast of the tissue, or resolve specific components of interest13,56. If agar gel was not desirable for a particular application, it could be replaced with other solid or elastic media that would be in equilibrium with the mounting liquid and relatively amorphous in its structure. For instance, in case of mounting with 96% ethanol, transparent candle crystal gel can be used instead of agar gel. In case of mounting with water, gelatin blocks or polyacrylamide blocks can also be used.

X-ray dose limit

One of the main drawbacks of wet embedding methods is the formation of bubbles due to dose rate or dose accumulation in the case of sCT57. These bubbles can move or damage the sample, in addition they can create strong artifacts, dramatically reducing the image quality58. Although bubbling is an issue with other standard preparation methods, e.g., paraffin embedding, it is not present with critical-point drying. In this protocol, the issue is mitigated by multiple degassing steps during the procedure, delaying the nucleation of bubbles during the scan and further mitigated by the use of high-energy X-rays with strong phase contrast. However, only a few beamlines are equipped to image soft tissue at high energy with sufficient coherence properties and propagation capabilities. Spatial coherence must be high enough that the propagation distance can be set to distinguish the density variation of the sample without geometric blurring. For low-energy X-ray tomography, the degassing steps must be performed conscientiously, and the dose rate controlled even more carefully because of the photodissociation of the water molecule.

Experimental designSample collection and fixation

Biological tissue can be collected by various methods. If the sample is collected directly from a biobank, surgery or dismissed transplantation (Step 1B), it can be taken directly to the sample fixation stage (Step 2). If the organs come from a donated body, organ extraction must be performed (Step 1A(ii)). Embalming of the body is carried out shortly after death, before organ extraction (by a licenced practitioner). The body is fixed by injecting formalin diluted in a solution containing lanolin into the right carotid artery (Step 1A(i)). After evisceration, complete fixation of the organ is ensured by immersing it in 4% neutral buffered formaldehyde (Step 2). The duration of fixation is defined by the size of the organ, e.g., 4 d for a human brain. When eviscerating the organ, ensure as much surrounding tissue as possible is removed to decrease the time of penetration of the fixative into the organ. The volume of fixative is also important, we recommend a volume at least four times the tissue volume.

Some organs, such as the lung, may require inflation. This can be partially accomplished at the fixation stage by using the instillation of formalin in the lungs under controlled pressure59. The lung is perfused with 4% formalin through the trachea using a 30 cm water column, the trachea can then be ligated to maintain the inflated configuration over a period of 2 d. The lung is subsequently immersed in a 4% formalin solution after extraction. Once fixed, the replacement of formalin by the successive baths of ethanol with vacuum pumping is coupled with the injection of ethanol in the bronchia to maintain a consistent 3D shape. Strictly controlled pressure with ethanol degassing is currently not possible with the proposed protocol, yet should become possible using the circulation of degassed ethanol instilled at a controlled pressure. Our results show that, even without strictly controlled inflation at the ethanol dehydration stage, the data still have high scientific utility.

Organ dehydration and degassing

The organ is dehydrated with multiple pre-degassed ethanol baths. The transition to 70% ethanol must be gradual to avoid shrinkage43. An ethanol concentration of 70% was chosen as the best compromise between controlled shrinkage and sufficient contrast to observe structures of interest using phase contrast; however, a different final ethanol concentration can be used depending on the application. During this step, degassing has to be performed to remove free and dissolved gas present in the tissue to avoid bubble formation or volume increase during imaging. For human organs, we present two methods, depending on the fragility of the organ for dehydration and degassing. The vacuum degassing (Step 4A(i–v)) can be used for most organs (heart, lung, liver, kidney, spleen). The organ is immersed at room temperature (20 °C) in four successive pre-degassed ethanol baths of concentrations of 50%, 60% and 70%, and then a second 70% bath to ensure equilibrium is reached. A degassing step is performed between baths. Equilibrium is generally reached in 4 d for each concentration for a human organ such as a lung or a heart. Degassing is performed with a vacuum pump and a desiccator by successive cycling down to an absolute pressure between 15 and 10 mbar. The degassing is considered suitable when no strong bubbling can be observed at 10 mbar. Alternatively, thermal cycling (between room temperature and 4 °C) (Step 4B(i)) was developed for fragile organs, such as human brain, as some damage was observed after using the vacuum degassing method36 if the bubbles were not able to find a way out of the brain. In this method, four thermal cycles are performed by immersing the organ in four successive baths of 50%, 60% and 70%, and a second 70% pre-degassed ethanol. Each thermal cycle consists of immersing the organ in the highly degassed ethanol bath at room temperature. The container has to be closed with care to avoid entrapping bubbles of air. It is then kept in a refrigerator for 4–5 d at 4 °C. During this period, the dissolved gas will diffuse into the surrounding ethanol, and the bubbles will progressively dissolve. After this time, the solutions and organ must be brought back to room temperature, and a new cycle can be started using a new strongly pre-degassed ethanol bath. For both methods, the minimum number of ethanol baths is four to reach a final concentration of 70% without having substantial shrinkage. The immersion times have to be adapted and optimized to the type and fragility of the organ. The result of the degassing can be tested by making a radiograph of the organ in its jar without the mounting media described hereafter. If some remaining bubbles are still visible, more thermal cycles can be performed with ethanol at 70%. Organs with adipose tissues, such as the brain, require a longer time to equilibrate with ethanol. At each stage, dehydration can be checked by disturbing the container with the organ inside and looking for streaks of different density (different transparency) forming in the surrounding ethanol solution, this indicates the presence of water in the ethanol.

Ethanol dehydration is sufficient to observe the structures of interest with µCT and sCT when using phase contrast; however, it should be possible to combine this protocol with the use of a contrast agent to resolve specific components of the sample or when using less sensitive imaging techniques56. The contrast agent must be miscible with 70% ethanol, or should be applied to the fixed organ before the ethanol dehydration. Contrast agents would increase the absorption of the sample, which may enhance bubble formation during imaging if using intense synchrotron X-ray beam.

In cases where the characterization technique is not be compatible with ethanol (for instance, MRI for diffusion), the same protocol for degassing and then for mounting can be applied using water with formalin at the desired concentration. For in situ applications requiring close-to-biological conditions, this protocol can also be used with water only, but the samples can be used for a only few hours as biological degradation would occur. Specific safety aspects (working under an adapted fume hood in a well-ventilated laboratory, wearing gloves, a laboratory coat, closed shoes and safety glasses) have to be taken as solutions of formalin or ethanol produces dangerous vapors.

Bubble formation

One of the main problems in wet embedding is the presence of bubbles. These can come from the mounting protocol (bubbles in the organs or trapped in the mounting media) and/or from very high X-ray doses resulting in evaporation of the ethanol (in case of sCT). In both cases they can create substantial artifacts (Fig. 3), and in case of bubbling from the X-ray dose, the movements of the bubbles during the scanning often renders the scans unusable60,61. Preliminary tests showed that degassing the sample before imaging removes trapped bubbles and delays the nucleation and growth of new bubbles. Hence, several degassing steps were incorporated in the protocol to mitigate the bubble formation.

Fig. 3: Example of bubble artifacts.

a–d, Artifacts due to the radiation dose (a,b) or to a stable bubble entrapped during mounting (c,d). a, A cross-section of a human liver with numerous bubbles that have developed due to too high a dose uptake due to a scan crashed with the beam on for several hours, causing artifacts in the images. ba, bubble artifact. b, A cross-section of the liver imaged after removal of bubbles by degassing the sample with vacuum degassing. c, A cross-section of a human brain with a bubble entrapped during mounting. d, A zoom on the bubble trapped inside the brain. The images were obtained at BM05 beamline at ESRF. All experiments followed the relevant governmental and institutional ethics regulations for human experiments.

Organ mounting

The purpose of this step is to maintain the organ in position during the scan to avoid motion artifacts and to ensure that more scans can be performed at a later stage with good 3D rigid registration between subsequent scans/experiments. The agar gel cannot be prepared with 70% ethanol directly, but must be done by slowly adding powder of agar agar in agitated hot demineralized water (~85 °C) up to 20 g/L. Once fully dissolved, the agar agar solution is poured into a suitable container and let to cool down for several hours. Once gelled, the mixture can be cut into small cubes (Step 8) and added to 96% ethanol (6 L:2 L ethanol:agar agar) (Step 9), ensuring a final ethanol concentration of 70%. After degassing (Step 12) and crushing of a part of the agar cubes (Step 14), the mixture is ready for mounting. If mounting with formalin is preferred to ethanol (e.g., to improve contrast with MRI), the 96% ethanol bath can be replaced with a 4% formalin bath. The agar gel can be prepared in advance and stored in an airtight container to prevent the reintroduction of gas into the solution after its degassing. The amount prepared depends on the size of the organ and container used for final mounting. An agar gel prepared with 70% ethanol ensures a drastic reduction of bubble formation during imaging. It is important to use a crushed agar gel and not a blended agar gel, as the blended gel does not hold the sample as firmly as the crushed agar gel and could lead to movement during imaging. Initially, only agar cubes were used to hold the sample in position; however, the cubes were found to be too rigid, creating deformations where they contacted the surface of soft organs such as lungs. Thus, the cubes should only be used at the bottom and top of the container. A few centimeters are used to create a solid base and avoid rotation of the sample (Step 16). Crushed agar is used in the remainder of the container to maintain the sample in position. The mixture of crushed agar in the mounting liquid (70% ethanol or 4% formalin) must be added to the container gently with a ladle to avoid gas bubble entrapping during the process (Fig. 4c). Rapid vacuum degassing should be performed at least three times when adding the agar crushed gel to remove entrapped bubbles. These vacuum cycles should be performed ideally down to 15 mbars. The dimensions of the container should be as close as possible to the specimen to minimize the amount of material that the X-ray beam has to pass through; however, the organ should not touch the side of the container to avoid artifacts in the images that would compromise the accuracy of the reconstruction (i.e., a minimum of 5 mm of crushed agar gel should surround the organ to avoid direct contact with the container). For sCT, the container used for the mounting must be made of a material resistant to X-rays and not too dense, such as polyethylene terephthalate. Glass should be avoided as its high density compared with the samples would result in strong absorption contrast artifacts. Once the agar gel has been compacted around the sample (Step 22) and properly degassed, the container can be sealed with a liquid-tight lid (Step 28). The mounting should be assessed to ensure that no movement of the sample in the container is possible and no bubbles can be seen inside. If some bubbles are entrapped, rapid vacuum degassing can be used to remove them. The same approach can be used to remove bubbles in an organ in case of a bubbling event due to high dose during sCT scanning (do not degas the s