Special issue: Biomedical imaging in comparative anatomy

Abstract

The last few decades have registered remarkable advancements in biomedical imaging, where novel platforms and, at times, existing ones with major upgrades have become available to researchers. Anatomy research has harvested the fruits of sophisticated imaging tools to its benefit that, in turn, have not only expanded the scope of anatomical research but have also enhanced the depth of information ingrained in imagery datasets, for example, by gaining higher resolution and deeper tissue permeation (Schramek et al., 2013). Such advanced platforms, coupled with extensive digitisation of datasets, present a range of unprecedented advantages, including data sharing that fosters collaboration and improved digital rendition with 3D comprehension of anatomical structures and investigation of internal tissue fabric with both qualitative and quantitative analyses (Wang et al., 2024). Advancements in algorithm programming are also game changers in the field, where user-friendly and often freely available software programmes provide fertile ground for further exploration.

While working on this Special Issue, it was not easy to capture a snapshot of the entire spectrum of imaging platforms that the current anatomy researchers are engaging in. A major component of the current biomedical imaging studies is focused on software programming and developing image analysis protocols around a multitude of biological systems, including whole-body imaging, imaging of real-life tissue samples with an understanding of tissue texture, and in vitro tissue mimics, such as tissue spheroids. Fortunately, these platforms enjoy extraordinary adaptability, and their utility is not limited to a single sample prototype. Therefore, the Special Issue needed to be inclusive of samples or species based upon which such a wide range of imaging platforms were developed, given there is enough space for collaboration and cross-fertilisation of ideas across the spectrum of biological samples that anatomy researchers currently deal with.

The scope of the term “comparative anatomy” from the perspective of this Special Issue was about comparing different species, including humans. Thus, a segment of the included papers was based on human specimens, while non-human species were also featured to strike a balance. Being a strong proponent of the One Health approach in medical education, isolating humans from non-human species risks missing the broader picture of how fundamental anatomical mechanisms work, while multiple organs and systems create a collaborative ecosystem for the vertebrates to thrive (Bhattacharjee et al., 2022). Furthermore, as applicable to many imaging platforms, although their utility is demonstrated in a particular species, they bear the potential for simultaneous uses in other species—both human and non-human. Hence, the aim was to forge a successful marriage between the distinct domains of “biomedical imaging” and “comparative anatomy”—domains otherwise independent enough in their rights, within the scope of this issue.

A considerable fraction of the articles in this issue were based on computed tomography (CT), which is fair given the progress achieved with its instrumentation, data rendition, and analysis. Its non-invasive nature and ability to reveal intricate structural details make it a popular choice for biological samples (Wolosker et al., 2021). Moreover, the emergence of micro-CT technique compatible with minute and, at times, fragile samples has added to its advantage (Jehoon et al., 2021; Thompson et al., 2020). Such a CT-based study investigated the protrusion of the infraorbital canal into the maxillary sinus with implications for infraorbital nerve injury due to, for example, trauma-induced maxillary fracture or iatrogenic injury like in endoscopic surgery and rhinoplasty (Karatag et al., 2024). The study included a cohort of 500 patients and reported an interesting array of CT data analysis, quantification, and validation steps that may help anatomy researchers.

The ability of CT to unravel information on delicate anatomical features was further exemplified in a study that investigated scapholunate ligament injury and how it impacted the complex biomechanics of the wrist joint (Promny et al., 2024). The authors converted the CT data into 3D models with statistical analysis that provided fascinating insights into the working of the carpal bones, supported by the scapholunate ligament, to achieve the versatile kinematics noticed in the wrist joint. The relevance of this article, especially the building of 3D models and how to analyse them, would be helpful for researchers, irrespective of whatever vertebrate species they are focused on.

Another emerging front in CT-based investigation of anatomy samples is the use of (primarily) iodine-based contrast agents for enhanced visualisation of the soft tissue—also known as diffusible iodine-based contrast-enhanced computed tomography or diceCT (Gignac et al., 2016; Gray et al., 2024). This is an interesting development in anatomy research that has the potential to revolutionise the current paradigms in soft tissue imaging. Although iodine dyes are currently being prioritised, other dyes for augmenting CT contrast, such as phosphotungstic acid (Lesciotto et al., 2020) and osmium tetroxide (Lanzetti & Ekdale, 2021), are also reported. These innovations are further opening up the possibilities and reach of diceCT, and this Special Issue is fortunate to showcase some of its relevant uses in (comparative) anatomy research.

However, a performance comparison between the variety of diceCT contrast agents remains scarce, and data are largely unavailable on how, for example, such iodine-based contrast agents, based on their chemistry, might influence the quality of data. Hence, a study was conducted that compared the performance of Lugol's iodine dissolved in water and ethanol as micro-CT contrast agents on 16 specimens that represented eight species from four vertebrate groups: amphibians, mammals, reptiles, and fishes (Crowell et al., 2024). The results provided insights into how solvent selection (water and ethanol in this case) while preparing Lugol's iodine impacted the imaging of different soft tissue structures, such as muscles, with adequate resolution. Such methodological research is important, especially in comparative anatomy, to refine the current protocols.

The prowess of diceCT in providing high-resolution imaging of soft tissue was further exploited in a study where 10% (w/v) Lugol's iodine solution was used for morphometric analysis of the muscular masses, muscle fibre lengths, and physiological muscle cross-sectional areas in the Egyptian fruit bat (Rousettus aegyptiacus) with comparison to a wide array of flying birds and non-flying mammals (Kissane et al., 2024). The data provided adequate information and enabled detailed analysis and comparison between the diverse species. Furthermore, it highlighted the curious anatomical adaptations that the muscles of Egyptian fruit bats have undergone to achieve powered flight.

The ability of diceCT to image muscles was also reported in an investigation on the anatomy of mystacial pads in harbour seals (Phoca vitulina), with a 3D rendition of both the extrinsic and intrinsic vibrissal muscles using 2.5% Lugol's iodine as a contrast agent (Elder et al., 2024). When supplemented with proper analysis, the data provided a thorough understanding of these muscles while indicating how they are responsible for vibrissal protraction underwater in these seals.

CT scans, along with surface scanning, were used to develop digital 3D (volume) renditions of the rotator cuff and teres major muscles in a range of hominoids, including gorillas (Gorilla gorilla) and bonobos (Pan paniscus), with further comparison to humans (van Beesel et al., 2025). The muscular surface areas (cm²) and volumes (cm³) were obtained through digital rendition and compared to similar data from real-life dissections. Diverse softwares were used in this study, such as 3D Slicer and computer-aided designing tools (e.g., Artec Studio 18 and Autodesk Maya), which can be an enriching resource for anatomy researchers. One of the major findings was a significant correlation between the areas of muscular origin and their volumes. Such tools can be useful in paleontological reconstructions in the absence of soft tissue in recovered specimens.

Given the global rise in musculoskeletal diseases, it is prudent to recognise the interest in using advanced imaging techniques on the musculoskeletal system. However, one of the major challenges with imaging, especially in the skeletal system, is the presence of calcified materials and dense tissue matrix that impede light penetration. Hence, researchers should consider combining various imaging modalities and harness the benefits from each of their strengths. Such a strategy was highlighted in an article where a workflow was proposed with a sequence of multiple imaging techniques, viz., micro-CT, MRI, second harmonic generation imaging, passive transparent lipid-exchanged acrylamide-hybridised rigid imaging tissue hydrogel, multiphoton and confocal microscopy, light sheet microscopy, brightfield microscopy, and wide-field fluorescence microscopy, used systematically to reveal structural information on a set of murine calvaria (König et al., 2025). The availability of such a workflow would appeal to researchers investigating the musculoskeletal system and guide future research endeavours to heighten their impact.

Other than CT, magnetic resonance imaging (MRI)-based non-invasive techniques, such as diffusion tensor imaging (DTI)—which assesses the magnitude and directionality of the (anisotropic) diffusion of water molecules—are also useful for a robust and reproducible understanding of anatomical structures (Assaf & Pasternak, 2008). DTI has been particularly popular in neuroanatomy research due to its ability to image white matter and has been used extensively in understanding cerebral anatomy and its functions, allied to tractography with applications in clinical medicine (Mukherjee et al., 2008; Podwalski et al., 2021). Currently, there is a significant niche for DTI applications in diseases like multiple sclerosis (Lopez-Soley et al., 2023), Parkinson's disease (Zhang & Burock, 2020), cerebral trauma (Jang & Cho, 2022), schizophrenia (Stämpfli et al., 2019), autism (Hrdlicka et al., 2019), ageing (Sullivan & Pfefferbaum, 2006), and neurodevelopmental disorders (Uddin & Karlsgodt, 2018).

A study included in this issue conducted on a similar line of application created an atlas of workable regions of interest on a DTI-derived dataset from 49 healthy subjects, where the diffusion of water molecules was estimated in 32 different directions (Dauleac et al., 2024). With further analysis, the authors were able to trace important cerebral tracts, including the spinocerebellar, corticospinal, corticoreticular, and rubrospinal tracts. The ensemble data resulted in a workable protocol that is helpful not only in generating new data in a species, but with the numerical read-outs extracted through quantification, it would also enable comparing different species.

Another study strived to expand the scope of DTI beyond neuroanatomical domains—which traditionally have remained its forte—into the cardiovascular system (Tornifoglio et al., 2025). It used DTI to understand and compare the tissue microstructures of ex vivo healthy, aneurysmal and type B chronic dissection (human) aortae, while a range of parameters (e.g., anisotropy, mean diffusivity, helical angle, tractography) were assessed. These measurements were further validated by comparing them to histological findings. Although still at a preliminary stage, such an opportunity for broadening the uses of DTI beyond neuroanatomy domains warranted urgent dissemination.

3D MRI was also effectively used in another study to investigate cerebral neuroplasticity (Spani et al., 2024). Here, the authors capitalised on the Generalised Procrustes Surface Analysis (GPSA) technique to assess the cortical surface areas in 40 healthy human volunteers subjected to quadrato motor training for 6 and 12 weeks, while the cerebral MRI data (with 3D rendering) were compared to six healthy volunteers unexposed to such training. The study mainly focused on the supplementary motor cortex, visual cortex, inferior frontal gyrus (with the anterior insula), and inferior parietal lobule. MRI data analysis demonstrated the ability of the GPSA-based approach to understand cortical surface alterations with potential for clinical applications. Furthermore, such quadrato motor training, with some modifications, can be used in neuroplasticity-related investigations on non-human vertebrates, especially in primates and avian species.

An in-depth transmission electron microscopy investigation was carried out on mice pulmonary tissue to understand the fabric of the glycocalyceal fluidic alveolar lining layer deposited on top of both the Type I and Type II alveolar epithelial cells (Gluhovic et al., 2024). Given the importance of glycocalyx in pulmonary diseases, including chronic obstructive pulmonary disease and (lower) respiratory tract infections (Timm et al., 2023)—with an unfortunate scarcity of adequate mechanistic insights—such studies fill out the current gap in comprehension of the glycocalyceal structure at nanoscale resolution and establishing adequate standard operating protocols in the field. After aldehyde tissue fixation, the authors compared alcian blue, ruthenium red, and lanthanum nitrate staining protocols. Furthermore, target-specific staining for sialic acid and fucose residues was achieved with Sambucus nigra lectin and Ulex europaeus agglutinin I, respectively. While various staining techniques revealed the ultrastructure of the alveolar glycocalyx layer to different granularity, this study would be a delight for electron microscopy enthusiasts with an interest in its use in soft tissue structures.

Other than X-ray-based microscopy, the Special Issue reported some excellent progress with light microscopic studies pivotal in revealing tissue information in both qualitative and quantitative terms. A major challenge during light microscopic tissue examination is the limited depth at which the light beams penetrate the sample. In an interesting technological innovation (Denyshchenko et al., 2024), an oblique light sheet was created by passing a laser beam through a glass microtome knife with a microscope mounted over the microtome. The generated light sheet enabled fluorescence microscopy in both liver tissue spheroids and green fluorescent protein-tagged Caenorhabditis elegans. The authors reported successful imaging up to a depth of 500 μm with adequate resolution in these spheroids, a remarkable improvement to confocal microscopy where reaching a depth of even 50 μm inside tissues is often challenging. The technique was also used to image the cellular nuclei and tight junctions of the spheroids with excellent 3D rendering. Similar success was also repeated in the C. elegans model. When combined with electron microscopy, such a sophisticated light microscopy technique, strengthened with unprecedented tissue permeation, can be a powerful tool for comparing various species or identifying their similarities (or their lack thereof) with 3D modelling.

In the light microscopic evaluation of tissue samples, two-photon microscopy, a non-linear microscopic technique, has enjoyed popularity both due to its ability to gauge higher depths within the tissue and limited phototoxicity (Helmchen & Denk, 2005; Rowlands et al., 2019). Its scope in biomedical imaging was further enhanced in a study, where in addition to stereoscopic fixation, it was used first for 3D visualisation and then to demarcate the cerebral cortical arteries from veins (with identification of blood flow direction and vasculature branching patterns) in awake mice (Liu et al., 2024). The study included normal animals who were compared to another group with induced (cerebral) thrombosis. Such studies can be useful to compare the dynamics of arterial and venous cerebral circulation in different species, and the knowledge can be beneficial in addressing neurological disorders like stroke.

Another study utilised fluorescence microscopy for 3D visualisation of the nerve fibres in a (foetal) uterus by fluorescence micro-optical sectioning tomography (Deng et al., 2024). The authors stained the uterine nerves initially with a polyclonal antibody followed by exposure to goat anti-rabbit IgG labelled with Alexa Fluor 488 dye. It revealed the radial distribution of nerves, with vine-like patterning, as they extended from the myometrium to the endometrium. The researchers also demarcated different uterine zones based on nerve fibre density: the cervix displaying higher density than the uterine body. Despite being a pilot study conducted on a single sample, the microscopic platform certainly bears promise. It can be used for the comprehension of filamentous structures, such as nerves in this case, in various organs, with further opportunities for quantification and comparison.

What undoubtedly distilled out as a challenge while handling the digital imaging data derived from a wide range of imaging platforms is to identify and then isolate the target organs or structures from their surroundings, that is, segmentation, possibly in 3D, for further analysis, including geometric assessments (Didziokas et al., 2024; Ma et al., 2024). While there are multiple ways of segmentation, including automated algorithms, for anatomical structures with complex geometry, such as cerebral ones, manual segmentation remains a valid practice. Unfortunately, such manual segmentation can be cumbersome and difficult to grasp for inexperienced users, especially while dealing with datasets with noise, artefacts, and structures with irregular geometry. An article in this issue addressed the challenges in (manual) segmentation and provided a detailed guide on how to deal with cerebral imagery datasets, including MRI studies (Darrault et al., 2024). Additionally, it proposed a roadmap while conducting manual segmentation that, with further refinement and adaptation, can evolve into a routine protocol for standardised data collection, analysis, and comparison with an emphasis on anatomy research.

Upon segmentation of anatomical structures, manual or automated, morphometric analysis often follows as a logical step, where acquiring a robust geometric segmentation is essential to extract useful information. To do so, an accurate placing of landmarks on the segmented structures is necessary, and fortunately, advanced automated landmarking pipelines, such as MALPACA, have facilitated it. In a study reported here (Thomas & Maga, 2025), 3D micro-CT scans of 425 mice hemi-mandibles were collected with further segmentation of the mandibles and polygon meshing followed by landmark placement and point-to-point correspondence using an enhanced MorphVQ Deep Functional Map tool. The study estimated the root mean square error of these landmarks and compared them to the ground-truth ones, simultaneously comparing the model performance against MALPACA. The refined MorphVQ pipeline was time-saving, and its unsupervised nature made it compatible across different datasets. The account also presented an in-depth 3D morphometric analysis, including deep learning systems, that might not only speed up the current state-of-the-art platforms but also augment the quality of obtained data—a necessity for further progress in the field—and, importantly, presented a template that future researchers can use and build upon.

While the developments in biomedical imaging continue to add more knowledge, their integration into medical pedagogy is crucial. As most of the current microscopic data are now derived and stored in digital formats, fortunately, it presents the possibility of student accessibility and coupling to artificial intelligence (AI) platforms (Ryan et al., 2024). An interesting study explored such possibilities and developed an extension to the QuPath software—widely used for the analysis of histopathology slides—known as QuPath Edu, and tailored it to learning environments by adding extra features, such as remote viewing and annotation of the slides, with linked questions, annotations, or media files (Yli-Hallila et al., 2024). The authors also created an alternate web-based platform for accessing QuPath with additional features, including (but not limited to) coupling with AI tools. A survey among the students confirmed their satisfaction and the efficacy of such emerging platforms in anatomy teaching.

Despite the progress achieved in biomedical imaging—with its necessity of integration into medical teaching realised well—the cost-effectiveness issue remains valid with many of the reported techniques, including CT and MRI, known to be expensive, requiring sophisticated instruments and trained personnel. Alternate and yet effective techniques of imaging and visualisation of anatomical structures, including whole body imaging with 3D rendition on embalmed specimens, would be thus welcome in anatomy teaching. A paper showcased the capability of photogrammetry as an alternative to CT or MRI in developing digital 3D anatomy models by first converting a series of 2D images into 3D mesh through the creation of a sparse point cloud, followed by decimation and mesh smoothing (Durrani et al., 2024). The photorealism of these 3D models, enhanced by real-life colouration, is commendable, and their scope can be further widened with online sharing through data cloud service. Photogrammetry is applicable in multiple species, potentially delivering a robust and visually arresting teaching tool that simultaneously stands affordable, informative, and collaborative.

The role of biomedical imaging in anatomical sciences is going through an overhaul, and the emergence of advanced platforms has further added to the discourse. This Special Issue has compiled some of the powerful imaging techniques with their applications, present or predicted, across a wide variety of (vertebrate) species, while the hope is that the (anatomy) research community will find it useful and rate it highly. In many ways, imaging science, not in isolation for anatomy only but as a discipline in its entirety, is at a crossroads now. There is no debate that plenty of novel platforms have emerged in recent times that have pushed the boundaries of biological (tissue) imaging. While this is certainly a welcome development, it also leaves us with plenty of unsettling questions. How shall we proceed from here? And how do we negotiate with the challenges, such as AI, that certainly leaves us in a dilemma: to be or not to be? How shall we filter out these new-age advances while excluding most, if not all, of their weaknesses? Unfortunately, there are no easy solutions or fixes for these frustrating riddles.

The Special Issue, in posterity, would leave some answers and maybe some loose yet visible cues for the researchers to brainstorm in that direction. The current generation of imaging researchers has tackled some of the challenges passed on by their predecessors and is now leaving a few more unresolved puzzles for future generations to solve in this eternal riverine flow of knowledge. The Special Issue, with hard work from its editorial and production teams, the authors, and last but not least—the anonymous reviewers (whose contributions often go unnoticed)—has strived to capture a topic that is relevant, timely, and exciting. While the onus of gauging the impact and contribution of the Special Issue lies on the research community, the dream is that, even in a distant future, it will continue guiding the anatomists in their pursuit of knowledge and excellence, like a lighthouse—with a flame that may be faint but still unvanquished, bright, and throbbing with vigour—standing at a bend in the river.