Micro-CT-based analysis of the murine osteochondral unit: a review and recommendations for standardized subchondral bone assessment

Abstract

Micro-CT enables high-resolution, 3D, and non-destructive visualization of bone microarchitecture in small preclinical animal models. It is widely applied to assess mineralized tissues in joints, making it a valuable tool for monitoring disease progression. Murine models are particularly prevalent in joint research due to their cost-effectiveness and disease tunability. Established guidelines for murine bone microstructure assessment provide a standardized framework for morphometric analysis and facilitate cross-study comparisons. However, these recommendations were developed for trabecular and cortical bone in long bones, and may not fully address the unique characteristics of mineralized tissues within joints. This review focuses on murine studies, and aims to: (1) examine reported methodologies for volume of interest selection; (2) outline commonly evaluated parameters; and (3) propose adaptations to expand existing guidelines for quantitative analysis of the osteochondral unit. In the absence of specific guidelines for osteochondral unit analysis, reported volumes of interest vary considerably and are influenced by anatomical differences across strains, sexes, ages, and disease stages. Many studies target the subchondral bone plate (SBP); however, the spatial resolution of desktop microCT is insufficient to distinguish SBP from calcified cartilage; hence, they are measured as 1 entity. To enhance reproducibility and comparability, we recommend standardized volumes of interest that include all mineralized tissues from the articular surface to the growth plate. Given the structural differences between SBP and subchondral trabecular bone, a consistent method for defining their boundary is strongly recommended. Data should be reported using robust, easily implementable metrics such as volume and thickness, with additional parameters (eg, porosity, mineral density) included as appropriate. Commonly used metrics such as SBP thickness and volume should continue to be used. When SBP and calcified cartilage cannot be distinguished, we recommend using the subchondral mineralized plate nomenclature.

Graphical Abstract

Graphical Abstract.

Introduction

Osteoarthritis (OA) is a chronic, degenerative joint disease, combining bone deterioration and cartilage thinning¹ resulting in pain, immobility, and reduced quality of life. Osteoarthritis has a higher prevalence in the elderly. With the aging of the global population, it is projected that by 2050, OA will affect up to double the population currently affected, heavily burdening healthcare systems.² The disease mainly impacts the weight-bearing joints of the lower limb, with the knee being the most commonly affected.³ The etiology of OA remains unclear, with ongoing debates regarding whether cartilage degradation⁴ or subchondral bone (SB) remodeling acts as the primary initiator of the disease. Despite extensive research on the mechanisms of OA pathogenesis⁵ and potential treatments for cartilage defect repair,⁶ a clear understanding of disease onset and progression has yet to be found. Consequently, OA treatments are largely focused on pain relief,⁷ improving overall health,⁸ and retaining functional mobility,⁹ rather than addressing the underlying OA pathophysiology.

Research to uncover the complex pathogeny of OA is an ongoing effort and relies mainly on preclinical models that develop the disease in a human-like way.^10,11 Among them, small rodents, such as mice and rats, are the most common subjects^12,13 due to relatively low costs¹⁴ and the ability to induce OA in a controllable and reproducible way. This allows to study the disease from its onset and enables the monitoring of degeneration over time.^10,15

The osteochondral unit (OCU), which comprises articular cartilage (AC) and the underlying SB, is a key region to evaluate. These 2 tissues, essential for joint health, are linked through mechanical and biochemical cross-talk and both undergo OA-related changes during disease progression.¹⁶ Furthermore, AC and SB show abnormal remodeling, even before the manifestation of clinical symptoms.^16,17

Articular cartilage has a hyaline component and a calcified component, called calcified cartilage (CC). The CC acts as a transitional layer between the hyaline cartilage (HC) and SB and plays a key role in communication between the tissues. Despite its relevance to joint function, CC is often overlooked, especially with medical imaging-based assessment.¹⁸ The SB can also be divided into 2 components, the SB plate (SBP) and the underlying subchondral trabecular bone (STB).

Micro-CT is widely used to quantify SB, because it allows non-destructive, 3D assessment of bone microstructure at the micrometer scale,¹⁹ revealing the presence of bone sclerosis, abnormal bone growth, osteophytes, and bone cysts.²⁰ MicroCT studies in rodent bones often refer to a set of guidelines by Bouxsein et al.,²¹ which ensure consistency across multiple research groups. The guidelines, which are widely adopted in the musculoskeletal imaging field, give recommendations on data acquisition, processing, and analysis steps. More specifically, they provide relevant recommendations on the minimum set of variables that need to be reported and on the volume of interests (VOIs) where these parameters should be evaluated. The recommendations were made with the evaluation of cortical and trabecular bone of long bones in mind. However, in diseases like OA, the analysis is primarily focused on the joints, rather than at the shaft of the bones. Hence, the guidelines do not specifically address the challenges of analyzing the OCU in detail, particularly the distinction between CC and SBP, and they do not provide specific VOI recommendations for joint regions. To accommodate these differences, modifications of the recommended set of parameters have been proposed.^22–26 Although helpful in answering research questions in individual studies, these modifications can undermine the standardization that the guidelines set out to achieve.

Therefore, the aim of this paper is to provide a comprehensive overview of how the SB region has been analyzed in micro-CT studies by: (1) examining reported methodologies for VOIs selection, including evaluating the extent to which CC is explicitly included; (2) providing an overview of study methodologies, including VOI choice and parameters assessed, and considering how they align with current guidelines or need to be adapted for SB; and (3) providing recommendations for a suitable VOI and a set of quantitative morphometric parameters appropriate for analyzing the mineralized tissues of the OCU, SBP, and CC, to aid standardization and improve cross-study comparisons. This is preceded by a brief overview of the OCU structure, highlighting the relevance of each tissue involved.

Structure of the OCU

The OCU is a functionally integrated structure where AC and SB interact dynamically. It plays a key role in biomechanical load distribution, with cartilage relying on bone for support and resilience. Mechanical and biochemical signaling between the 2 regulates joint adaptation and disease progression.²⁷ For instance, in OA, thickening of the SB (sclerosis), accelerates cartilage degeneration,²⁸ while cartilage degradation releases inflammatory mediators that alter the bone remodeling process, contributing to joint diseases.²⁹ Although cartilage is avascular and relies mainly on synovial fluid for nutrients, its deeper layers exchange metabolites through SB.

On a structural level, the OCU has a somewhat similar organization in rodents and humans, although dimensions are substantially larger in humans. In mice, HC is around 30 μm and CC is approximately equal to this (Figure 1), while in humans, HC is around 2 mm and CC is around 100 μm.^30,31

Figure 1.

Organization of the osteochondral unit in mice. The articular cartilage includes the hyaline cartilage, the tidemark, and the calcified cartilage. Below the articular cartilage is the subchondral bone, divided into the plate and the trabecular bone tissue [created with BioRender.com].

Subchondral bone

In synovial joints, SB is located beneath the cartilage layer. It is divided into an SBP and STB. The SBP is a thin cortical bone layer;³² hence, it is sometimes referred to as the subchondral cortical plate.³³ In human knees, it has a non-uniform thickness ranging from 300 to 1500 μm,^34,35 whereas in mice and rats, it measures around 150 and 300 μm, respectively.³⁶ The SBP acts as a transitional layer between soft AC and the trabecular bone at the epiphyseal region and it adapts according to the loading directions.³⁷ The SBP functions to distribute joint loads and maintain joint shape.^38,39 In joint diseases, some of the earliest manifestations involve changes in SBP structure,⁴⁰ such as plate thickening and the development of osteophytes and bone cysts.⁴¹ This makes the SBP a critical tissue for investigating both disease onset and development, as well as for potential treatment strategies.¹²

Articular cartilage

Articular cartilage is a highly specialized connective tissue that provides a smooth and lubricated surface to facilitate load transmission across the joint.⁴² It is composed of chondrocytes embedded within a dense extra cellular matrix, made up of collagen, proteoglycans, and water.

Articular cartilage shows a layered structure, comprising 4 zones: superficial, middle, deep, and calcified.⁴³ The first 3 layers consist of HC, which is separated from the underlying CC by the tidemark, a structure that plays an important role in joint health.⁴⁴ These layers exhibit a gradual increase in mechanical stiffness and resistance to deformation from the joint surface to the SB. In particular, the deep zone provides resistance to compressive forces due to its higher proteoglycan content, vertical collagen arrangement, and increased chondrocyte density.⁴⁵

In rodents, the distinction between superficial, middle, and deep zones is less obvious, due to the small size of the region.³⁰ Another key feature of AC is its avascular, aneural, and alymphatic nature, which limits its capacity for healing and repair, especially during pathologies such as OA.⁴⁶

Calcified cartilage

The CC serves as a transitional zone between the softer HC and the stiffer SB. Calcified cartilage is separated from the overlying HC by the tidemark, an undulating basophilic 3D structure,⁴⁷ and from the SB by the cement line, a structural interface that is more distinct in humans than rodents.

Calcified cartilage shows hybrid characteristics, sharing features of both cartilage and bone. Cartilage cells (chondrocytes) are present, but in a hypertrophic state, reflecting the transitional nature of this layer. Calcified cartilage retains some collagen type II, though in significantly lower amounts than HC, ~20% vs 60%, respectively. Simultaneously, CC displays mineralized properties similar to bone, with hydroxyapatite constituting ~65% of its dry weight, whereas in SB this is 85%.⁴⁸

Functionally, the wavy structure of the tidemark is thought to facilitate force transmission between softer cartilage and stiffer SB.⁴⁹ Mechanically, CC acts as a biomechanical bridge between these tissues, which is evident in its elastic modulus that is 2 orders of magnitude higher than HC, and 1 lower than SB.⁵⁰

In joint disease models such as OA, alterations in CC have been observed. These include tidemark multiplication⁵¹ and the invasion of blood vessels from SB, which disrupt normal remodeling and compromise the integrity of the OCU.⁵²

Visualizing the OCU with micro-CT

Micro-CT is a well-suited imaging technique for visualizing mineralized tissues because X-ray absorption strongly correlates with mineral content. This property, along with the high spatial resolution, has made micro-CT an effective tool for assessing bone in rodent OCU studies, providing non-invasive and non-destructive 3D bone assessment.⁵³ Several commercially available systems have been developed to evaluate bone, including for in vivo studies.²¹

Given the similar mineral content of CC and SBP, these layers cannot be distinguished using desktop micro-CT.⁵⁴ This is due to limitations in both contrast and spatial resolution which typically range from 5 to 10 μm (Figure 2A). This challenge is more pronounced in rodents, particularly in mice, where the thin mineralized tissue further complicates the identification of the 2 layers.^39,55 High-resolution scans can improve the visualization of CC and SBP, as well as fine structures such as lacunae (Figure 2B), but these are often conducted ex vivo due to longer scanning times and higher radiation doses, which raise animal welfare concerns.^18,56

Figure 2.

(A) Ex vivo micro-CT image of a mouse knee joint at 10 μm resolution. (B) High-resolution micro-CT image (0.7 μm) of the osteochondral unit from a mouse tibia, allowing improved visualization of fine structures (image courtesy of Anik Banerjee, KU Leuven). Abbreviations: CC, calcified cartilage; SBP, subchondral bone plate; STB, subchondral trabecular bone.

Synchrotron-radiation micro-CT enable submicron-scale visualization and can discern levels of mineralization, improving the delineation between CC and SBP (Figure 3).^57–59

Figure 3.

Mouse knee joint image using X-ray based techniques, adapted from⁵⁷. (A) Whole-joint desktop micro-CT image, in which CC cannot be distinguished from SBP. (B) Whole-joint synchrotron-based phase contrast CT image of the same joint. (C) Synchrotron-based phase-contrast CT image acquired on the same system. The magnified region highlights the distinction between CC, HC, and underlying SBP and STB. The red dotted line represents the position of the HC based on histological analysis. Abbreviations: CC, calcified cartilage; HC, hyaline cartilage; SBP, subchondral bone plate; STB, subchondral trabecular bone.

Due to the much lower X-ray absorption of HC, it is not directly visible on micro-CT. To enable soft tissue assessment, contrast agents, typically positively or negatively charged, can be introduced to enhance tissue contrast (Figure 4).⁶⁰ Because the cartilage extracellular matrix has a negative charge due to its proteoglycan content, the distribution of these agents depends on electrostatic interactions. Positively charged (cationic) contrast agents accumulate within the tissue in proportion to its proteoglycan content, thereby increasing attenuation and allowing the cartilage to be directly visualized with micro-CT (Figure 4A).^60,61 In contrast, negatively charged (anionic) agents are excluded from the matrix and provide indirect morphological information by outlining the cartilage boundaries (Figure 4B).⁶² Although contrast agents can effectively help with the visualization of non-mineralized tissues, they cannot distinguish CC from the SBP.⁶⁰

Figure 4.

(A) Contrast-enhanced micro-CT imaging of a mouse joint with a positively charged contrast agent from⁶⁰ showing the layer of hyaline cartilage brighter than the background. The white box gives a magnified view of the interface with the mineralized tissues. (B) Contrast-enhanced microCT imaging of a mouse joint with a negatively charged contrast agent. It is possible to observe the layer of hyaline cartilage as a dark band, indicated by the arrow, compared to bone and the joint space that show higher attenuation. Image adapted from⁶² (scale bar not available).²³

Despite advancements in medical imaging technologies, histology remains the gold standard for the evaluation of OA progression and tissue-level pathological changes.⁴³ It provides cellular-level detail and a clear delineation of cartilage from SB using tissue-specific stains. In histology, the CC layer is identified as part of the cartilage compartment, whereas in micro-CT it is considered together with bone. While histology is a destructive technique, and is limited to the assessment to 2 dimensions, micro-CT enables non-destructive 3D assessment of the mineralized tissues but lacks soft-tissue contrast. Combining histology with micro-CT can provide complementary insights into joint structures (Figure 5).

Figure 5.

Mouse microCT images of the sagittal (A) and frontal plane (B) with the corresponding histological section stained with toluidine blue (C) from⁶³ (scale bar not available).

Evaluated literature

A literature search was conducted on PubMed to assess the current state of research focusing on how VOIs are selected and morphometric parameters are reported in micro-CT studies. Specifically, we evaluated studies that performed quantitative morphometric analysis (QMA) of mineralized tissues at the knee OCU, considering only papers on murine preclinical models published after the release of the 2010 guidelines.

Table 1 summarizes studies published up to 2025 that cited the existing guidelines.²¹ When multiple studies from the same research group addressed a similar question, the most recent and comprehensive was kept to avoid redundancy. This set of papers was further complemented with additional relevant studies that did not necessarily reference the guidelines, identified during related murine knee OCU literature searches using “MicroCT” and “Subchondral plate” as key words, in order to provide a more comprehensive overview of the current practice.

Table 1.

Selected rodents studies that analyze the SBP changes with morphometric indices.

Authors	Scanning parameters			VOI choice	STB/SBP separation	QMA parameters	Species
Guidelines according to Bouxsein et al. 2010 21	Voxel size	X-ray tube peak energy	Exposure time (ET)	Based on anatomical landmarks or bone length percentage	No recommendation for subchondral cortical/trabecular segmentation	Tt.Ar, Ct.Ar, Ct.Ar/Tt.Ar, Ct.Th	Rodents
Adebayo et al. 201739	10 μm	55 kVp	600 ms (IT)	SBP is manually contoured	Global thresholds method	SBP.Th.Med/Lat, TMD	B6 mice
Bergman et al. 202564	8.9 μm	Not specified	Not specified	Manually contoured SCB from images of the medial/lateral femoral condyles	Manual delineation of the SBP and STB	SCB.Th, TMD	B6 mice
Besler et al. 201765	12 μm,	50 kVp	900 ms (IT)	Identification of cortical bone with morphological operations	Identification of STB/SBP compartments with morphological operations	Ct.Th; bone volume fraction, %BV	B6 mice
Botter et al. 201166	9 μm	40 keV	2356 ms (ET)	Fixed volume (500 μm × 750 μm × 350 μm), measured on the medial/lateral tibial condyles.	Identification of cortical bone based on bone volume fraction	SBP.Th.Med/Lat, SBP.V.Med/Lat, BMD	B6 mice
Brown et al. 202067	Not specified	9.9 W, 90 μA	200 ms	Tibial epiphysis, Tibial subchondral bone excluding the trabeculae	VOI subtraction	SBP.V SBP.Po	Lewis rats
Butterfield et al. 202168	2 μm	70 kVp	1500 ms (IT)	VOI scaled according to the size of the tibia, placed at the same position on medial/lateral side according to the plateau edges	No separation	SB.BV/TV, SB.BMC, SM.TMD	WT mice
Chan et al. 202333	2 μm	70 keV	1500 ms	Manually contoured SBP from images of the medial/lateral femoral condyles	Manual delineation of the SBP and STB	SBP.Th.Med/Lat, SBP.V.Med/Lat, BMDD	B6 mice
Chen et al. 202369	9 μm	60 kVp	1000 ms (ET)	Manually contoured SBP from images of the medial/lateral tibial plateau	Automatically separated using CTAn software	SBP.Th.Med/Lat	STR/ort mice
Christiansen et al. 201270	15 μm	55 kVp	900 ms (IT)	VOI 600 μm distal to the proximal tibia, excluding the STB	Manual delineation of the SBP and STB	Cortical thickness and bone tissue mineral density	B6 mice
Das Neves Borges et al. 201771	5 μm	50 keV	1600 ms	Fixed volume (500 μm × 750 μm x 350 μm), placed on the center of medial/lateral tibial condyles	Identification of cortical bone based on macro-porosity	SBP.Th.Med/Lat, SBP.V.Med/Lat, BMD	B6 mice
Fouasson-Chailloux et al. 202172	5 μm	50 kVp	620 ms	Subchondral bone of the tibial epiphysis	No explicit indication of the choice	SBP.Th	B6 mice
Guss et al. 201973	10 μm	55 kVp	600 ms (IT)	No explicit indication of the choice	STB extending from the end of the SBP to the start of the GP	SBP.Th, TMD	B6 mice
Hamann et al. 201474	6 μm	55 kVp	400 ms (IT)	1 mm width pre-segmented VOI	Identification of STB/SBP compartments with automatic image analysis algorithm73	SBP.Th, SBP.Po, BMD	Sprague-Dawley rats
Higuchi et al. 201775	5 μm	Not specified	Not specified	Fixed geometry 500 μm mediolateral width, 1.0 mm ventrodorsal length on the medial tibial plateau	No explicit indication of the choice	SBP. BV/TV, SBP.Th	B6 mice
Hildebrandt et al. 202376	5.1 μm	70 kVp	Not specified	Manually contoured SBP from images of the medial tibial plateau	Manual delineation of the SBP and STB	Ct.V, Bone and Pore density, Avg.Po	WT mice
Hislop et al. 202278	10 μm	70 kVp	200 ms (IT)	Manually contoured SBP from 1 mm images of the proximal tibial epiphysis	No explicit indication of the choice	SCB.Th.Med/Lat; SCB.TMD.Med/Lat	B6 mice
Hu et al. 201881	9 μm	50 kVp	Not specified	Subchondral bone medial compartment	Not specified	SBP.Th	B6 mice
Huang et al. 201724	10 μm	Not specified	Not specified	Manually contoured SBP from images of the medial tibial condyles	Manual delineation of the SBP and STB	SBP.Th.Med	B6 mice
Iijima et al. 201582	21 μm	43 kV	7 min (scan time)	Fixed geometry mediolateral width of 500 μm and a ventrodorsal length of 1 mm in the frontal plane in the medial tibia	No explicit indication of the choice	SBP.BV/TV.Med, SBP.Th.Med	Wistar rats
Ji et al. 202083	9 μm	50 kV	Not specified	Anterior subchondral bone of the medial tibial plateau	No separation	BV/TV on the whole region	B6 mice
Jia et al. 201879	6 μm	Not specified	Not specified	Manually contoured SBP from images	Automatically separated according to thresholds	SBP.Th	Rosa-Tomato mice
Kihara et al. 201784	20 μm	90 kVp	900 ms (IT)	SBP VOI extends 150 μm distal to the proximal point of the tibia and excluding the underlying trabecular bone	STB is manually contoured	Cortical thickness, bone tissue mineral density	Homozygous p21 mice
Kim et al. 201385	18 μm	40 keV	2356 ms	Manually contoured SBP from images	STB is manually contoured	3D SBP thickness	B6 mice
Li et al. 202386	20 μm	90 keV	Not specified	Fixed geometry (500 μm × 800 μm × 400 μm) on medial plateau	Automatically separated according to thresholds	BV/TV, BMD	B6 mice
Liu et al. 202580	10 μm	70 kVp	350 ms (IT)	VOI selected following65	Separation done using previously developed algorithm62	Ct.Th, Ct.Po	B10 mice
Lockwood et al. 201487	10 μm	55 kpV	900 ms (IT)	500 μm distal to the proximal point of the tibia, excluding the STB and any osteophytes	No explicit indication of the choice	SBP.Th, BMD	B6 mice
Luna et al. 202088	10 μm	55 kVp	600 ms (IT)	No explicit choice of the SBP VOI	Automatically separated according to thresholds	SBP.Th, TMD	B6 mice
Maertz et al. 201689	12 μm	70 kVp	250 ms (IT)	Manually contoured SBP from images	Automatically separated using erosion function	SBP.Th.Med, SBP.Th, BV/TV, TMD	Lewis rats
McCann et al. 201790	20 μm	80 kVp	4500 (ET)	500 μm in mediolateral length and 1150 μm, ventrodorsal length, centered on the medial tibial plateau	Manual delineation of the SBP and STB	SBP.Th.Med, BMD	CD-1 mice
McCulloch et al. 201991	4.5 μm	50 kV	Not specified	Fixed geometry (500 μm × 900 μm × 900 μm) in the center of load in the medial and lateral tibial plateau	Manual delineation of the SBP and STB	SBP.BV/TV	B6 mice
Menges et al. 202492	17.6 μm	65 kV	Not specified	Fixed geometries selected on tibia and femur	No separation	BV/TV	Lister Hooded SPF rats
Mohan et al. 201125	8.7 μm	60 kVp	4.7 s	Fixed geometries selected 1.5 mm medio-lateral in width, 2.5 mm in length from the posterior side on the medial/lateral tibial plateau	Identification of cortical bone based on bone volume fraction from63	SBP.Th, SBP.Po	Wistar rats
Oliviero et al. 202322	4.35 μm	49 keV	1180 ms	Fixed volume cross-section (500 μm × 500 μm), extended along the axis of the long bone to the growth plate, measured on the medial/lateral tibial plateau	Manual delineation of the SBP and STB	SBP.Th.Med/Lat	B6 mice
Osterberg et al. 201793	9 μm	49 keV	Not specified	Defined by the cortical boundary surface facing the joint space and the transition zone between the cortical and trabecular bone within the epiphyseal section	Not specified	SBP.Th.Med on tibia and femur condyle	STR/ort and B6 mice
Poudel et al. 202494	9.7 μm	100 kV	Not specified	Proximal tibia	Not specified	SBP.Th, BMD	UM-HET3 mice
Poulet et al. 201595	5 μm	40 kV	Not specified	Manually contoured SBP from images on medial and lateral side of both femur and tibia	Manual delineation of the SBP and STB	SBP.Th.Med/Lat	CBA mice
Reece et al. 201896	16 μm	45 kVp	200 ms (IT)	Manually contoured SBP from images on medial central and lateral third of the tibial plateau	Manual delineation of the SBP and STB	SBP.V, SBP.Th, SBP.Po, BMD	Lewis rats
Rosch et al. 202297	3.4 μm	90 kVp	1500 ms (IT)	Semiautomatic segmentation process to locate the CC and the SBP layer	Fixed geometry for STB analysis placed the lower margin of the SBP	SBP.Th	B6 mice
Samvelyan et al. 202198	5 μm	50 kV	Not specified	Hand-drawn of the SBP and epiphyseal trabecular bone for each tibial lateral and medial compartments were selected	Manual delineation of the SBP and STB	SBP.Th.Med/Lat, SBP.BV/TV	B6 mice
Takebe et al. 201599	21 μm	45 kVp	300 ms (IT)	Manual delineation of the SBP on the tibial plateau	Manual delineation of the SBP and STB	SBP.Th	B6 mice
White et al. 2022100	12.25 μm	70 kV	Not specified	Fixed cubed geometry (1.04 mm) centered in each of the medial and lateral regions of the femur and tibia	Manual delineation of the SBP on medial/lateral tibial plateau	SBP.Th.Med/Lat on both femur and tibia SBP.Po.Med/Lat on both femur and tibia	Long Evans rats
Xu et al. 2015101	Not specified	Not specified	Not specified	Subchondral bone medial compartment	Not specified	SBP.Th	DBA/1 J mice
Zhou et al. 2022102	12 μm	285 kVp	200 ms (IT)	Epiphyseal cortical bone chosen automatically using a watershed algorithm	Manual delineation of the SBP and STB	Ct.Ar, Tt.Ar, SA/V,103 surface curvature, TMD	DBA/1 J mice
Ziemian et al. 2021104	10 μm	55 kVp	600 ms (IT)	Manual delineation of the SBP on the tibial plateau	Not specified	SBP.Th	LC and pOC-ERαKO mice

Abbreviations: Avg.Po, subchondral bone plate porosity; Ct.Ar, cortical bone area; Ct.Ar/Tt.Ar cortical bone area fraction; GP, growth plate; IT, integration time; SA/V, surface-to-volume ratio; SBP or SCB, subchondral bone plate; SBP.Ar, subchondral bone plate area; SBP.Po, subchondral bone plate porosity; SBP.Th, subchondral bone plate thickness; SBP.V, subchondral bone plate volume; STB, subchondral trabecular bone; TMD, tissue mineral density; Tt.Ar, total area.

The main findings, including adherence to and deviation from the existing guidelines, are summarized in Table 1. The table also includes key guideline aspects, namely, reported scanning parameters, VOI choice, and the methodology to separate STB from SBP. In addition, information about the rat or mouse strain is specified.

Discussions

Scanning parameters

Among the information that must be reported, scanning parameters are particularly important, as micro-CT image quality is highly dependent on the chosen settings. The guidelines recommend reporting voxel size, X-ray tube peak energy, and exposure time (ET). However, as much as 36% of the considered studies did not report all recommended scanning parameters (Table 1). For instance, some studies only report voxel size.^24,64,75,83 Complete reporting of voxel size, X-ray energy, and ET is critical for interpreting results and evaluating radiation exposure, particularly for potential in vivo applications.⁵⁶

VOI selection

A wide range of VOI selection approaches are reported in OCU studies to account for biological variability. Some studies prefer manual VOI selection, either by contouring the entire SB region^{24,39,78,83,85,89,95,96,99,104} (Figure 6A) or by drawing a fixed-volume VOI based on anatomical landmarks, such as the center of the femoral condyles or the medial and lateral sides of the tibial plateau^{22,33,70,82,91,98,100} (Figure 6B and C). These methods allow flexibility, but they are labor-intensive and prone to operator bias, thereby limiting reproducibility.¹⁰⁸

Figure 6.

Examples of subchondral bone volume of interest selection: (A) manual from CTAnalyser software, (B and C) fixed geometry on femur condyles and tibial plateau from¹⁰⁰ (scale bar not available), (D) automatic segmentation from.¹¹¹ Legend: blue: epiphysis, yellow: growth plate, green: primary spongiosa, red: secondary spongiosa.

Volume of interests are often positioned relative to the medial and lateral sides of the joint to capture the distinct pathological progression patterns that can occur in each region. Failing to account for these differences and averaging metrics for these regions, results in a loss of information.⁵⁵ Some studies analyze only 1 side, more often the medial^{24,75,82,83,86,90} while others report data averaged across both regions^{67,70,73,74,79,85,87,94,99,101,104} or, if separated, fail to describe the method used to perform the medial and lateral cut.^33,69,78 To ensure consistency across specimens, this division should be performed according to anatomical landmarks, such as the intercondylar fossa for the femur and the intercondylar eminence for the tibia.

Fixed-volume VOIs can offer a time-efficient approach, but they may miss important pathological features, such as abnormal bone growth and osteophyte development at the edges of the femoral condyles or tibial plateau.^105,119

To address these challenges, automatic and semi-automatic methods have been developed (Figure 6D). Most semi-automatic approaches are based on interpolation algorithms, though their effectiveness depends on the sampling step; larger step sizes are faster, but less accurate.¹⁰⁶ Recently, machine-learning algorithms have shown promising results for automatic and reproducible VOI selection.^107,109,110 While powerful, these algorithms are sensitive to input data and often require training on large and diverse datasets, or specialized software expertise.

Moreover, several studies do not explicitly report their VOI choice, using vague terms such as “tibial epiphysis,”^67,72 “subchondral bone medial compartment,”^81,101 or “proximal tibia,”⁹⁴ without specifying if the selection has been performed automatically or manually. In some cases, the VOI choice is omitted entirely.^73,88 This lack of clarity further complicates comparisons and reproducibility across studies.

Separating STB from SBP

Since standard desktop micro-CT systems cannot reliably distinguish CC from SBP, these layers should be considered a single structure, which can be referred to as the subchondral mineralized plate (SMP), ensuring that the frequent inclusion of CC is not overlooked when only SBP is mentioned. The term SBP should be reserved for cases where it can be clearly separated from CC, such as with high-resolution synchrotron radiation-based micro-CT.

Separating SMP from STB is another critical but inconsistently reported step. The interface between these regions is gradual, making consistent delineation challenging.¹⁰⁹

Manual separation is often performed without detailed reporting of the criteria used,^{22,24,70,90,95,98,100,102} introducing variability and operator bias into the analysis.⁷⁷ Some studies report using automatic approaches based on image properties, such as porosity,⁷¹ morphological operations,^25,66,89 thresholding techniques,^39,79,86,88 or in-house algorithms,^69,74 though they are highly dependent on algorithm robustness and image quality. More recently, machine-learning approaches have also been applied.^111,112

The guidelines²¹ recommend reporting all steps and algorithms used during both processing and analysis. However, Table 1 reveals that a significant number of studies fail to describe this step, even when subsequent analyses of SBP and STB are performed separately.^{75,78,82,87,93,94,104} Some studies skip this separation entirely and analyze both tissues together.^83,92

Quantitative morphometric analysis

Quantitative morphometric analysis in micro-CT studies generally follows the metrics set out by Bouxsein et al.,²¹ which were originally developed for murine trabecular and cortical bone. For STB, these standard parameters are directly applicable. In contrast, parameters designed for the assessment of cortical bone, including total cross-sectional area inside the periosteal envelope (Tt.Ar), cortical thickness (Ct.Th), cortical bone area (Ct.Ar), and cortical bone area fraction (Ct.Ar/Tt.Ar) require adaptation for the SMP. The SMP differs both structurally and functionally from diaphyseal cortical bone; it has a curved, irregular geometry, and high mechanical adaptability, due to its position at the joint and the combined presence of CC and SBP.

Commonly reported measures include SBP thickness (SBP.Th) which is also referred to as SCB.Th⁷⁸ or cortical thickness.^65,70 Other frequently reported metrics are SBP volume (SBP.V), bone mineral density (BMD), and tissue mineral density (TMD). These parameters can be easily implemented in most micro-CT in-house and open-source bone analysis software.^113,114,116 Porosity assessment (SBP.Po) is often reported in rat studies^25,96,100 where the larger size of the joint allows for more detailed bone visualization. In Zhou et al. (2021) surface area-to-volume ratio (SA/V) and surface curvature are also reported.¹⁰²

Bone mineral density and TMD provide information on tissue mineralization and are useful for tracking aging and the impact of diseases such as OA and osteoporosis.¹⁹ However, these metrics are highly dependent on the X-ray source and scanning settings, which can complicate comparisons between studies.^117,118 This issue can be mitigated by reporting technical details, including the calibration process and values, as well as scanning parameters.

Conclusions

In summary, rather than adopting a single universal VOI definition, VOI should be based on anatomical landmarks, with boundaries, thickness, and spatial extent explicitly specified and demonstrated to be reproducible across subjects. In addition, we recommend performing the separation between STB and SMP using a well-described and reproducible method until a field-wide consensus is established regarding how and where these 2 tissues should be separated. Fully reporting acquisition and analysis parameters, including voxel size, X-ray tube peak energy, and ET, is crucial for ensuring consistency across studies.

Additionally, the separation between the medial and lateral regions of the joint should be performed using anatomical landmarks (eg, the intercondylar fossa/eminence) as a reference. Commonly reported QMA parameters, such as thickness and volume, should continue to be included to facilitate reproducibility and comparability. However, since the SBP cannot be reliably separated from the CC in the majority of studies using desktop micro-CT, we recommend that these parameters be reported instead with reference to the mineralized plate, specifically as SMP.Th and SMP.V, and that the term “SMP” is used instead of “SBP” in this context. Additional metrics, such as BMD/TMD, may provide additional information on tissue mineralization, but require careful calibration and detailed reporting of acquisition and processing procedures, in line with.²¹ Measures of porosity (SMP.Po) are typically feasible, at standard desktop micro-CT resolution, only in larger murine joints (eg, rats). Updating existing guidelines to include standardized VOIs and SMP-based metrics will enhance methodological consistency and improve the evaluation of subchondral changes in murine joint disease models.

Updating existing guidelines²¹ to include standardized VOIs and SMP-based metrics will enhance methodological consistency and improve the evaluation of subchondral changes in murine joint disease models.

Author contributions

Cecilia Liberati (Conceptualization, Data curation, Investigation, Methodology, Writing—original draft) and G. Harry van Lenthe (Resources, Supervision, Writing—review & editing)

Funding

This work was supported by funding from FWO and F.R.S.-FNRS under the Excellence of Science (EOS) program (EOS No. 40007553).

Conflicts of interest

All authors state that they have no conflicts of interest.

Data availability

No datasets were generated or analyzed during the current study.