A Comprehensive Review on MRI-based Knee Joint Segmentation and Analysis Techniques

Using magnetic resonance imaging (MRI) in osteoarthritis pathogenesis research has proven extremely beneficial. However, it is always challenging for both clinicians and researchers to detect morphological changes in knee joints from magnetic resonance (MR) imaging since the surrounding tissues produce identical signals in MR studies, making it difficult to distinguish between them. Segmenting the knee bone, articular cartilage and menisci from the MR images allows one to examine the complete volume of the bone, articular cartilage, and menisci. It can also be used to assess certain characteristics quantitatively. However, segmentation is a laborious and time-consuming operation that requires sufficient training to complete correctly. With the advancement of MRI technology and computational methods, researchers have developed several algorithms to automate the task of individual knee bone, articular cartilage and meniscus segmentation during the last two decades. This systematic review aims to present available fully and semi-automatic segmentation methods for knee bone, cartilage, and meniscus published in different scientific articles. This review provides a vivid description of the scientific advancements to clinicians and researchers in this field of image analysis and segmentation, which helps the development of novel automated methods for clinical applications. The review also contains the recently developed fully automated deep learning-based methods for segmentation, which not only provides better results compared to the conventional techniques but also open a new field of research in Medical Imaging.

Quantitative CT of the knee in the IMI-APPROACH osteoarthritis cohort: Association of bone mineral density with radiographic disease severity, meniscal coverage and meniscal extrusion

OBJECTIVE

Osteoarthritis (OA) is a highly prevalent chronic condition. The subchondral bone plays an important role in onset and progression of OA making it a potential treatment target for disease-modifying therapeutic approaches. However, little is known about changes of periarticular bone mineral density (BMD) in OA and its relation to meniscal coverage and meniscal extrusion at the knee. Thus, the aim of this study was to describe periarticular BMD in the Applied Public-Private Research enabling OsteoArthritis Clinical Headway (APPROACH) cohort at the knee and to analyze the association with structural disease severity, meniscal coverage and meniscal extrusion.

DESIGN

Quantitative CT (QCT), MRI and radiographic examinations were acquired in 275 patients with knee osteoarthritis (OA). QCT was used to assess BMD at the femur and tibia, at the cortical bone plate (Cort) and at the epiphysis at three locations: subchondral (Sub), mid-epiphysis (Mid) and adjacent to the physis (Juxta). BMD was evaluated for the medial and lateral compartment separately and for subregions covered and not covered by the meniscus. Radiographs were used to determine the femorotibial angle and were evaluated according to the Kellgren and Lawrence (KL) system. Meniscal extrusion was assessed from 0 to 3.

RESULTS

Mean BMD differed significantly between each anatomic location at both the femur and tibia (p < 0.001) in patients with KL0. Tibial regions assumed to be covered with meniscus in patients with KL0 showed lower BMD at Sub (p < 0.001), equivalent BMD at Mid (p = 0.07) and higher BMD at Juxta (p < 0.001) subregions compared to regions not covered with meniscus. Knees with KL2-4 showed lower Sub (p = 0.03), Mid (p = 0.01) and Juxta (p < 0.05) BMD at the medial femur compared to KL0/1. Meniscal extrusion grade 2 and 3 was associated with greater BMD at the tibial Cort (p < 0.001, p = 0.007). Varus malalignment is associated with significant greater BMD at the medial femur and at the medial tibia at all anatomic locations.

CONCLUSION

BMD within the epiphyses of the tibia and femur decreases with increasing distance from the articular surface. Knees with structural OA (KL2-4) exhibit greater cortical BMD values at the tibia and lower BMD at the femur at the subchondral level and levels beneath compared to KL0/1. BMD at the tibial cortical bone plate is greater in patients with meniscal extrusion grade 2/3.

A Registration Method for Three-Dimensional Analysis of Bone Mineral Density in the Proximal Tibia

Although alterations in bone mineral density (BMD) at the proximal tibia have been suggested to play a role in various musculoskeletal conditions, their pathophysiological implications and their value as markers for diagnosis remain unclear. Improving our understanding of proximal tibial BMD requires novel tools for three-dimensional (3D) analysis of BMD distribution. Three-dimensional imaging is possible with computed tomography (CT), but computational anatomy algorithms are missing to standardize the quantification of 3D proximal tibial BMD, preventing distribution analyses. The objectives of this study were to develop and assess a registration method, suitable with routine knee CT scans, to allow the standardized quantification of 3D BMD distribution in the proximal tibia. Second, as an example of application, the study aimed to characterize the distribution of BMD below the tibial cartilages in healthy knees. A method was proposed to register both the surface (vertices) and the content (voxels) of proximal tibias. The method combines rigid transformations to account for differences in bone size and position in the scanner's field of view and to address inconsistencies in the portion of the tibial shaft included in routine CT scan, with a nonrigid transformation locally matching the proximal tibias. The method proved to be highly reproducible and provided a comprehensive description of the relationship between bone depth and BMD. Specifically it reported significantly higher BMD in the first 6 mm of bone than deeper in the proximal tibia. In conclusion, the proposed method offers promising possibilities to analyze BMD and other properties of the tibia in 3D.

Cohort profile: The Applied Public-Private Research enabling OsteoArthritis Clinical Headway (IMI-APPROACH) study: a 2-year, European, cohort study to describe, validate and predict phenotypes of osteoarthritis using clinical, imaging and biochemical markers

PURPOSE

The Applied Public-Private Research enabling OsteoArthritis Clinical Headway (APPROACH) consortium intends to prospectively describe in detail, preselected patients with knee osteoarthritis (OA), using conventional and novel clinical, imaging, and biochemical markers, to support OA drug development.

PARTICIPANTS

APPROACH is a prospective cohort study including 297 patients with tibiofemoral OA, according to the American College of Rheumatology classification criteria. Patients were (pre)selected from existing cohorts using machine learning models, developed on data from the CHECK cohort, to display a high likelihood of radiographic joint space width (JSW) loss and/or knee pain progression.

FINDINGS TO DATE

Selection appeared logistically feasible and baseline characteristics of the cohort demonstrated an OA population with more severe disease: age 66.5 (SD 7.1) vs 68.1 (7.7) years, min-JSW 2.5 (1.3) vs 2.1 (1.0) mm and Knee injury and Osteoarthritis Outcome Score pain 31.3 (19.7) vs 17.7 (14.6), except for age, all: p<0.001, for selected versus excluded patients, respectively. Based on the selection model, this cohort has a predicted higher chance of progression.

FUTURE PLANS

Patients will visit the hospital again at 6, 12 and 24 months for physical examination, pain and general health questionnaires, collection of blood and urine, MRI scans, radiographs of knees and hands, CT scan of the knee, low radiation whole-body CT, HandScan, motion analysis and performance-based tests.After two years, data will show whether those patients with the highest probabilities for progression experienced disease progression as compared to those wit lower probabilities (model validation) and whether phenotypes/endotypes can be identified and predicted to facilitate targeted drug therapy.

TRIAL REGISTRATION NUMBER

NCT03883568.

Impact of meniscal coverage on subchondral bone mineral density of the proximal tibia in female subjects - A cross-sectional in vivo study using QCT

OBJECTIVE

To verify earlier data in cadavers that in female subjects with OA meniscal coverage is associated with lowered bone mineral density of the underlying subchondral bone in the proximal tibia by investigating the local bone mineral density (BMD) distribution within the epiphysis.

METHODS

BMD of the subchondral bone of the tibia was measured by QCT in 67 elderly females diagnosed with OA (Kellgren-Lawrence grades 2-3). The epiphysis was subdivided along the axis of the tibia into a subchondral-epiphyseal VOI covering the first 5-6 mm below the subchondral bone plate, a mid-epiphyseal VOI covering the adjacent 7-8 and a juxtaphyseal VOI of another 7-8 mm that bordered the growth plate. These VIOs were further divided into lateral and medial and then into anterior, mid and posterior sub-VOIs. Finally, all subVOIs were divided in one subVOI covered by the menisci (CM) and another not covered by the menisci (nCM). BMD ratios of these two subVOIs were compared.

RESULTS

In the subchondral epiphysis BMD was significantly lower (Medial: mean BMD_diff = 125 mg/cm³, p<0.001; Lateral: mean BMD_diff = 56 mg/cm³p < 0.001) in subVOIs covered by the meniscus compared to subVOIs not covered by the meniscus. The BMD difference was no longer significant in the mid epiphysis (Medial: mean BMD_diff = 10 mg/cm³, p>0.82; Lateral: mean BMD_diff = 7 mg/cm³, p=0.99) and was reversed in the juxtaphysis. With a few exceptions these BMD differences were independent of the lateral-medial and the anterior-mid-posterior position. BMD significantly (p<0.05) decreased with age independent on whether the location was covered or uncovered by the meniscus, however the BMD ratio of the corresponding nCM and CM subVOIs did not significantly (p>0.1) change with age.

CONCLUSION

In-vivo QCT measurements of the BMD distribution in the proximal tibia indicate a protective effect of the menisci in the subchondral bone close to the joint. This protective effect is age independent despite the overall age-related decrease of BMD.

Automatic allograft bone selection through band registration and its application to distal femur

Clinical reports suggest that large bone defects could be effectively restored by allograft bone transplantation, where allograft bone selection acts an important role. Besides, there is a huge demand for developing the automatic allograft bone selection methods, as the automatic methods could greatly improve the management efficiency of the large bone banks. Although several automatic methods have been presented to select the most suitable allograft bone from the massive allograft bone bank, these methods still suffer from inaccuracy. In this paper, we propose an effective allograft bone selection method without using the contralateral bones. Firstly, the allograft bone is globally aligned to the recipient bone by surface registration. Then, the global alignment is further refined through band registration. The band, defined as the recipient points within the lifted and lowered cutting planes, could involve more local structure of the defected segment. Therefore, our method could achieve robust alignment and high registration accuracy of the allograft and recipient. Moreover, the existing contour method and surface method could be unified into one framework under our method by adjusting the lift and lower distances of the cutting planes. Finally, our method has been validated on the database of distal femurs. The experimental results indicate that our method outperforms the surface method and contour method.

Crystallographic orientation of the c-axis of biological apatite as a new index of the quality of subchondral bone in knee joint osteoarthritis

The aim of the present study was to investigate the preferred orientation of biological apatite (BAp) as a new index of the quality of subchondral bone (SB) in knee joint osteoarthritis (OA). Ten OA and five normal knee joints were obtained. Thickness, quantity and bone mineral density (BMD) of SB were analyzed at the medial condyle of the femur in dry conditions by peripheral quantitative computed tomography. In addition, the preferred crystallographic orientation of the c-axis of BAp was evaluated as bone quality parameter using a microbeam X-ray diffractometer technique. BMD and thickness of SB were significantly increased in OA specimens compared to normal knee specimens (P < 0.01), and the preferred orientation of the c-axis of BAp along the normal direction of SB surface was significantly higher in OA specimens (P < 0.01), reflecting the change in stress of concentration in the pathological portion without cartilage. SB sclerosis in OA results in both proliferation of bone tissues and enhanced degree of preferential alignment of the c-axis of BAp. Our findings could have major implications for the diagnosis of clinical studies, including pathologic elucidation in OA.

Advanced Knee Structure Analysis (AKSA): a comparison of bone mineral density and trabecular texture measurements using computed tomography and high-resolution peripheral quantitative computed tomography of human knee cadavers

Background

A change of loading conditions in the knee causes changes in the subchondral bone and may be a cause of osteoarthritis (OA). However, quantification of trabecular architecture in vivo is difficult due to the limiting spatial resolution of the imaging equipment; one approach is the use of texture parameters. In previous studies, we have used digital models to simulate changes of subchondral bone architecture under OA progression. One major result was that, using computed tomography (CT) images, subchondral bone mineral density (BMD) in combination with anisotropy and global homogeneity could characterize this progression.

The primary goal of this study was a comparison of BMD, entropy, anisotropy, variogram slope, and local and global inhomogeneity measurements between high-resolution peripheral quantitative CT (HR-pQCT) and CT using human cadaveric knees. The secondary goal was the verification of the spatial resolution dependence of texture parameters observed in the earlier simulations, two important prerequisites for the interpretation of in vivo measurements in OA patients.

Method

The applicability of texture analysis to characterize bone architecture in clinical CT examinations was investigated and compared to results obtained from HR-pQCT. Fifty-seven human knee cadavers (OA status unknown) were examined with both imaging modalities. Three-dimensional (3D) segmentation and registration processes, together with automatic positioning of 3D analysis volumes of interest (VOIs), ensured the measurement of BMD and texture parameters at the same anatomical locations in CT and HR-pQCT datasets.

Results

According to the calculation of dice ratios (>0.978), the accuracy of VOI locations between methods was excellent. Entropy, anisotropy, and global inhomogeneity showed significant and high linear correlation between both methods (0.68 < R² < 1.00). The resolution dependence of these parameters simulated earlier was confirmed by the in vitro measurements.

Conclusion

The high correlation of HR-pQCT- and CT-based measurements of entropy, global inhomogeneity, and anisotropy suggests interchangeability between devices regarding the quantification of texture. The agreement of the experimentally determined resolution dependence of global inhomogeneity and anisotropy with earlier simulations is an important milestone towards their use to quantify subchondral bone structure. However, an in vivo study is still required to establish their clinical relevance.

Characterization of knee osteoarthritis-related changes in trabecular bone using texture parameters at various levels of spatial resolution-a simulation study

Articular cartilage and subchondral bone are the key tissues in osteoarthritis (OA). The role of the cancellous bone increasingly attracts attention in OA research. Because of its fast adaptation to changes in the loading distribution across joints, its quantification is expected to improve the diagnosis and monitoring of OA. In this study, we simulated OA progression-related changes of trabecular structure in a series of digital bone models and then characterized the potential of texture parameters and bone mineral density (BMD) as surrogate measures to quantify trabecular bone structure. Five texture parameters were studied: entropy, global and local inhomogeneity, anisotropy and variogram slope. Their dependence on OA relevant structural changes was investigated for three spatial resolutions typically used in micro computed tomography (CT; 10 μm), high-resolution peripheral quantitative CT (HR-pQCT) (90 μm) and clinical whole-body CT equipment (250 μm). At all resolutions, OA-related changes in trabecular bone architecture can be quantified using a specific (resolution dependent) combination of three texture parameters. BMD alone is inadequate for this purpose but if available reduces the required texture parameter combination to anisotropy and global inhomogeneity. The results are summarized in a comprehensive analysis guide for the detection of structural changes in OA knees. In conclusion, texture parameters can be used to characterize trabecular bone architecture even at spatial resolutions below the dimensions of a single trabecula and are essential for a detailed classification of relevant OA changes that cannot be achieved with a measurement of BMD alone.

A Digital Model to Simulate Effects of Bone Architecture Variations on Texture at Spatial Resolutions of CT, HR-pQCT, and μCT Scanners

The quantification of changes in the trabecular bone structure induced by musculoskeletal diseases like osteoarthritis, osteoporosis, rheumatoid arthritis, and others by means of a texture analysis is a valuable tool which is expected to improve the diagnosis and monitoring of a disease. The reaction of texture parameters on different alterations in the architecture of the fine trabecular network and inherent imaging factors such as spatial resolution or image noise has to be understood in detail to ensure an accurate and reliable determination of the current bone state. Therefore, a digital model for the quantitative analysis of cancellous bone structures was developed. Five parameters were used for texture analysis: entropy, global and local inhomogeneity, local anisotropy, and variogram slope. Various generic structural changes of cancellous bone were simulated for different spatial resolutions. Additionally, the dependence of the texture parameters on tissue mineralization and noise was investigated. The present work explains changes in texture parameter outcomes based on structural changes originating from structure modifications and reveals that a texture analysis could provide useful information for a trabecular bone analysis even at resolutions below the dimensions of single trabeculae.

Bone marrow lesions identified by MRI in knee osteoarthritis are associated with locally increased bone mineral density measured by QCT

OBJECTIVE

Bone marrow lesions (BMLs) in the knee are associated with pain and compartment-specific joint space narrowing. However, the correlation of BMLs with bone mineral density (BMD) has rarely been investigated. The aim of the present study was to examine whether BMD in BMLs is altered compared to the surrounding bone.

DESIGN

Thirty-four BMLs detected in osteoarthritis (OA) knees (KL grade 2 and 3) of 26 patients were investigated. A 3D-segmentation was used to determine BML volumes of interest (VOI) and their surrounding bone in MR images. These VOIs were registered to corresponding single-energy QCT images and a BMD analysis was performed. The same VOIs were transferred to control datasets (19 OA patients without BMLs) by an elastic registration, where the BMD analysis was repeated. To account for the dependence of bone marrow composition on BMD measures derived using single-energy QCT, simulations were performed to evaluate how changing fat-water compositions likely occurring with BML development may influence BMD measures and observed BMD differences between patients with and without BMLs. The association between loading in the knee and the occurrence of BMLs was investigated by medial to lateral (M:L) BMD ratios.

RESULTS

BMD was significantly increased at BML locations, even with a fat-to-water conversion rate of 39%. The M:L BMD ratio was significantly increased in bones with medial BMLs.

CONCLUSIONS

BMD was examined exactly at BML locations and surrounding bone using highly accurate segmentation and registration methods. BMD was significantly increased at BML locations (P < 0.05).

Advances in bone imaging for osteoporosis

The diagnosis and management of osteoporosis have been improved by the development of new quantitative methods of skeletal assessment and by the availability of an increasing number of therapeutic options, respectively. A number of imaging methods exist and all have advantages and disadvantages. Dual-energy X-ray absorptiometry (DXA) is the most widely available and commonly utilized method for clinical diagnosis of osteoporosis and will remain so for the foreseeable future. The WHO 10-year fracture risk assessment tool (FRAX(®)) will improve clinical use of DXA and the cost-effectiveness of therapeutic intervention. Improved reporting of radiographic features that suggest osteoporosis and the presence of vertebral fracture, which are powerful predictors of future fractures, could increase the frequency of appropriate DXA referrals. Quantitative CT remains predominantly a research tool, but has advantages over DXA--allowing measurement of volumetric density, separate measures of cortical and trabecular bone density, and evaluation of bone shape and size. High resolution imaging, using both CT and MRI, has been introduced to measure trabecular and cortical bone microstructure. Although these methods provide detailed insights into the effects of disease and therapies on bone, they are technically challenging and not widely available, so they are unlikely to be used in clinical practice.

CT imaging for the investigation of subchondral bone in knee osteoarthritis

In osteoarthritis, magnetic resonance imaging is the method of choice to image articular cartilage and "bone marrow lesions." However, the calcified cartilage, the subchondral bone plate, and trabecular subchondral bone that are mineralized tissues strongly attenuate X-rays and are therefore potentially accessible for analyses using computed tomography (CT). CT images nicely show osseous cardinal signs of advanced osteoarthritis such as osteophytes, subchondral cysts, and subchondral bone sclerosis. But more importantly, CT can help us to better understand the pathophysiology of knee osteoarthritis from the measurement of the density and structure of subchondral mineralized tissues in vivo. For that purpose, we recently developed dedicated image analysis software called Medical Image Analysis Framework (MIAF)-Knee. In this manuscript, our aims are to present current knowledge on CT imaging of the subchondral bone in knee osteoarthritis and to provide a brief introduction to basic technical aspects of MIAF-Knee as well as preliminary results we obtained in patients with knee osteoarthritis as compared to control subjects.