Objectives This study aimed at developing a convolutional neural network (CNN) able to automatically quantify and characterize the level of degeneration of rotator cuff (RC) muscles from shoulder CT images including muscle atrophy and fatty infiltration. Methods One hundred three shoulder CT scans from 95 patients with primary glenohumeral osteoarthritis undergoing anatomical total shoulder arthroplasty were retrospectively retrieved. Three independent radiologists manually segmented the premorbid boundaries of all four RC muscles on standardized sagittal-oblique CT sections. This premorbid muscle segmentation was further automatically predicted using a CNN. Automatically predicted premorbid segmentations were then used to quantify the ratio of muscle atrophy, fatty infiltration, secondary bone formation, and overall muscle degeneration. These muscle parameters were compared with measures obtained manually by human raters. Results Average Dice similarity coefficients for muscle segmentations obtained automatically with the CNN (88% ± 9%) and manually by human raters (89% ± 6%) were comparable. No significant differences were observed for the subscapularis, supraspinatus, and teres minor muscles (p > 0.120), whereas Dice coefficients of the automatic segmentation were significantly higher for the infraspinatus (p < 0.012). The automatic approach was able to provide good-very good estimates of muscle atrophy (R 2 = 0.87), fatty infiltration (R 2 = 0.91), and overall muscle degeneration (R 2 = 0.91). However, CNN-derived segmentations showed a higher variability in quantifying secondary bone formation (R 2 = 0.61) than human raters (R 2 = 0.87). Conclusions Deep learning provides a rapid and reliable automatic quantification of RC muscle atrophy, fatty infiltration, and overall muscle degeneration directly from preoperative shoulder CT scans of osteoarthritic patients, with an accuracy comparable with that of human raters. Key Points • Deep learning can not only segment RC muscles currently available in CT images but also learn their pre-existing locations and shapes from invariant anatomical structures visible on CT sections. • Our automatic method is able to provide a rapid and reliable quantification of RC muscle atrophy and fatty infiltration from conventional shoulder CT scans. • The accuracy of our automatic quantitative technique is comparable with that of human raters.
Including structural information of trabecular bone improves the prediction of bone strength and fracture risk. However, this information is available in clinical CT scans, only for peripheral bones. We hypothesized that a correlation exists between the shape of the bone, its volume fraction (BV/TV) and fabric, which could be characterized using statistical modeling. High-resolution peripheral computed tomography (HR-pQCT) images of 73 proximal femurs were used to build a combined statistical model of shape, BV/TV and fabric. The model was based on correspondence established by image registration and by morphing of a finite element mesh describing the spatial distribution of the bone properties. Results showed no correlation between the distribution of bone shape, BV/TV and fabric. Only the first mode of variation associated with density and orientation showed a strong relationship (R>0.8). In addition, the model showed that the anisotropic information of the proximal femur does not vary significantly in a population of healthy, osteoporotic and osteopenic samples. In our dataset, the average anisotropy of the population was able to provide a close approximation of the patient-specific anisotropy. These results were confirmed by homogenized finite element (hFE) analyses, which showed that the biomechanical behavior of the proximal femur was not significantly different when the average anisotropic information of the population was used instead of patient-specific fabric extracted from HR-pQCT. Based on these findings, it can be assumed that the fabric information of the proximal femur follows a similar structure in an elderly population of healthy, osteopenic and osteoporotic proximal femurs.
Image-based modeling is a popular approach to perform patient-specific biomechanical simulations. Accurate modeling is critical for orthopedic application to evaluate implant design and surgical planning. It has been shown that bone strength can be estimated from the Bone Mineral Density (BMD) and trabecular bone architecture. However, these findings cannot directly be fully transferred to patient-specific modeling since only BMD can be derived from clinical CT. Therefore, the objective of this study was to propose a method to predict the trabecular bone structure using a µCT atlas and an image registration technique. The approach has been evaluated on femurs and patellae under physiological loading. The displacement and ultimate force for femurs loaded in stance position were predicted with an error of 2.5% and 3.7% respectively, while predictions obtained with an isotropic material resulted in errors of 7.3% and 6.9%. Similar results were obtained for the patella, where the strain predicted using the registration approach resulted in an improved mean squared error compared to the isotropic model. We conclude that the registration of anisotropic information from of a single template bone enables more accurate patient-specific simulation from clinical image datasets than isotropic model.
Background
Higher‐resolution MRI of the patellofemoral cartilage under loading is hampered by subject motion since knee flexion is required during the scan.
Purpose
To demonstrate robust quantification of cartilage compression and contact area changes in response to in situ loading by means of MRI with prospective motion correction and regularized image postprocessing.
Study Type
Cohort study.
Subjects
Fifteen healthy male subjects.
Field Strength
3 T.
Sequence
Spoiled 3D gradient‐echo sequence augmented with prospective motion correction based on optical tracking. Measurements were performed with three different loads (0/200/400 N).
Assessment
Bone and cartilage segmentation was performed manually and regularized with a deep‐learning approach. Average patellar and femoral cartilage thickness and contact area were calculated for the three loading situations. Reproducibility was assessed via repeated measurements in one subject.
Statistical Tests
Comparison of the three loading situations was performed by Wilcoxon signed‐rank tests.
Results
Regularization using a deep convolutional neural network reduced the variance of the quantified relative load‐induced changes of cartilage thickness and contact area compared to purely manual segmentation (average reduction of standard deviation by ∼50%) and repeated measurements performed on the same subject demonstrated high reproducibility of the method. For the three loading situations (0/200/400 N), the patellofemoral cartilage contact area as well as the mean patellar and femoral cartilage thickness were significantly different from each other (P < 0.05). While the patellofemoral cartilage contact area increased under loading (by 14.5/19.0% for loads of 200/400 N), patellar and femoral cartilage thickness exhibited a load‐dependent thickness decrease (patella: –4.4/–7.4%, femur: –3.4/–7.1% for loads of 200/400 N).
Data Conclusion
MRI with prospective motion correction enables quantitative evaluation of patellofemoral cartilage deformation and contact area changes in response to in situ loading. Regularizing the manual segmentations using a neural network enables robust quantification of the load‐induced changes.
Level of Evidence: 2
Technical Efficacy: Stage 2
J. Magn. Reson. Imaging 2019;50:1561–1570.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.