![]() The objective of our study was to develop deep convolutional neural networks (CNN) to identify an axial image slice from a routine axial T1-weighted PM MRI and another CNN to segment the PM muscle on an axial view. However, no prior studies have evaluated a workflow to select a specific MR image of the pectoral major muscle and to determine the PMM-CSA. Multiple approaches have been described to segment upper extremity muscles, mostly including rotator cuff muscles using manual or semi-automated time-consuming strategies 4, 15 and recently using deep learning as an accurate method 16. Therefore, there is an opportunity to evaluate the application of emerging deep learning-based MRI algorithms for complex segmentations, such as the pectoralis major muscle described in this work. These inconsistencies can affect the surgical planning for anatomic repair of the PM and make meaningful evaluation of repair techniques and treatment outcomes not optimal 14. Moreover, the literature lacks an injury classification system that is consistently applied and accurately reflects surgically relevant, anatomic injury patterns. As a result, published descriptions of PM ruptures have been inconsistent with the actual musculotendinous morphology. The PMM and tendon have a complex musculotendinous anatomy that is often misunderstood by both radiologists and surgeons non-familiar with its morphology. MRI is the gold standard for diagnosis of an acute pectoralis major tear useful to identify the location, size and severity of the lesion and to also for treatment planning 13. Imaging evaluation of the pectoralis major is paramount when an injury is suspected. The Pectoralis major muscle (PMM) and tendon (PMT) are main contributors to bench press movement, as shown in electroneuromyography studies 11, 12. Bench press exercise is a multi-joint movement commonly used for improving upper body performance. Sports activities including weightlifting require powerful movement of the upper body. Magnetic resonance imaging (MRI) has been used to assess muscle size and found to significantly correlate with joint power in single-joint 10 and multi-joint movements 1. During the assessment of sports performance, power output determined by muscle strength and joint velocity are important for determining the optimal load for resistance and power training 5, 6, 7, 8, 9. Muscle size is a determinant of muscle strength during single-joint and multi-joint movements 1, 2, 3, 4. Our results show an overall accurate selection of PMM-CSA and automated PM muscle segmentation using a combination of deep CNN algorithms. The results of Step B showed top-3 accuracy > 98% to select an appropriate axial image with the greatest PMM-CSA. The segmentation model (Step A) produced an accurate pectoralis muscle segmentation with a Mean Dice score of 0.94 ± 0.01. A top-3 accuracy evaluated this method on 8 axial T1-weighted PM MRIs internal test cases. If one of the selected was in the top-3 from the ground truth, then we considered it to be a success. Then, we selected the top-3 slices with the largest cross-sectional area and compared them with the ground truth. In step B, we used the OpenCV2 (version 4.5.1, ) framework to calculate the PMM-CSA of the model predictions and ground truth. Mean-dice score determined the segmentation score on 8 internal axial T1-weighted PM MRIs. The segmentation model was trained from scratch (MONAI/Pytorch SegResNet, 4 mini-batch, 1000 epochs, dropout 0.20, Adam, learning rate 0.0005, cosine annealing, softmax). In step A, we manually segmented the PMM on 134 axial T1-weighted PM MRIs. Our method is based on two steps: (A) segmentation model, (B) PMM-CSA selection. We hypothesized a CNN technique can accurately perform both tasks compared with manual reference standards. To develop and validate a deep convolutional neural network (CNN) method capable of selecting the greatest Pectoralis Major Cross-Sectional Area (PMM-CSA) and automatically segmenting PMM on an axial Magnetic Resonance Imaging (MRI).
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |