A novel mean shape based post-processing method for enhancing deep learning lower-limb muscle segmentation accuracy

Abstract

This study aims at improving the lower-limb muscle segmentation accuracy of deep learning approaches based on Magnetic Resonance Imaging (MRI) scans, crucial for the diagnostic and therapeutic processes in musculoskeletal diseases. In general, segmentation methods such as U-Net deep learning neural networks can achieve good Dice Similarity Coefficient (DSC) values, e.g. around 0.83 to 0.91 on various cohorts. Some generic post-processing strategies have been studied to incorporate connectivity constraints into the resulting masks for the purpose of further improving the segmentation accuracy. In this paper, a novel mean shape (MS) based post-processing method is proposed, utilizing Statistical Shape Modelling (SSM) to fine-tune the segmentation output, taking into consideration the muscle anatomical shape. The methodology was compared to existing post-processing techniques and a commercial semi-automatic tool on MRI scans from two cohorts of post-menopausal women (10 Training, 8 Testing, voxel size 1.0x1.0x1.0 mm3). The MS based method obtained a mean DSC of 0.83 across the different analysed muscles and the best performance for the Hausdorff Distance (HD, 20.6 mm) and the Average Symmetric Surface Distance (ASSD, 2.1 mm). These findings highlight the feasibility and potential of using anatomical mean shape in post-processing of human lower-limb muscle segmentation task and indicate that the proposed method can be popularized to other biological organ segmentation mission.

Fig 1. Experimental design pipeline.

Fig 2. Pipeline for mean shape based post-processing method.

Initially, the muscle segmentation results (original predictions) from the neural network were stacked, and subsequently, each muscle label was extracted to generate distinct 3D shapes (a-1). Using the ShapeWorks software, the mean shape for each muscle was generated (a-2) using gold standard manual segmentation from PMW-1. To reduce substantial spatial disparities between the original prediction and the mean shape, a rigid registration based on an iterative closest point algorithm (ICP) was performed in MATLAB (fixed shape: original prediction; moving shape: mean shape; function: pcregistericp (moving, fixed)) (b-1). Variations in muscle (e.g., TFL) slices at corresponding positions of both, from the knee to the hip, are depicted (b-2). In the final step, the two shapes underwent deformable registration in ShIRT, and the registered mean shape was retained as the post-processing segmentation results (c-1/2/3).

Fig 3. The mask and segmentation visualization of atlas and new subjects in Mimics.

Fig 4. Performance comparison of each method.

Box plots for the DSC (top left), RVE (top right), HD (bottom left), and ASSD (bottom right) for each method (16 medium values of each muscle calculated from 128 labels) are shown (* indicates p<0.05, ** indicates p<0.01). Lower scores indicate better performance in HD and ASSD. U-Net denotes original prediction from U-Net; U-Net+2D denotes 2D post-processing; U-Net+CRF denotes fully connected conditional random fields; U-Net+MS denotes mean shape based post-process method; Mimics denotes the results from Mimics semi-automatic muscle segmentation toolbox.

Fig 5. Performance comparison of each method after pooling all labels into one.

Box plots for the DSC (top left), RVE (top right), and ASSD (bottom left) for each method (8 values from 8 subject under each metric) are shown (* indicates p<0.05, ** indicates p<0.01).

Fig 6. Segmentation performance visualization of each method.

The figure exhibits the segmentation performance of each method on one subject from a location close to knee to hip, rows a-d. Column 0 denotes the golden standard, and columns 1–5 denote different segmentation results.

Fig 7. Segmentation performance visualization of each method after pooling all labels into one.

The figure exhibits the segmentation performance of each method on one testing subject from a location close to knee to hip, rows a-d. Column 0 denotes the golden standard, and columns 1–5 denote different segmentation results.

Fig 8. Comparison between ground truth and segmentation results with/without post-processing step.

GT (in purple) and segmented results (in green) are overlapped and visualized by converting them into point clouds in MATLAB. Parts a-i are the comparison of VI, VL, VM, SMB, SMT, GRA, AM, AB and GM from one testing subject, respectively. In each part, the left side is the comparison of U-Net results with GT, and the right side is the comparison of U-Net+MS results with GT.

Fig 9. Comparison between ground truth and segmentation results from Mimics.

In each part, the ground truth (left), segmentation result (middle) and the overlapping section of both (right) are shown. Parts a and b represent the results of muscle vastus intermedius and sartorius of same testing subject.

Fig 10. Relationship between muscle volume and DSC score under 4 methods (U-Net (top left), U-Net+2D (top right), U-Net+CRF (bottom left) and U-Net+MS (bottom right)) of test subjects.

The muscles were arranged in ascending order according to the mean muscle volume of PMW-2 cohort. The error bar of each point denotes the standard deviation of muscle volume among 8 test subjects.