Key result
Two-stage deep learning pipeline achieves ~0.93 average Dice similarity coefficient for thigh CT segmentation.
Why the study?
Deep learning segmentation of thigh CT images requires substantial labeled data, but manual annotation is costly, motivating methods to train with limited human labels.
Approximated handcrafted pseudo labels can provide good initialization for deep neural networks in CT image segmentation, reducing the need for extensive human expert labeled data.
May aid efficient CT segmentation workflows; leaves open prospective validation before clinical adoption.
Purpose: Muscle, bone, and fat segmentation from thigh images is essential for quantifying body composition. Voxelwise image segmentation enables quantification of tissue properties including area, intensity, and texture. Deep learning approaches have had substantial success in medical image segmentation, but they typically require a significant amount of data. Due to the high cost of manual annotation, training deep learning models with limited human label data is desirable, but it is a challenging problem. Approach: Inspired by transfer learning, we proposed a two-stage deep learning pipeline to address the thigh and lower leg segmentation issue. We studied three datasets, 3022 thigh slices and 8939 lower leg slices from the BLSA dataset and 121 thigh slices from the GESTALT study. First, we generated pseudo labels for thigh based on approximate handcrafted approaches using CT intensity and anatomical morphology. Then, those pseudo labels were fed into deep neural networks to train models from scratch. Finally, the first stage model was loaded as the initialization and fine-tuned with a more limited set of expert human labels of the thigh. Results: We evaluated the performance of this framework on 73 thigh CT images and obtained an average Dice similarity coefficient (DSC) of 0.927 across muscle, internal bone, cortical bone, subcutaneous fat, and intermuscular fat. To test the generalizability of the proposed framework, we applied the model on lower leg images and obtained an average DSC of 0.823. Conclusions: Approximated handcrafted pseudo labels can build a good initialization for deep neural networks, which can help to reduce the need for, and make full use of, human expert labeled data.
No takes yet. Share an insight, caveat, or question.
Yang et al. (2022) studied Muscle, bone, and fat segmentation from thigh images. Two-stage deep learning pipeline with pseudo labels was evaluated on Average Dice similarity coefficient (DSC) across muscle, internal bone, cortical bone, subcutaneous fat, and intermuscular fat. A two-stage deep learning pipeline using approximated handcrafted pseudo labels achieved an average Dice similarity coefficient of 0.927 for thigh CT image segmentation.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: