Deep learning methods for automatic segmentation of lower leg muscles and bones from MRI scans of children with and without cerebral palsy

Cerebral palsy is a neurological condition that is known to affect muscle growth. Detailed investigations of muscle growth require segmentation of muscles from MRI scans, which is typically done manually. In this study, we evaluated the performance of 2D, 3D, and hybrid deep learning models for automatic segmentation of 11 lower leg muscles and two bones from MRI scans of children with and without cerebral palsy. All six models were trained and evaluated on manually segmented T-1-weighted MRI scans of the lower legs of 20 children, six of whom had cerebral palsy. The segmentation results were assessed using the median Dice similarity coefficient (DSC), average symmetric surface distance (ASSD), and volume error (VError) of all 13 labels of every scan. The best performance was achieved by H-DenseUNet, a hybrid model (DSC 0.90, ASSD 0.5 mm, and VError 2.6 cm(3)). The performance was equivalent to the inter-rater performance of manual segmentation (DSC 0.89, ASSD 0.6 mm, and VError 3.3 cm(3)). Models trained with the Dice loss function outperformed models trained with the cross-entropy loss function. Near-optimal performance could be attained using only 11 scans for training. Segmentation performance was similar for scans of typically developing children (DSC 0.90, ASSD 0.5 mm, and VError 2.8 cm(3)) and children with cerebral palsy (DSC 0.85, ASSD 0.6 mm, and VError 2.4 cm(3)). These findings demonstrate the feasibility of fully automatic segmentation of individual muscles and bones from MRI scans of children with and without cerebral palsy.

Automated summary

The study evaluated deep learning models for automatic segmentation of muscles and bones from MRI scans of children with and without cerebral palsy. The hybrid model showed the best performance, and models trained with the Dice loss function outperformed those trained with the cross-entropy loss function. Additionally, near-optimal performance could be achieved with training on only 11 scans.