Estimating Full Regional Skeletal Muscle Fibre 2 Curvature from b - Mode Ultrasound Images Using 3 Convolutional - Deconvolutional Neural Networks 4

Direct measurement of strain within muscle is important for understanding muscle 8 function in health and disease. Current technology (kinematics, dynamometry, electromyography) 9 provides limited ability to measure strain within muscle. Regional fiber orientation and length are 10 related with active/passive strain within muscle. Currently, ultrasound imaging provides the only 11 non-invasive means of observing regional fiber orientation within muscle during dynamic tasks. 12 Previous attempts to automatically estimate fiber orientation from ultrasound are not adequate, 13 often requiring manual region selection, feature engineering, providing low-resolution estimations 14 (one angle per muscle), and deep muscles are often not attempted. Here, we propose 15 deconvolutional neural networks (DCNN) for estimating fiber orientation at the pixel-level. 16 Dynamic ultrasound images sequences of the calf muscles were acquired (25 Hz) from 8 healthy 17 volunteers (4 male, ages: 25–36, median 30). A combination of expert annotation and 18 interpolation/extrapolation provided labels of regional fiber orientation for each image. We then 19 trained DCNNs both with and without dropout using leave one out cross-validation. Our results 20 demonstrated robust estimation of regional fiber orientation with approximately 3° error, which 21 was an improvement on previous methods. The methods presented here provide new potential to 22 study muscle in disease and health. 23