Automatic Tracking of Muscle Cross‐Sectional Area Using Convolutional Neural Networks with Ultrasound

Abstract

Objectives The purpose of this study was to develop an automatic tracking method for the muscle cross‐sectional area (CSA) on ultrasound (US) images using a convolutional neural network (CNN). The performance of the proposed method was evaluated and compared with that of the state‐of‐the art muscle segmentation method. Methods A real‐time US image sequence was obtained from the rectus femoris muscle during voluntary contraction. A CNN was built to segment the rectus femoris muscle and calculate the CSA in each US frame. This network consisted of 2 stages: feature extraction and score map reconstruction. The training of the network was divided into 3 steps with output score map resolutions of one‐fourth, one‐half, and all of the original image. We evaluated the segmentation performance of our method with 5‐fold cross‐validation. The mean precision, recall, and dice similarity score were calculated. Results The mean precision, recall, and Dice's coefficient (DSC) ± SD were 0.936 ± 0.029, 0.882 ± 0.045, and 0.907 ± 0.023, respectively. Compared with the state‐of‐the‐art muscle segmentation method (constrained mutual‐information–based free‐form deformation), the proposed method using CNN showed high performance. Conclusions The automated method proposed in this study provides an accurate and efficient approach to the estimation of the muscle CSA during muscle contraction.

ACSAuto-semi-automatic assessment of human vastus lateralis and rectus femoris cross-sectional area in ultrasound images

“…While this technique seems to be able to detect changes in muscle size and responses to musculoskeletal training 33 , 34 , ACSA measurements in muscles exceeding the field of view of the ultrasound probe might be limited in meaningfulness. Chen et al 15 recently demonstrated an automatic ACSA segmenting algorithm using a deep learning model. Deep learning is a type of machine learning that uses a deep neural network 15 .…”

“…Chen et al 15 recently demonstrated an automatic ACSA segmenting algorithm using a deep learning model. Deep learning is a type of machine learning that uses a deep neural network 15 . The algorithm segments the ACSA of the RF in ultrasound images and test images were recorded during contraction of the muscle 15 .…”

“…These programs were mainly designed for image sequences derived from dynamic ultrasound measurements 11 – 13 . More recently, Seynnes and Cronin 14 and Cronin et al 15 , 16 published fully automated open source programs able to calculate fascicle length, pennation angle and thickness of muscles from static ultrasound pictures. Nonetheless, very few automated programs measuring the ACSA of a whole muscle in ultrasound images are currently available 15 , 17 .…”

“…More recently, Seynnes and Cronin 14 and Cronin et al 15 , 16 published fully automated open source programs able to calculate fascicle length, pennation angle and thickness of muscles from static ultrasound pictures. Nonetheless, very few automated programs measuring the ACSA of a whole muscle in ultrasound images are currently available 15 , 17 . For example, Salvi and colleagues 17 used an image segmentation approach to measure ASCA of several muscles in static B-mode images.…”

“…For example, Salvi and colleagues 17 used an image segmentation approach to measure ASCA of several muscles in static B-mode images. Chen et al 15 used a deep learning approach to train a neural network in evaluating m. rectus femoris (RF) ACSA in an EFOV image sequence acquired during voluntary contraction. Yet, none of these fully automated programs are openly accessible.…”