Abstract
Aim: This study aimed to develop a system to automatically measure the movement of feature points on transabdominal ultrasound images that can evaluate rectus abdominis contraction. Methods: Transabdominal ultrasound images during rectus abdominis contraction were obtained from 33 women, and the ratio of well-visible images was calculated for 15 candidate indicators on the images. The movement distances of the well-visible indicators between rest and contraction of the rectus abdominis were measured manually and by the developed feature point tracking software. An expert visually classified the strength of the rectus abdominis contractions (strong or weak). Results: We identified three feature points: the intersection of the lateral wall of the bladder head and the bladder base, a point on the vaginal lumen, and a point on the anterior rectal wall. For each of these, when an expert visually classified a strong contraction, the movement distance of the feature points was significantly greater than that for weak contractions. The regression coefficient of movement distances between manually and automatically measured was 0. 88-1. 17. Conclusion: We identified the feasibility of using one of the three feature types of feature points to evaluate rectus abdominis contraction and developed a system to automatically track these feature points.
