Automatic Musculoskeletal Segmentation in Medical Images

Many people around the world, especially soldiers and military personnel who require a lot of physical activity to perform combat tasks, face with musculoskeletal injuries. To automatically diagnose musculoskeletal disorders in the medical images, the first step is to segment the bones and the muscles in these images. The aim of this study is an automatically segmentation of bones and skeletal muscles in the medical images.

Various medical imaging methods such as CT scan can be used to obtain images of different parts of the body for identification and assessment of injuries and diseases. In this research, 1200 CT-Scan images from NAJA staff were used to segment muscles and bones. These datasets were taken from the imaging center of Hazrat Vali-e-Asr Hospital. There are different image processing algorithms for medical image segmentation. In this study, the fuzzy clustering algorithm was used. In the proposed method, depending on the anatomical position of the slices and the presence of dense or spongy bone in the slices, a different number of classes were defined for the fuzzy algorithm according to the image brightness histogram. There was one intensity class for the muscles in all slices.

The results of muscles and bones segmentation are shown in 2D and 3D. Two and Three-dimensional segmentation allows the observation and assessment of broken bones and changes in muscles volume due to various injuries and during treatment.

The use of different medical image processing methods for automatic musculoskeletal segmentation in these images can help physicians in diagnosing and evaluating the healing process of musculoskeletal injuries among military personnel.