Affiliations: [a]
Department of Computer Engineering, Silla University, Busan, Korea
| [b]
Department of Radiology, School of Medicine, Pusan National University, Busan, Korea
| [c]
Division of Software Convergence, Cheongju University, Cheongju, Korea
| [d]
Department of Computer Games, Yong-In SongDam College, Yongin, Korea
| [e]
Department of Neurosurgery, School of Medicine, Pusan National University, Busan, Korea
Correspondence:
[*]
Corresponding author. Kwang Baek Kim, Department of Computer Engineering, Silla University, Busan 46958, Korea. E-mail: gbkim@silla.ac.kr.
Abstract: Rotator cuff tear is a common cause of shoulder pain and disability. It is especially critical for elderly people since such injuries are strongly related with age and reduce the quality of life due to the shoulder pain and weakness with shoulder flexion and abduction. Ultrasound of the shoulder is widely used in examining the state of rotator cuff but is often criticized as operator-dependent because of the complexity of the shoulder anatomy or anisotropy and lack of intensity contrast from ultrasonographic images. Automatic segmentation of the related tendon tear by computer assisted software will be the answer for that problem. In this paper, we propose a fully automatic extractor of partial/full thickness tear of rotator cuff tendon with Fuzzy C-Means based quantization for pixel classification and fuzzy stretching for image contrast enhancement. In experiment, our method exhibits sufficient agreement with human expert. For 12 partial thickness tear cases, the sensitivity was 96.5% and specificity was 91.1% whereas the sensitivity was 92.6% and the specificity was 96.4% for 44 full thickness tear cases.
