Video based CTS Screening
Screening Carpal Tunnel Syndrome and Comparing Multiple Diseases with Video Analysis of Hand Movement
松井良太,井原拓哉,塚本和矢, 小山恭史,藤田浩二,杉浦裕太
Ryota Matsui, Takuya Ibara, Kazuya Tsukamoto, Takafumi Koyama, Koji Fujita, Yuta Sugiura

Reference / 引用はこちら

Ryota Matsui, Takuya Ibara, Kazuya Tsukamoto, Takafumi Koyama, Koji Fujita, Yuta Sugiura, Video Analysis of Hand Gestures for Distinguishing Patients with Carpal Tunnel Syndrome, The 2022 ACM Interactive Surfaces and Spaces Conference (ACM ISS 2022), ACM.

Video based CM Screeningプロジェクトにおいて,手指のグーパー動作の解析により,頚髄症の疑いを発見するスクリーニング手法を提案した.この手法は,10秒テストとよばれる既存の頚髄症スクリーニング手法に基づいているが, 手指運動の障害は他の疾患の症状としてもみられる.本研究では,同手法を手根管症候群のスクリーニング,さらには疾患同士の比較に応用しうるかについて検討する.同手法では,グーパー動作の様子を撮影した映像から,五指の指先や関節などの特徴点を時系列データとして抽出する.機械学習を用い,このデータの周波数特性を患者と健常者に分類し,疾患の疑いを発見する.有効性を評価するため,東京医科歯科大学病院から提供された各疾患の患者ならびに健常者の映像を用い,交差検証を実施した.その結果,手根管症候群と健常者の比較では,従来のスクリーニング手法を概ね上回る感度,特異度が得られた.さらに疾患同士(手根管症候群,頚髄症)の比較にも応用しうることが示唆された.

"In the previous study, we suggested a screening method for cervical myelopathy (CM) with video analysis of the gripping and releasing (G&R) gestures whose project name is Video based CM Screening. While this screening method is based on the conventional 10-s G&R test, some other diseases also cause hand dysfunctions. This work studies if we can apply the screening method to screen carpal tunnel syndrome (CTS) and compare the diseases. We record videos of the G&R gestures to extract feature points such as fingertips and finger joints. Machine learning models classify frequency components of the feature points into the patients and the controls as the screening results. We used the videos of the patient and control groups for cross-validation to evaluate the performance. Our method showed generally higher sensitivity and specificity than the conventional screening methods. Furthermore, the results suggested that our method may also be applied to the comparison between CTS and CM.