RGBカメラを用いた頚髄症スクリーニング手法の提案
Video-Based Hand Tracking for Screening Cervical Myelopathy
2021
松井良太，小山恭史，藤田浩二，斎藤英雄，杉浦裕太
Ryota Matsui, Takafumi Koyama, Koji Fujita, Hideo Saito, Yuta Sugiura
Reference / 引用するならこちら
Matsui R., Koyama T., Fujita K., Saito H., Sugiura Y. (2021) Video-Based Hand Tracking for Screening Cervical Myelopathy. In: Bebis G. et al. (eds) Advances in Visual Computing. ISVC 2021. Lecture Notes in Computer Science, vol 13018. Springer, Cham. [DOI]
本研究では、グーパー運動の様子を録画した映像を、機械学習を用いて分析する頚髄症スクリーニング手法を提案する。机上に置かれたスマートフォンの直上で内蔵カメラで録画する。次に画像処理を用い、この映像から手指の動きを特徴量として抽出する。この特徴量は前処理の後に、サポートベクタマシン（SVM）分類モデルを用いて患者と健常者の 2クラスに分類される。
Cervical myelopathy (CM) is a pathology caused by cervical spinal cord compression. Spinal surgeons often use the 10-sec grip and release (G&R) test to screen hand disorders, a typical symptom of CM. We propose a screening method for CM based on videos of the G&R test and machine learning. Each patient holds their hand above a smartphone to record the G&R movement as a video with the built-in camera. We use an image-processing framework to obtain feature values of the hand movement. A support vector machine classifier estimates if these feature values suggest any characteristics of CM patients. We conducted a user experiment on 20 CM patients and 15 controls to evaluate our method. As a result, sensitivity, specificity, and area under the curve were 90.0%, 93.3%, and 0.947, respectively. This performance is higher than the conventional methods.