Laboratory of Biomedical Informatics and Artificial Intelligence, Department of Neurosurgery, National Medical Research Center for Neurosurgery named after N.N. Burdenko, Moscow, Russian Federation.
Faculty of Computational Mathematics and Cybernetics, Lomonosov Moscow State University, Moscow, Russian Federation.
Stud Health Technol Inform. 2024 Aug 22;316:934-938. doi: 10.3233/SHTI240564.
Objective evaluation of microsurgical technique quality is vital for successful training in neurosurgery. This study aimed to assess the accuracy of automatically detecting a neurosurgeon's proper posture and hand positioning using computer vision. We employed the RTMPose neural network model to identify key anatomical points in the neurosurgeon's projection and calculated various angles formed by connecting these points. By utilizing machine learning on these angles, we were able to classify images of the surgeon's posture and hands into correct positions and various types of errors with an accuracy of at least 0.9. Computer vision enables successful detection and objective assessment of the neurosurgeon's posture and hand positions. The high accuracy of this detection can pave the way for a new training approach in neurosurgery.
客观评估显微外科技术质量对于神经外科的成功培训至关重要。本研究旨在评估计算机视觉自动检测神经外科医生正确姿势和手部位置的准确性。我们使用 RTMPose 神经网络模型来识别神经外科医生投影中的关键解剖点，并计算连接这些点形成的各种角度。通过对这些角度进行机器学习，我们能够将外科医生姿势和手部的图像分类为正确位置和各种类型的错误，准确率至少为 0.9。计算机视觉可以成功检测和客观评估神经外科医生的姿势和手部位置。这种检测的高精度可以为神经外科的新培训方法铺平道路。
