JIGSAWS (JHU-ISI Gesture and Skill Assessment Working Set)

JHU-ISI 手势和技能评估工作集 (JIGSAWS) 是用于人体运动建模的手术活动数据集。这些数据是通过约翰霍普金斯大学 (JHU) 和 Intuitive Surgical, Inc. (Sunnyvale, CA. ISI) 在一项 IRB 批准的研究中的合作收集的。该数据集的发布已获得约翰霍普金斯大学 IRB 的批准。该数据集是使用达芬奇手术系统从八位具有不同技能水平的外科医生在台式模型上执行三项基本手术任务的五次重复获得的：缝合、打结和穿针，这是大多数外科手术的标准组件技能培训课程。 JIGSAWS 数据集由三个部分组成：
运动学数据：描述机械手运动的笛卡尔位置、方向、速度、角速度和夹持器角度。
视频数据：从内窥镜摄像机捕获的立体视频。可以在此处下载 JIGSAWS 任务的示例视频：缝合、打结、穿针。
手动注释包括：手势（原子手术活动段标签）。
技能（使用修改后的客观结构化技术技能评估的全球评分）。
实验装置：一种标准化的交叉验证实验装置，可用于评估自动手术手势识别和技能评估方法。我们小组一直在使用这个实验装置来评估和比较我们开发的方法。

JHU-ISI Gesture and Skill Assessment Working Set (JIGSAWS) is a surgical activity dataset for human motion modeling. These data were collected via a collaboration between Johns Hopkins University (JHU) and Intuitive Surgical, Inc. (Sunnyvale, CA, ISI) under an IRB-approved study, and its release has been approved by the Johns Hopkins University IRB.

The dataset was acquired using the da Vinci Surgical System, with five repetitions of three basic surgical tasks performed by eight surgeons with varying skill levels on a bench model: suturing, knot tying, and needle passing, which are standard component skills in most surgical skill training curricula.

The JIGSAWS dataset consists of three components:
1. Kinematic data: Cartesian positions, orientations, velocities, angular velocities, and gripper angles describing manipulator motion.
2. Video data: Stereoscopic video captured from an endoscopic camera. Example videos of JIGSAWS tasks (suturing, knot tying, needle passing) can be downloaded here.
3. Manual annotations, including:
- Gesture: labels for atomic surgical activity segments
- Skill: global rating scores using the modified Objective Structured Technical Skills Assessment

Experimental setup: A standardized cross-validation experimental setup that can be used to evaluate automated surgical gesture recognition and skill assessment methods. Our research team has consistently utilized this setup to evaluate and compare the methods we have developed.

搜集汇总

背景与挑战

背景概述

JIGSAWS是一个手术活动数据集，用于人体运动建模，由约翰霍普金斯大学与Intuitive Surgical合作收集，通过达芬奇手术系统采集八位外科医生执行缝合、打结和穿针任务的运动学数据、立体视频和手动注释。该数据集旨在支持自动手术手势识别和技能评估方法的实验评估。