Date of Award

8-2024

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Electrical and Computer Engineering

Committee Chair/Advisor

Dr. Richard E. Groff

Committee Member

Dr. Ravikiran Singapogu

Committee Member

Dr. Xiaolong Ma

Committee Member

Dr. Ian D. Walker

Abstract

Surgical suturing skill assessment is a crucial part of surgical education. Vascular surgery educators have developed a simulation-based examination called Fundamentals of Vascular Surgery, which includes a clock-face model for assessing open surgical suturing skills. The clock-face model, however, requires the valuable time of expert surgeons to determine examinees' skills. Moreover, expert surgeons have different judgments for appropriate sutures, which leads to inconsistent grading. These limitations motivate us to use sensors to measure examinees' needle motions and hand motions during the clock-face suturing exercises, and then use the measurements for objective suturing skill assessment.

To assess suturing skills based on needle motions, computer vision is used to track the needle trajectory. Skill assessment metrics are then calculated based on the estimated needle trajectory. Improving on a prior design, a more robust suturing simulator was designed and constructed. The new simulator was used to collect data from 97 participants, with different years of surgical experience, who were separated into novice, intermediate, and expert groups. Statistical analysis shows that all the image-based metrics have significantly different means between the novice and the remaining two groups. The groups are even more distinguished when analyzing suturing at the depth condition, for which six of the seven metrics have significantly different means between the intermediate and expert groups. Further, the metric classification performance is examined by ROC curves. The results show that the median metric value for a suturing exercise has higher classification accuracy than the metric value for a suture.

To assess suturing skills based on hand motion, a deep-learning program is developed to analyze hand-motion videos, and then estimate hand rotation at the roll axis. The program includes a hand detection algorithm and a hand roll estimation algorithm. The roll motion estimate is then used to calculate hand-roll metrics for suturing skill assessment. The hand detection algorithm receives videos and then localizes participants' dominant hand in video frames. Results show that the hand detection algorithm precisely localizes participants' dominant hand in videos with different brightness and backgrounds. The hand roll estimation algorithm receives cropped hand images and then outputs the corresponding hand roll angle or roll velocity. Results show that the hand roll estimation algorithm has consistent performance at different locations. To examine if the algorithm output can be used for surgical suturing skill assessment, the algorithm output and the IMU measurements are used to calculate hand-roll metrics that capture participants' hand roll during the clock-face suturing task. Results show that the estimated hand roll angle has similar statistics as the IMU measurement when calculating the hand-roll metrics. Specifically, 4 out of the 5 metrics have significantly different means between the novice group and the remaining two groups. Also, 4 out of the 5 metrics have significant mean differences between the resident surgeons' and attending surgeons' groups.

Share

COinS
 
 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.