Automated segmentation of phases, steps, and tasks in laparoscopic cholecystectomy using deep learning.

BACKGROUND: Video-based review is paramount for operative performance assessment but can be laborious when performed manually. Hierarchical Task Analysis (HTA) is a well-known method that divides any procedure into phases, steps, and tasks. HTA requires large datasets of videos with consistent definitions at each level. Our aim was to develop an AI model for automated segmentation of phases, steps, and tasks for laparoscopic cholecystectomy videos using a standardized HTA. METHODS: A total of 160 laparoscopic cholecystectomy videos were collected from a publicly available dataset known as cholec80 and from our own institution. All videos were annotated for the beginning and ending of a predefined set of phases, steps, and tasks. Deep learning models were then separately developed and trained for the three levels using a 3D Convolutional Neural Network architecture. RESULTS: Four phases, eight steps, and nineteen tasks were defined through expert consensus. The training set for our deep learning models contained 100 videos with an additional 20 videos for hyperparameter optimization and tuning. The remaining 40 videos were used for testing the performance. The overall accuracy for phases, steps, and tasks were 0.90, 0.81, and 0.65 with the average F1 score of 0.86, 0.76 and 0.48 respectively. Control of bleeding and bile spillage tasks were most variable in definition, operative management, and clinical relevance. CONCLUSION: The use of hierarchical task analysis for surgical video analysis has numerous applications in AI-based automated systems. Our results show that our tiered method of task analysis can successfully be used to train a DL model.

Overcoming Health Literacy Barriers by Developing Standardized Surgical Discharge Instructions.

INTRODUCTION: Approximately one-third of surgical patients exhibit low health literacy, and 39% of our patients are primary Spanish speakers. We first evaluated the current content of our arteriovenous fistula/graft discharge instruction (DCI) templates. Using the Plan-Do-Study-Act cycle quality improvement methodology, we then aimed to optimize the readability and formally translate new DCI and evaluate usage and inappropriate bouncebacks following implementation. METHODS: Current arteriovenous fistula/graft template content was reviewed by the literacy department for readability and vascular faculty for completeness and accuracy. The literacy department edits were categorized by word choice, added/removed content, format change, and grammatical errors. Two vascular surgeons rated completeness and accuracy on a Likert scale (1-5). Retrospective chart review was performed for telephone calls and emergency department bouncebacks for 3 mo flanking new DCI implementation. RESULTS: Of the 10 templates, all were in English and word count ranged from 192 to 990 words. Despite each template including all necessary subcategories, the median number of edits per 100 words was 9.2 [7.0-9.5]. Approximately half of the edits (5.4 [5.1-5.5]) were word choice edits. Overall, experts rated completeness at 3.9 [3.2-4.2] and accuracy at 4.0 [3.7-4.1]. Highest template utilization occurred during post-implementation months 1 (90%) and 3 (100%) with orientation sessions. There was a significant increase in concordant Spanish DCI use (P < 0.01) and no inappropriate bouncebacks after implementation. CONCLUSIONS: Our study demonstrated notable variability in the content and readability of our vascular access instruction templates. New DCI had strong usage and language concordance; continued use may decrease bouncebacks.