Large-scale surgical workflow segmentation for laparoscopic sacrocolpopexy

Sophia Bano,Jan Deprest,Danail Stoyanov,Ann-Sophie Page,Yitong Zhang,Francisco Vasconcelos

doi:10.1007/s11548-021-02544-5

Abstract

PurposeLaparoscopic sacrocolpopexy is the gold standard procedure for the management of vaginal vault prolapse. Studying surgical skills and different approaches to this procedure requires an analysis at the level of each of its individual phases, thus motivating investigation of automated surgical workflow for expediting this research. Phase durations in this procedure are significantly larger and more variable than commonly available benchmarks such as Cholec80, and we assess these differences.MethodologyWe introduce sequence-to-sequence (seq2seq) models for coarse-level phase segmentation in order to deal with highly variable phase durations in Sacrocolpopexy. Multiple architectures (LSTM and transformer), configurations (time-shifted, time-synchronous), and training strategies are tested with this novel framework to explore its flexibility.ResultsWe perform 7-fold cross-validation on a dataset with 14 complete videos of sacrocolpopexy. We perform both a frame-based (accuracy, F1-score) and an event-based (Ward metric) evaluation of our algorithms and show that different architectures present a trade-off between higher number of accurate frames (LSTM, Mode average) or more consistent ordering of phase transitions (Transformer). We compare the implementations on the widely used Cholec80 dataset and verify that relative performances are different to those in Sacrocolpopexy.ConclusionsWe show that workflow segmentation of Sacrocolpopexy videos has specific challenges that are different to the widely used benchmark Cholec80 and require dedicated approaches to deal with the significantly larger phase durations. We demonstrate the feasibility of seq2seq models in Sacrocolpopexy, a broad framework that can be further explored with new configurations. We show that an event-based evaluation metric is useful to evaluate workflow segmentation algorithms and provides complementary insight to the more commonly used metrics such as accuracy or F1-score.

Highlights

Half of the women above 50 years of age suffer from pelvic organ prolapse, with a lifetime prevalence of 30 ± 50% [24], and in particular, vaginal vault prolapse is a frequent occurrence after hysterectomy [10]
Laparoscopic sacrocolpopexy aims at fixating the vaginal vault using a mesh implant that is permanently sutured or stapled to the sacral promontory
As the evaluation metrics used in most of other works, the accuracy here is calculated based on the whole videos rather than a per phase macro-average

Summary

Conclusions

We show that workflow segmentation of Sacrocolpopexy videos has specific challenges that are different to the widely used benchmark Cholec and require dedicated approaches to deal with the significantly larger phase durations. We demonstrate the feasibility of seq2seq models in Sacrocolpopexy, a broad framework that can be further explored with new configurations. We show that an event-based evaluation metric is useful to evaluate workflow segmentation algorithms and provides complementary insight to the more commonly used metrics such as accuracy or F1-score. Keywords Surgical workflow segmentation · Machine learning · Laparoscopic sacrocolpopexy · Long short-term memory networks · Transformer networks

Introduction

Experimental setup

Evaluation metrics

Results and discussion

Method

Conclusion