English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Real-time algorithms for automatically recognizing surgical phases are neededto develop systems that can provide assistance to surgeons, enable bettermanagement of operating room (OR) resources and consequently improve safetywithin the OR. State-of-the-art surgical phase recognition algorithms usinglaparoscopic videos are based on fully supervised training. This limits theirpotential for widespread application, since creation of manual annotations isan expensive process considering the numerous types of existing surgeries andthe vast amount of laparoscopic videos available. In this work, we propose anew self-supervised pre-training approach based on the prediction of remainingsurgery duration (RSD) from laparoscopic videos. The RSD prediction task isused to pre-train a convolutional neural network (CNN) and long short-termmemory (LSTM) network in an end-to-end manner. Our proposed approach utilizesall available data and reduces the reliance on annotated data, therebyfacilitating the scaling up of surgical phase recognition algorithms todifferent kinds of surgeries. Additionally, we present EndoN2N, an end-to-endtrained CNN-LSTM model for surgical phase recognition and evaluate theperformance of our approach on a dataset of 120 Cholecystectomy laparoscopicvideos (Cholec120). This work also presents the first systematic study ofself-supervised pre-training approaches to understand the amount of annotationsrequired for surgical phase recognition. Interestingly, the proposed RSDpre-training approach leads to performance improvement even when all thetraining data is manually annotated and outperforms the single pre-trainingapproach for surgical phase recognition presently published in the literature.It is also observed that end-to-end training of CNN-LSTM networks boostssurgical phase recognition performance.

Less is More: Surgical Phase Recognition with Less Annotations through Self-Supervised Pre-training of CNN-LSTM Networks