LoViT: Long Video Transformer for surgical phase recognition

Online surgical phase recognition plays a significant role towards building contextual tools that could quantify performance and oversee the execution of surgical workflows. Current approaches are limited since they train spatial feature extractors using frame-level supervision that could lead to incorrect predictions due to similar frames appearing at different phases, and poorly fuse local and global features due to computational constraints which can affect the analysis of long videos commonly encountered in surgical interventions. In this paper, we present a two-stage method, called Long Video Transformer (LoViT), emphasizing the development of a temporally-rich spatial feature extractor and a phase transition map. The temporally-rich spatial feature extractor is designed to capture critical temporal information within the surgical video frames. The phase transition map provides essential insights into the dynamic transitions between different surgical phases. LoViT combines these innovations with a multiscale temporal aggregator consisting of two cascaded L-Trans modules based on self-attention, followed by a G-Informer module based on ProbSparse self-attention for processing global temporal information. The multi-scale temporal head then leverages the temporally-rich spatial features and phase transition map to classify surgical phases using phase transition-aware supervision. Our approach outperforms state-of-the-art methods on the Cholec80 and AutoLaparo datasets consistently. Compared to Trans-SVNet, LoViT achieves a 2.4 pp (percentage point) improvement in video-level accuracy on Cholec80 and a 3.1 pp improvement on AutoLaparo. Our results demonstrate the effectiveness of our approach in achieving state-of-the-art performance of surgical phase recognition on two datasets of different surgical procedures and temporal sequencing characteristics. The project page is available at https://github.com/MRUIL/LoViT.