SEDMamba: Enhancing Selective State Space Modelling with Bottleneck Mechanism and Fine-to-Coarse Temporal Fusion for Efficient Error Detection in Robot-Assisted Surgery

TL;DR

SEDMamba is a hierarchical model that improves surgical error detection in long videos by efficiently capturing long-term dependencies using a novel bottleneck mechanism and multi-scale temporal fusion, supported by a new in-vivo error dataset.

Contribution

The paper introduces SEDMamba, a novel hierarchical model with a bottleneck mechanism and fine-to-coarse temporal fusion for efficient long sequence error detection in surgical videos, and provides the first in-vivo error dataset.

Findings

SEDMamba outperforms state-of-the-art methods by at least 1.82% AUC and 3.80% AP.

The model achieves linear complexity in long sequence modeling.

A new in-vivo surgical error dataset is introduced.

Abstract

Automated detection of surgical errors can improve robotic-assisted surgery. Despite promising progress, existing methods still face challenges in capturing rich temporal context to establish long-term dependencies while maintaining computational efficiency. In this paper, we propose a novel hierarchical model named SEDMamba, which incorporates the selective state space model (SSM) into surgical error detection, facilitating efficient long sequence modelling with linear complexity. SEDMamba enhances selective SSM with a bottleneck mechanism and fine-to-coarse temporal fusion (FCTF) to detect and temporally localize surgical errors in long videos. The bottleneck mechanism compresses and restores features within their spatial dimension, thereby reducing computational complexity. FCTF utilizes multiple dilated 1D convolutional layers to merge temporal information across diverse scale ranges, accommodating errors of varying duration. Our work also contributes the first-of-its-kind, frame-level, in-vivo surgical error dataset to support error detection in real surgical cases. Specifically, we deploy the clinically validated observational clinical human reliability assessment tool (OCHRA) to annotate the errors during suturing tasks in an open-source radical prostatectomy dataset (SAR-RARP50). Experimental results demonstrate that our SEDMamba outperforms state-of-the-art methods with at least 1.82% AUC and 3.80% AP performance gains with significantly reduced computational complexity. The corresponding error annotations, code and models are released at https://github.com/wzjialang/SEDMamba.