Class-Incremental Domain Adaptation with Smoothing and Calibration for Surgical Report Generation

TL;DR

This paper introduces a class-incremental domain adaptation framework using smoothing and calibration techniques to improve surgical report generation across different robotic surgery domains, handling new classes and domain shifts effectively.

Contribution

It proposes a novel multi-layer transformer-based model with class-incremental learning, contrastive loss, Gaussian smoothing, and label smoothing for robust surgical report generation.

Findings

Improved performance in surgical report generation across domains.

Effective handling of unseen classes with few-shot learning.

Enhanced domain invariance and calibration in predictions.

Abstract

Generating surgical reports aimed at surgical scene understanding in robot-assisted surgery can contribute to documenting entry tasks and post-operative analysis. Despite the impressive outcome, the deep learning model degrades the performance when applied to different domains encountering domain shifts. In addition, there are new instruments and variations in surgical tissues appeared in robotic surgery. In this work, we propose class-incremental domain adaptation (CIDA) with a multi-layer transformer-based model to tackle the new classes and domain shift in the target domain to generate surgical reports during robotic surgery. To adapt incremental classes and extract domain invariant features, a class-incremental (CI) learning method with supervised contrastive (SupCon) loss is incorporated with a feature extractor. To generate caption from the extracted feature, curriculum by one-dimensional gaussian smoothing (CBS) is integrated with a multi-layer transformer-based caption prediction model. CBS smoothes the features embedding using anti-aliasing and helps the model to learn domain invariant features. We also adopt label smoothing (LS) to calibrate prediction probability and obtain better feature representation with both feature extractor and captioning model. The proposed techniques are empirically evaluated by using the datasets of two surgical domains, such as nephrectomy operations and transoral robotic surgery. We observe that domain invariant feature learning and the well-calibrated network improves the surgical report generation performance in both source and target domain under domain shift and unseen classes in the manners of one-shot and few-shot learning. The code is publicly available at https://github.com/XuMengyaAmy/CIDACaptioning.