AutoPET III Challenge: PET/CT Semantic Segmentation

TL;DR

This paper presents a two-stage deep learning approach with model ensembling for precise lesion segmentation in PET/CT images, aiming to improve robustness and accuracy in the AutoPET III challenge.

Contribution

It introduces a novel multi-stage segmentation framework combining DynUNet and ensemble models for enhanced PET/CT lesion segmentation.

Findings

Achieved improved segmentation accuracy through multi-stage approach

Enhanced robustness with data augmentation techniques

Demonstrated effectiveness of model ensembling in medical image segmentation

Abstract

In this study, we implemented a two-stage deep learning-based approach to segment lesions in PET/CT images for the AutoPET III challenge. The first stage utilized a DynUNet model for coarse segmentation, identifying broad regions of interest. The second stage refined this segmentation using an ensemble of SwinUNETR, SegResNet, and UNet models. Preprocessing involved resampling images to a common resolution and normalization, while data augmentation techniques such as affine transformations and intensity adjustments were applied to enhance model generalization. The dataset was split into 80% training and 20% validation, excluding healthy cases. This method leverages multi-stage segmentation and model ensembling to achieve precise lesion segmentation, aiming to improve robustness and overall performance.