Automated Lesion Segmentation in Whole-Body PET/CT in a multitracer setting

TL;DR

This paper presents an automated workflow for lesion segmentation in multitracer PET/CT images using specialized preprocessing and YOLOv8 for classification, aiming to improve accuracy across different tracers.

Contribution

It introduces a novel preprocessing pipeline and segmentation approach tailored for FDG and PSMA PET/CT images in a multitracer setting.

Findings

Enhanced segmentation accuracy demonstrated for both tracers

Effective classification of PET/CT images using YOLOv8

Open-source code available for reproducibility

Abstract

This study explores a workflow for automated segmentation of lesions in FDG and PSMA PET/CT images. Due to the substantial differences in image characteristics between FDG and PSMA, specialized preprocessing steps are required. Utilizing YOLOv8 for data classification, the FDG and PSMA images are preprocessed separately before feeding them into the segmentation models, aiming to improve lesion segmentation accuracy. The study focuses on evaluating the performance of automated segmentation workflow for multitracer PET images. The findings are expected to provide critical insights for enhancing diagnostic workflows and patient-specific treatment plans. Our code will be open-sourced and available at https://github.com/jiayiliu-pku/AP2024.