Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Enhancing Lesion Segmentation in PET/CT Imaging with Deep Learning and Advanced Data Preprocessing Techniques

Jiayi Liu, Qiaoyi Xue, Youdan Feng, Tianming Xu, Kaixin Shen, Chuyun Shen, Yuhang Shi

TL;DR

This work tackles the challenge of accurate lesion segmentation in PET/CT imaging for FDG and PSMA tracers, addressing the scarcity of public PET data by evaluating standardized preprocessing and augmentation strategies. It proposes a two-stage pipeline: a YOLOv8 tracer classifier to identify the tracer, followed by DynUNet-based 3D segmentation trained separately on FDG and PSMA data. Key findings show that basic preprocessing yields moderate Dice scores, with substantial gains from Gaussian sharpening and intensity clipping (ClipValMax=280), alongside non-zero normalization and GammaTransform adjustments, highlighting the importance of data preparation in PET/CT segmentation. The study provides evidence that tailored preprocessing and augmentation can meaningfully boost diagnostic accuracy, and it contributions include open-source code to facilitate reproducibility and standardization across the field.

Abstract

The escalating global cancer burden underscores the critical need for precise diagnostic tools in oncology. This research employs deep learning to enhance lesion segmentation in PET/CT imaging, utilizing a dataset of 900 whole-body FDG-PET/CT and 600 PSMA-PET/CT studies from the AutoPET challenge III. Our methodical approach includes robust preprocessing and data augmentation techniques to ensure model robustness and generalizability. We investigate the influence of non-zero normalization and modifications to the data augmentation pipeline, such as the introduction of RandGaussianSharpen and adjustments to the Gamma transform parameter. This study aims to contribute to the standardization of preprocessing and augmentation strategies in PET/CT imaging, potentially improving the diagnostic accuracy and the personalized management of cancer patients. Our code will be open-sourced and available at https://github.com/jiayiliu-pku/DC2024.

Enhancing Lesion Segmentation in PET/CT Imaging with Deep Learning and Advanced Data Preprocessing Techniques

TL;DR

This work tackles the challenge of accurate lesion segmentation in PET/CT imaging for FDG and PSMA tracers, addressing the scarcity of public PET data by evaluating standardized preprocessing and augmentation strategies. It proposes a two-stage pipeline: a YOLOv8 tracer classifier to identify the tracer, followed by DynUNet-based 3D segmentation trained separately on FDG and PSMA data. Key findings show that basic preprocessing yields moderate Dice scores, with substantial gains from Gaussian sharpening and intensity clipping (ClipValMax=280), alongside non-zero normalization and GammaTransform adjustments, highlighting the importance of data preparation in PET/CT segmentation. The study provides evidence that tailored preprocessing and augmentation can meaningfully boost diagnostic accuracy, and it contributions include open-source code to facilitate reproducibility and standardization across the field.

Abstract

The escalating global cancer burden underscores the critical need for precise diagnostic tools in oncology. This research employs deep learning to enhance lesion segmentation in PET/CT imaging, utilizing a dataset of 900 whole-body FDG-PET/CT and 600 PSMA-PET/CT studies from the AutoPET challenge III. Our methodical approach includes robust preprocessing and data augmentation techniques to ensure model robustness and generalizability. We investigate the influence of non-zero normalization and modifications to the data augmentation pipeline, such as the introduction of RandGaussianSharpen and adjustments to the Gamma transform parameter. This study aims to contribute to the standardization of preprocessing and augmentation strategies in PET/CT imaging, potentially improving the diagnostic accuracy and the personalized management of cancer patients. Our code will be open-sourced and available at https://github.com/jiayiliu-pku/DC2024.
Paper Structure (8 sections, 1 figure, 1 table)

This paper contains 8 sections, 1 figure, 1 table.

Table of Contents

  1. Introduction
  2. Methods
  3. Data and preprocessing
  4. Model architecture and training
  5. Yolo model
  6. DynUNet model
  7. Results
  8. Conclusion

Figures (1)

  • Figure 1: The workflow of automated lesion segmentation of FDG PET images and PSMA PET images.