Assessing the impact of a pre-processing pipeline on a tumor localization model in prostate MRI within a large multi-institutional dataset (NRG-GU005)
Multi-parametric magnetic resonance imaging (mpMRI) is increasingly recognized as a valuable tool for characterizing prostate cancer, integrating T2-weighted (T2w), diffusion-weighted (DWI), and dynamic contrast-enhanced (DCE) imaging. Despite its high sensitivity in localizing tumors, the specificity of mpMRI was shown to be hindered by benign conditions that mimic cancerous tissue features. The aim of this study is to investigate the effect of a pre-processing pipeline integrating state-of-the-art registration, bias field correction and normalization tools. We validated this pre-processing pipeline on a large multi-site dataset of 468 patients from 109 institutions. After performing all pre-processing, tumor localization was determined using a model-based tumor localization approach that takes both multi-parametric MRI and prior clinical knowledge features as input. Results show deformable image registration yielded a significant improvement in tumor localization accuracy, both for the diameter analysis as well as the area under the curve comparison for the subset of patients with ground truth tumor delineations.
Authors
- Arthur Frazier
- Uulke A. van der Heide (ORCID: https://orcid.org/0000-0002-4146-6419)
- Damien Olivié
- Felix Yi-Chung Feng (ORCID: https://orcid.org/0000-0002-0963-7687)
- Scott Carlyle Morgan (ORCID: https://orcid.org/0000-0002-6568-3346)
- Nelson Leong (ORCID: https://orcid.org/0000-0001-9410-756X)
- Abhishek A. Solanki (ORCID: https://orcid.org/0000-0002-7762-8205)
- Samuel Kadoury (ORCID: https://orcid.org/0000-0002-3048-4291)
- Marion Tonneau (ORCID: https://orcid.org/0000-0003-4776-115X)
- Thomas Boike (ORCID: https://orcid.org/0000-0003-0870-6764)
- C.A. Lawton
- Alina Mihai (ORCID: https://orcid.org/0000-0002-3558-3109)
- Clare M. Tempany (ORCID: https://orcid.org/0000-0002-1642-9720)
- Jeff M. Michalski
- Daniel J. Pennington (ORCID: https://orcid.org/0000-0001-7889-9923)
- Peter L. Choyke (ORCID: https://orcid.org/0000-0003-1086-8826)
- Stephanie Alley
- Howard Sandler
- Rodney J. Ellis
- Cynthia Ménard (ORCID: https://orcid.org/0000-0003-2609-9208)
- Baris I. Turkbey
Institutions
- Brigham and Women's Hospital (US)
- National Institutes of Health (US)
- Harvard University (US)
- University of South Florida (US)
- The Netherlands Cancer Institute (NL)
- Southeast Clinical Oncology Research Consortium (US)
- Michigan Healthcare Professionals (US)
- Polytechnique Montréal (CA)
- Université de Montréal (CA)
Publication Details
- Published
- 2025-04-02
- DOI
- https://doi.org/10.1117/12.3043835
- Primary Topic
- Radiomics and Machine Learning in Medical Imaging
- Type
- article
- Field-Weighted Citation Impact
- 0.00