Related Experiment Video
Updated: Oct 4, 2025
Dynamic Contrast Enhanced Magnetic Resonance Imaging of an Orthotopic Pancreatic Cancer Mouse Model
Published on: April 18, 2015
MRI Radiomics Model Predicts Pathologic Complete Response of Rectal Cancer Following Chemoradiotherapy
Jaeseung Shin1, Nieun Seo1, Song-Ee Baek1
1From the Department of Radiology and Research Institute of Radiological Science, Severance Hospital, Yonsei University College of Medicine, 50-1 Yonsei-ro, Seodaemun-gu, Seoul 03722, South Korea (J.S., N.S., S.E.B., J.S.L., S.K.); Data Science Team, Center for Digital Health, Yongin Severance Hospital, Yonsei University College of Medicine, Yongin, South Korea (N.H.S.); and Departments of Surgical Oncology (N.K.K.) and Radiation Oncology (W.S.K.), Yonsei University College of Medicine, Seoul, South Korea.
An MRI radiomics model accurately predicts pathologic complete response (pCR) in locally advanced rectal cancer (LARC) after neoadjuvant chemoradiotherapy (nCRT). This AI-driven approach outperforms radiologists in diagnosing pCR, aiding organ preservation decisions.
Area of Science:
- Oncology
- Radiology
- Medical Imaging Analysis
Background:
- Accurate preoperative assessment of pathologic complete response (pCR) is crucial for organ preservation in locally advanced rectal cancer (LARC) patients undergoing neoadjuvant chemoradiotherapy (nCRT).
- Large-scale validation of MRI radiomics models for predicting pCR in LARC is currently lacking.
Purpose of the Study:
- To evaluate the performance of MRI radiomics models (T2-weighted and diffusion-weighted imaging) in predicting pCR after nCRT in LARC.
- To compare the diagnostic accuracy of these radiomics models against visual assessment by experienced radiologists.
Main Methods:
- Retrospective analysis of 898 LARC patients who underwent post-nCRT MRI and resection.
- Radiomic features extracted from T2-weighted and apparent diffusion coefficient (ADC) MRI images to build prediction models.
- Comparison of radiomics model performance (AUC) with pooled radiologist assessments using the DeLong method.
Main Results:
- The merged T2-weighted and ADC radiomics model achieved an AUC of 0.82, comparable to the T2-weighted only model (AUC 0.82).
- The T2-weighted radiomics model demonstrated superior classification performance (AUC 0.82) compared to pooled radiologists (AUC 0.74).
- The T2-weighted model showed higher sensitivity (80.0%) but lower specificity (68.4%) than radiologists (15.6% sensitivity, 98.6% specificity).
Conclusions:
- MRI-based radiomics models, particularly using T2-weighted imaging, show significant potential for predicting pCR in LARC patients post-nCRT.
- The developed radiomics model exhibited better classification performance than experienced radiologists, suggesting its utility in clinical decision-making for organ preservation.
