Related Experiment Video
Updated: Sep 18, 2025
11:31
Using Micro-computed Tomography for the Assessment of Tumor Development and Follow-up of Response to Treatment in a Mouse Model of Lung Cancer
Published on: May 20, 2016
10.8K
Radiomics Models to Predict Tumor Response and Pneumonitis in Non-Small Cell Lung Cancer Patients Treated with
Monica Yadav1, Wongi Woo2, Young Kwang Chae1
1Feinberg School of Medicine, Northwestern University, NMH/Arkes Family Pavilion Suite 800, 676 N Saint Clair, Chicago, IL 60611, USA.
Journal of Clinical Medicine
|June 26, 2025
Summary
Artificial intelligence (AI) radiomics can predict checkpoint inhibitor-associated pneumonitis (CIP) and tumor response in non-small cell lung cancer (NSCLC) patients undergoing immunotherapy. These AI models show potential for personalized treatment strategies.
Area of Science:
- Oncology
- Radiology
- Artificial Intelligence
- Medical Imaging
Background:
- Checkpoint inhibitor-associated pneumonitis (CIP) is a significant challenge for patients with non-small cell lung cancer (NSCLC) receiving immunotherapy.
- Predicting CIP and treatment response is crucial for optimizing patient management and outcomes in advanced NSCLC.
Purpose of the Study:
- To investigate the potential of artificial intelligence (AI) algorithms analyzing radiomic features from pre-treatment CT scans to predict CIP occurrence in NSCLC patients.
- To evaluate the ability of these radiomic models to predict tumor response to immunotherapy.
- To assess the association between clinical factors and the development of CIP.
Main Methods:
- Analysis of pre-treatment CT scans from 159 stage III-IV NSCLC patients undergoing immunotherapy.
- Extraction of 3D radiomic features from tumors and surrounding regions using LIFEx software.
- Application of a linear mixed-effect model for radiomics harmonization and a random forest algorithm for predictive model development.
- Evaluation of model performance using the area under the curve (AUC).
Main Results:
- Radiomics analysis demonstrated predictability for CIP with an AUC of 0.60 (95% CI 0.55-0.66).
- Predictive models for tumor response achieved AUCs of 0.63 (irRECIST) and 0.66 (RECIST 1.1).
- Patients who developed pneumonitis were more likely to be male, have less adenocarcinoma histology, and exhibit higher tumor mutational burden.
Conclusions:
- Pre-treatment CT-based radiomic models show promise in predicting CIP and tumor response in NSCLC patients treated with immunotherapy.
- These findings suggest that radiomics could aid in identifying patients at risk for CIP and predicting treatment efficacy.
- Further validation is warranted to integrate radiomic predictions into clinical decision-making for NSCLC immunotherapy.
