WCLC 2025 - Posters & ePosters-P2.06.98 Deep Learning-Based 18F-FDG PET/CT Radiomics Model for Predicting Pathological Response to Chemoimmunotherapy in NSCLC

P2.06.98 Deep Learning-Based 18F-FDG PET/CT Radiomics Model for Predicting Pathological Response to Chemoimmunotherapy in NSCLC

Back to course

Pdf Summary

This study developed a deep learning-based radiomics model using 18F-FDG PET/CT imaging to predict pathological response to chemoimmunotherapy in non-small cell lung cancer (NSCLC) patients. Immune checkpoint inhibitors (ICIs), particularly targeting PD-1/PD-L1, have revolutionized lung cancer treatment, but current biomarkers like PD-L1 expression and tumor mutational burden (TMB) provide limited predictive power for ICI response. Radiomics, which extracts quantitative features from medical images to capture tumor heterogeneity non-invasively, has emerged as a promising tool to improve prediction of immunotherapy outcomes.<br /><br />The authors integrated multimodal PET/CT radiomics features with deep learning techniques to create a model termed the Mask-MLP score. The model was trained and validated on patient imaging data, as outlined in the study’s data acquisition and model development workflows. Performance evaluation using ROC curves, confusion matrices, and predicted probability distributions demonstrated the model’s strong ability to discriminate responders from non-responders to chemoimmunotherapy.<br /><br />Importantly, the Mask-MLP score offers an individualized, non-invasive biomarker that can potentially guide clinical decision-making by predicting which NSCLC patients are more likely to benefit from chemoimmunotherapy. This approach aims to overcome the limitations of existing biomarkers and facilitate personalized immunotherapy strategies. Overall, the study establishes the clinical value of deep learning-enhanced PET/CT radiomics as an effective predictor of pathological response in NSCLC treated with combined chemoimmunotherapy.

Asset Subtitle

ao li

Meta Tag

Speaker ao li

Topic Pathology and Biomarkers

Keywords

deep learning

radiomics

18F-FDG PET/CT

non-small cell lung cancer

NSCLC

immune checkpoint inhibitors

PD-1/PD-L1

chemoimmunotherapy

Mask-MLP score

pathological response prediction