Deep Learning Approaches for Predicting Treatment Response in Lung Cancer

Overview

This study developed a multitask deep learning framework to predict treatment response and progression-free survival (PFS) in lung cancer patients using sequential cone-beam computed tomography (CBCT) imaging.

Background

Lung cancer is the most commonly diagnosed cancer and a leading cause of cancer-related deaths globally. Early identification of treatment response is crucial for optimizing therapeutic strategies in patients undergoing radiotherapy. Traditional prognostic models often fail to account for the biological variability in treatment responses, highlighting the need for advanced predictive methodologies.

Data Highlights

A total of 142 lung cancer patients were analyzed, with the model achieving an area under the receiver operating characteristic curve (AUC) of 0.858 ± 0.078 and a concordance index (C-index) of 0.672 ± 0.058 using only the first on-treatment CBCT.

Key Findings

• The multitask deep learning framework was trained end-to-end for treatment response classification and survival prediction.
• The first on-treatment CBCT scan provided the strongest predictive signal for treatment response classification.
• Adding subsequent CBCT scans degraded classification performance due to increased dimensionality relative to cohort size.
• Ablation analysis confirmed the contributions of attention fusion and multitask training to model performance.
• The study benchmarked the proposed model against 12 baseline methods using 5-fold cross-validation.

Clinical Implications

The findings indicate that utilizing the first on-treatment CBCT scan may enhance early prediction of treatment response in lung cancer patients.

Conclusion

The study demonstrates the potential of deep learning frameworks in predicting treatment outcomes in lung cancer radiotherapy.

Related Resources & Content

1. ASCO AI in Oncology, ASCO AI, 2026 -- Improved Immunotherapy Response Prediction in NSCLC With Deep-Learning Radiomic Biomarker
2. The ASCO Post, The ASCO Post, 2026 -- Deep-Learning CT Biomarker Predicts Survival Better Than Traditional Measures in Immunotherapy-Treated Advanced NSCLC
3. Durvalumab after Chemoradiotherapy in Stage III Non–Small-Cell Lung Cancer | New England Journal of Medicine
4. asco ai in oncology — Improved Immunotherapy Response Prediction in NSCLC With Deep-Learning Radiomic Biomarker
5. The ASCO Post — Deep-Learning CT Biomarker Predicts Survival Better Than Traditional Measures in Immunotherapy-Treated Advanced NSCLC
6. Frontiers in Oncology — CT-based radiomics combined with deep learning for predicting radiation pneumonitis in patients with esophageal cancer: A two-center study
7. RECIST 1.1 - RECIST
8. The Image Biomarker Standardization Initiative: Standardized Convolutional Filters for Reproducible Radiomics and Enhanced Clinical Insights - PubMed
9. Durvalumab after Chemoradiotherapy in Stage III Non–Small-Cell Lung Cancer | New England Journal of Medicine