Immunotherapy efficacy prediction through a feature re-calibrated 2.5D neural network
Haipeng Xu1, Chenxin Li2, Longfeng Zhang1
1Clinical Oncology School of Fujian Medical University, Fujian Cancer Hospital, Fujian 350014, China.
Computer Methods and Programs in Biomedicine
|April 3, 2024
Summary
This study introduces an advanced AI model for predicting lung cancer immunotherapy effectiveness using medical images. The system shows high accuracy, potentially improving treatment decisions for non-small cell lung cancer (NSCLC) patients.
Area of Science:
- Oncology
- Radiology
- Artificial Intelligence
Background:
- Lung cancer remains a major global cause of cancer mortality.
- Immunotherapy offers promise for advanced non-small cell lung cancer (NSCLC), but treatment response varies.
- Current biomarkers for immunotherapy selection have limitations, highlighting the need for novel predictive tools.
Purpose of the Study:
- To develop an automated system for predicting immunotherapy efficacy in lung cancer patients.
- To leverage imaging-based biomarkers for improved prediction of treatment outcomes.
- To enhance clinical decision-making for NSCLC immunotherapy.
Main Methods:
- An advanced 2.5D neural network architecture was employed, integrating 2D intra-slice and 3D inter-slice feature extraction.
- A lesion-focused prior and attention-based re-calibration were utilized for feature enhancement.
- An accumulated back-propagation strategy was designed for memory-efficient parameter optimization.
Main Results:
- The proposed model achieved superior performance on an in-house clinical dataset compared to state-of-the-art methods.
- The model demonstrated increased inference efficiency per subject.
- Ablation experiments validated the effectiveness of individual model components.
Conclusions:
- The developed model shows significant potential to improve physicians' diagnostic performance in predicting immunotherapy efficacy.
- The findings suggest substantial clinical application value for personalized lung cancer treatment.
- Further research is motivated by the model's effectiveness in immunotherapy efficacy prediction.
