2023 World Conference on Lung Cancer (Posters)-P1.13. A Multi-Stage Imaging-Based Deep Learning Classifier for EGFR-TKI Response Prediction in EGFR mutant Non-small Cell Lung Cancer

P1.13. A Multi-Stage Imaging-Based Deep Learning Classifier for EGFR-TKI Response Prediction in EGFR mutant Non-small Cell Lung Cancer - PDF(Slides)

Back to course

Pdf Summary

Researchers have developed a deep learning model called MSRNet that can predict the response to epidermal growth factor receptor (EGFR)-tyrosine kinase inhibitor (TKI) therapy in patients with EGFR mutation-positive non-small cell lung cancer (NSCLC). The model was trained using CT images collected from patients before treatment and one month after treatment from three healthcare providers. MSRNet consists of a CT preprocessing pipeline, a CT feature extractor, and a drug resistance risk predictor. The study included patients with histopathology-confirmed stage IIIB-IV lung adenocarcinoma and Exon19 del or Exon21 L858R mutation who received TKI targeted therapy as a first-line treatment. The patients were divided into one training cohort and two independent external validation cohorts. The lightweight CNN backbone of MSRNet extracted semantic features from the CT images, and the end-to-end risk score prediction network was optimized using a survival loss function. The trained model achieved high accuracy in predicting the risk of progression in EGFR TKI-treated NSCLC patients. The non-invasive and convenient nature of the model provides doctors with ideas for choosing individualized treatment plans and can potentially buy patients more time before the development of drug resistance. The model has strong operability with low input requirements, making it suitable for dynamic chest CT imaging. The study was conducted by a multidisciplinary team of researchers from different healthcare institutions. Further details about the study's methodology, results, and conclusion can be found in the provided document.

Asset Subtitle

Yang Xia

Meta Tag

Speaker Yang Xia

Topic Screening & Early Detection: Biomarkers/Imaging Technology

Keywords

MSRNet

deep learning model

EGFR-tyrosine kinase inhibitor

non-small cell lung cancer

CT images

drug resistance risk predictor

Exon19 del

Exon21 L858R mutation

individualized treatment plans

dynamic chest CT imaging