Predicting brain metastases originating from different pathological subtypes of lung cancer based on multimodal MRI using DL approach

Motivation: It is still a challenge to confirm the origin pathological subtypes of lung cancer through brain metastases (BMs). Goal(s): To investigate the feasibility of a deep learning (DL) approach based on multimodal MRI to predict BMs originating from different pathological subtypes of lung cancer. Approach: We employ DL approach ResNet-18 as the basic classification framework and perform classification detection on T2 FLAIR, DWI, ADC, and T1CE sequences. The discrimination performances of those sequences were accessed by using receiver operating characteristic curve analysis. Results: DL serves as a non-invasive tool for the exceptional prediction of BMs originating from different pathological subtypes of lung cancer. Impact: Deep learning based on multimodal MRI can be a helpful tool for predicts brain metastases originating from different pathological subtypes of lung cancer. Apparent diffusion coefficient maps and T1-weighted contrast enhancement sequence exhibiting optimal predictive performance.