Fig. 1From: Exploring non-invasive precision treatment in non-small cell lung cancer patients through deep learning radiomics across imaging features and molecular phenotypesOverall study design. Step 1: CT image preprocessing, three ROIs are utilized as input images. Step 2: CT deep learning model schematic. The pre-trained DenseNet121 is employed as the backbone network to encode input images into network features. Step 3: Radiomics model workflow. Step 4: Development and integration details of the deep learning radiomics model. ROI: region of interestBack to article page