Primary Tumor and Lung Cancer Prediction on Dual Energy CT Images for Nodal Metastasis using Deep Learning Technique

One of the most important clinical responsibilities for lung cancer (LC) survival and recurrence is detecting lymph node metastases (LNM). It is still challenging to identify medical strategies and pretreatment choices for patients with malignancies based on preoperative identification of nodal metastases. The quantitative properties of shape, first-order, second-order, and higher-order textures in computed tomography (CT) are used to identify tumors. Statistical and machine learning (ML) studies are utilized to integrate clinical data and assess the attributes of their individual. This research discovered several radiomic features associated with distant, nodal, and histological spread, which has the great capability as an imaging biomarker for pathological diagnosis and target therapy selection. To detect nodal metastases (Nmets) from primary LC tumors, a new deep prediction approach with a Size-Related Damper Block (SR-DB) has been developed using Gemstone Spectral Imaging (GSI) dual-energy CT. In an implementation, a recommended strategy for Nmet prediction trained on the 40 keV dataset offers an accuracy of 0.89 percent and a sensitivity score of 0.95 percent. The results show that tumor heterogeneity and size aid in determining whether nodal metastasis from the initial tumor is absent or present in the proposed model.