Integrating deep learning and radiomics in the differentiation of major histological subtypes of invasive non-mucinous lung adenocarcinoma using positron emission tomography and computed tomography.

ABSTRACT:

Background

Accurate identification of the main subtypes of invasive non-mucinous adenocarcinoma (INMA) is essential for individualized treatment. Given the limitations of preoperative biopsy pathological diagnosis, this study aimed to evaluate the feasibility of using positron emission tomography and computed tomography (PET/CT) radiomics features, deep learning (DL) features, and their combined models for non-invasive identification of the four main INMA subtypes prior to surgery.

Methods

A total of 386 patients from hospital one and 32 patients from hospital two with preoperative PET/CT images were enrolled as the training, internal validation cohorts and external validation cohorts, retrospectively. Radiomics features were extracted from CT, PET and PET/CT images to build

SUBMITTER: Wang D 

PROVIDER: S-EPMC12541866 | biostudies-literature | 2025 Sep

REPOSITORIES: biostudies-literature