Use of CT radiomics to differentiate minimally invasive adenocarcinomas and invasive adenocarcinomas presenting as pure ground-glass nodules larger than 10 mm

• DOI: 10.1016/j.ejrad.2021.109772
• Publication Date: 2021-05-12T15:55:17Z
• Authors: Ziqi Xiong; Yining Jiang; Siyu Che; Wenjing Zhao; Yan Guo; Guosheng Li; Ailian Liu; Zhiyong Li
• Keywords: Univariate; Univariate analysis; Lasso

PurposeThis study aimed to develop a model based on radiomics features extracted from computed tomography (CT) images effectively differentiate between minimally invasive adenocarcinomas (MIAs) and (IAs) manifesting as pure ground-glass nodules (pGGNs) larger than 10 mm.MethodThis retrospective included patients who underwent surgical resection for persistent pGGN November 2012 June 2018 diagnosed with MIAs or IAs. The were randomly assigned the training test cohorts. correlation coefficient method least absolute shrinkage selection operator (LASSO) applied select useful constructing whose performance was assessed by area under receiver operating characteristic curve (AUC-ROC). compared standard CT (shape, volume mean value of largest cross-section) combined radiomics-standard using univariate multivariate logistic regression analysis.ResultsThe showed better discriminative ability (training AUC, 0.879; 0.877) 0.820; 0.828). 0.870) did not demonstrate improved model. Radiomics_score an independent predictor invasiveness following analysis.ConclusionsFor pGGNs mm, demonstrated superior diagnostic in differentiating IAs MIAs, which may be clinicians diagnosis treatment selection.