A T2 weighted imaging-based radiomics nomogram for the classification of hepatic blood-rich lesions: hepatocellular carcinoma and benign liver lesions.

purposeTo develop a radiomics nomogram based on T2-weighted imaging (T2WI) to distinguish hepatocellular carcinoma (HCC) from blood-rich benign liver lesions (BLLs) and to assess its value as an adjunct to conventional imaging.

methodsThis study analyzed imaging and clinical data from 220 patients with pathologically confirmed HCC (n = 140) and BLLs (n = 80). Patients were divided into training cohort (n = 100), test cohort (n = 44), and validation cohort (n = 76). We developed a radiomics model, a clinical model, and a fusion model integrating Rad-score with clinical factors, and identified the best predictive model among them in the test set, which was then compared with the diagnostic efficiency of two radiologists. The efficacy was assessed by the area under the receiver operating characteristic curve (AUC).

resultsFour radiomics features and two clinical factors (age, liver cirrhosis) were selected for the construction of radiomics and clinical models, respectively. The fusion model exhibited superior discriminative ability in the training set (AUC = 0.984), test set (AUC = 0.948), and validation set (AUC = 0.955). In the validation set, a statistical difference was observed between the fusion model and the radiomics model (P = 0.001). Furthermore, the diagnostic performance of the fusion model based on T2WI was comparable to that of conventional imaging based on enhanced MRI. In the cohort with conventional imaging errors, the AUCs of the clinical model, radiomics model, and fusion model were 0.841, 0.737, and 0.905, respectively.

conclusionsThe T2WI-based radiomics nomogram effectively differentiated HCC from blood-rich BLLs, and might be an adjunct to conventional imaging.