The diagnostic value of a nomogram based on enhanced CT radiomics for differentiating between intrahepatic cholangiocarcinoma and early hepatic abscess

Objective This study aimed to investigate the value of a CT-enhanced scanning radiomics nomogram in distinguishing between early hepatic abscess (EHA) and intrahepatic cholangiocarcinoma (ICC) and to validate its diagnostic efficacy. Materials and Methods Clinical and imaging data on 112 patients diagnosed with EHA and ICC who underwent double-phase CT-enhanced scanning at our hospital were collected. The contours of the lesions were delineated layer by layer across the three phases of CT scanning and enhancement using 3D Slicer software to define the region of interest (ROI). Subsequently, the contours were merged into 3D models, and radiomics features were extracted using the Radiomics plug-in. The data were randomly divided into training (n = 78) and validation (n = 34) cohorts at a 7:3 ratio, using the R programming language. Standardization was performed using the Z-score method, and LASSO regression was used to select the best λ-value for screening variables, which were then used to establish prediction models. The rad-score was calculated using the best radiomics model, and a joint model was constructed based on the rad-score and clinical scores. A nomogram was developed based on the joint model. The diagnostic efficacy of the models for distinguishing ICC and EHA was assessed using receiver operating characteristic (ROC) curve and area under the curve (AUC) analyses. Calibration curves were used to evaluate the reliability and accuracy of the nomograms, while decision curves and clinical impact curves were utilized to assess their clinical value. Results Compared with the ICC group, significant differences were observed in clinical data and imaging characteristics in the EHA group, including age, centripetal enhancement, hepatic pericardial depression sign, arterial perfusion abnormality, arterial CT value, and arteriovenous enhancement ( p < 0.05). Logistic regression analysis identified centripetal enhancement, hepatic pericardial depression sign, arterial perfusion abnormality, arterial CT value, and arteriovenous enhancement as independent influencing factors. Three, five, and four radiomics features were retained in the scanning, arterial, and venous phases, respectively. Single-phase models were constructed, with the radiomics model from the arterial phase demonstrating the best diagnostic efficacy. The rad-score was calculated using the arterial-phase radiomics model, and nomograms were drawn in conjunction with the clinical model. The nomogram based on the combined model exhibited the highest differential diagnostic efficacy between EHA and ICC (training cohort: AUC of 0.972; validation cohort: AUC of 0.868). The calibration curves indicated good agreement between the predicted and pathological results, while decision curves and clinical impact curves demonstrated higher clinical utility of the nomograms. Conclusion The CT-enhanced scanning radiomics nomogram demonstrates high clinical value in distinguishing between EHA and ICC, thereby enhancing the accuracy of preoperative diagnosis.