Establishment of a predictive model of CT value in the diagnosis and differential diagnosis of pulmonary cryptococcosis

Abstract

ObjectiveTo investigate the application value of computed tomography (CT) value (HU) in the diagnosis and differential diagnosis of pulmonary cryptococcosis (PC) and to construct a prediction model. MethodsRetrospective analysis of the clinical data of 73 patients who presented with nodular/mass-type occupations on lung CT and confirmed by histopathology in our hospital from January 2019 to May 2022 were divided into PC group (23 patients) and non-PC group (50 patients) according to the pathological findings, and the CT values of each patient's lung lesions were measured. The differences in age, gender, symptoms, lesion involvement in one/both lungs, lung lobe distribution, number of lesions, maximum lesion diameter (cm), lesion margin condition, and CT value results were compared between the two groups. Independent risk factors for PC were analyzed for indicators with statistically significant differences, clinical prediction models were constructed and column line plots were drawn, C (correction) indices were calculated, subject characteristics (ROC) curves were drawn, calibration curves and clinical decision curve analysis (DCA) were performed to further evaluate the predictive efficacy of the models. ResultsComparative analysis of patient data between the two groups showed statistically significant differences in central, peripheral and global CT values (P < 0.05), and multiple regression analysis indicated that central CT value, peripheral CT value and global CT value could be used as independent risk factors for the diagnosis and differential diagnosis of PC. The area under the ROC curve of the model predicting PC was 0.814 (95 % CI: 0.7011–0.9267), and the corrected C-index (Bootstrap = 1000) was 0.781; the actual curve overlapped well with the calibration curve; the DCA results indicated that the column line graph model has high clinical application value. ConclusionsCT value measurements of lesions can be used as an independent risk factor for PC, and clinical prediction models based on the above factors are predictive for the diagnosis and differential diagnosis of PC.