Abstract:Objective: To explore the clinical value of multimodal ultrasound combined with radiomics in the differential diagnosis of lung squamous cell carcinoma and pulmonary tuberculosis. Methods: A total of 113 patients with lung squamous cell carcinoma and pulmonary tuberculosis confirmed by pathology were retrospectively analyzed, with a total of 121 lesions, including 71 lesions of lung squamous cell carcinoma and 50 lesions of pulmonary tuberculosis. All patients underwent gray scale ultrasound and contrast-enhanced ultrasound, and the results were compared and analyzed. According to the ratio of 7: 3, it was randomly divided into a training set(85 cases)and a validation set(36 cases), and the corresponding radiomics features were extracted from these two types of images, respectively. After standardization, the features were filtered and the dimension was reduced to establish a multimodal ultrasound radiomics label. Logistic regression was used to establish clinical diagnosis, multimodal ultrasound imaging group and combined diagnosis model. The receiver operating characteristic(ROC)curve was used to evaluate the performance of each model, and the area under the curve(AUC)was compared by Delong test. The clinical decision curve was drawn to evaluate its clinical practical value, and the calibration curve was drawn to analyze the consistency of the combined diagnostic model in identifying lung squamous cell carcinoma and pulmonary tuberculosis. Results: Multivariate Logistic regression analysis showed that the lesion shape, size and contrast-enhanced ultrasound enhancement mode were independent risk factors for the differential diagnosis of lung squamous cell carcinoma and pulmonary tuberculosis. The multimodal radiomics model and the joint model were established by using two-dimensional ultrasound and contrast-enhanced ultrasound images. The AUC of the joint model in the training set and the validation set were 0.995 and 0.908, respectively, which were higher than those of the single classification model. Conclusion: The combined model based on multimodal ultrasound radiomics can better identify and diagnose lung squamous cell carcinoma and pulmonary tuberculosis, and has a good application prospect.