The value of synthetic MRI combined with diffusion weighted imaging in differential diagnosis of benign and malignant breast lesions / 中华放射学杂志

ABSTRACT

Objective:To evaluate the value of synthetic MRI combined with DWI in the diagnosis of benign and malignant breast lesions.Methods:The data of 184 consecutive patients with suspected breast lesions in Yunnan Cancer Hospital from July to September 2019 were prospectively analyzed. All patients were randomly assigned to training group ( n=110) and validation group ( n=74), and underwent conventional MRI and synthetic MRI respectively before and after contrast injection. At the maximum slice of the lesion, the ROI was drawn along the edge and recorded as "tumor". In the solid area with the most obvious tumor enhancement, the second ROI was drawn and recorded as "local". At the same time, ADC values (ADC local and ADC tumor) and relaxation time values (T local and T tumor) were measured. T and T + represented the relaxation time value of the ROI pre-and post-contrast scanning. &Delta;T% represented the relative change rate in T value between pre-and post-contrast scanning.The rank sum test was used to test the quantitative parameters of benign and malignant breast lesions in the training group and the validation group, and the variables with P<0.05 were included in the binary logistic regression analysis to screen the independent variables and establish the prediction model. The area under ROC curve was used to evaluate the discrimination of parameters and models. The clinical applicability of model was analyzed by decision curve analysis (DCA). Results:In the training group, univariate analysis showed that there were significant differences in T 1tumor, T 1+tumor, &Delta;T 1% tumor, T 2local, T 2+local, T 2tumor and T 2+tumor, ADC local, ADC tumor between benign and malignant breast lesions ( P<0.05). Multivariate logistic regression analysis showed that T 1+tumor, &Delta;T 1% tumor, T 2tumor, ADC local, ADC tumor were independent variables in the diagnosis of breast cancer. The relaxation time model (model A T 1+tumor, &Delta;T 1% tumor, T 2tumor) and ADC model (model B ADC local, ADC tumor) established by combining the above variables had the same diagnostic efficiency (AUC=0.905, 0.914, Z=-1.874, P=0.062), and the multi-parameter combination model (model C T 1+tumor, &Delta;T 1% tumor, T 2tumor, ADC local, ADC tumor) had the highest diagnostic efficiency (AUC=0.965). DCA analysis showed that when the threshold probability ranges between 21%-99% (training cohort) and 15%-99% (validation cohort), the net benefit of model C was better than model A and B. Conclusion:The multi-parameter combined prediction model established based on the relaxation time value and ADC can identify breast cancer efficiently and can be used as an auxiliary diagnostic tool.