Correlation Analysis of Breast Cancer DWI Combined with DCE-MRI Imaging Features with Molecular Subtypes and Prognostic Factors

• Published in: Journal of medical systems Vol. 43; no. 4; pp. 83 - 10
• Main Authors: Yuan, Congru, Jin, Feng, Guo, Xiuling, Zhao, Sheng, Li, Wei, Guo, Haidong
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.04.2019; Springer Nature B.V
• Subjects: Artificial Intelligence Application in Health Informatics; Bayesian analysis; Breast cancer; Cancer; Correlation analysis; Discriminant analysis; ErbB-2 protein; Gene amplification; Health Informatics; Health Sciences; Image & Signal Processing; Image enhancement; Information processing; Lymph nodes; Magnetic resonance imaging; Medical imaging; Medical prognosis; Medicine; Medicine & Public Health; Metastases; Parameters; Physical characteristics; Post-production processing; Signal processing; Statistics for Life Sciences; Workstations; Diffusion-weighted imaging; Prognostic factor; Breast cancer; DCE-MRI; Molecular subtype
• ISSN: 0148-5598; 1573-689X; 1573-689X
• DOI: 10.1007/s10916-019-1197-5

This study aimed to deeply analyze the application of DWI and DCE-MRI imaging in breast cancer, the correlation between the imaging characteristics of DWI and DCE-MRI and the molecular subtypes and prognostic factors of breast cancer was studied. Firstly, DWI and DCE-MRI scans of all patients before interventional therapy were performed, and relevant information of the subjects was introduced in turn. Secondly, molecular subtypes were determined according to immunohistochemical results and gene amplification. Siemens 3.0 T post-processing workstation was used for image post-processing. The time signal curve (TIC), early enhancement rate (EER) and ADC values were measured, morphological characteristics were recorded, and the correlation between each image feature and molecular subtypes, prognostic factors (tumor size, pathological grade, lymph node metastasis, ER, PR, HER2, Ki67) was analyzed. The results showed that parameters such as ADC value, EER, lobulation sign, burr sign, enhancement way and TIC type were correlated with prognostic factors and molecular subtypes. And Bayesian model discriminant analysis showed that the above imaging parameters couldn’t well predict the expression of immunohistochemical factors and molecular subtypes. However, the above characteristics had a good effect on the prediction of pathological grade, with a false diagnosis rate of 9.69%.