Blood oxygenation level dependent magnetic resonance imaging and diffusion weighted MRI imaging for benign and malignant breast cancer discrimination

Blood oxygenation level dependent magnetic resonance imaging and diffusion weighted MRI imaging for benign and malignant breast cancer discrimination:



Publication date: Available online 17 October 2020

Source: Magnetic Resonance Imaging

Author(s): Roberta Fusco, Vincenza Granata, Paolo Pariante, Vincenzo Cerciello, Claudio Siani, Maurizio Di Bonito, Marika Valentino, Mario Sansone, Gerardo Botti, Antonella Petrillo



Abstract

Purpose

The purpose of this study is to assess Blood oxygenation level dependent Magnetic Resonance Imaging (BOLD-MRI) and Diffusion Weighted Magnetic Resonance Imaging (DW-MRI) in the differentiation of benign and malignant breast lesions.



Methods

Fifty-nine breast lesions (26 benign and 33 malignant lesions) pathologically proven in 59 patients were included in this retrospective study. As BOLD parameters were estimated basal signal S0 and the relaxation rate R2*, diffusion and perfusion parameters were derived by DWI (pseudo-diffusion coefficient (Dp), perfusion fraction (fp) and tissue diffusivity (Dt)). Wilcoxon-Mann-Whitney U test and Receiver operating characteristic (ROC) analyses were calculated and area under ROC curve (AUC) was obtained. Moreover, pattern recognition approaches (linear discrimination analysis (LDA), support vector machine, k-nearest neighbours, decision tree) with least absolute shrinkage and selection operator (LASSO) method and leave one out cross validation approach were considered.



Results

A significant discrimination was obtained by the standard deviation value of S0, as BOLD parameter, that reached an AUC of 0.76 with a sensitivity of 65%, a specificity of 85% and an accuracy of 76%. No significant discrimination was obtained considering diffusion and perfusion parameters. Considering LASSO results, the features to use as predictors were all extracted parameters except that the mean value of R2* and the best result was obtained by a LDA that obtained an AUC = 0.83, with a sensitivity of 88%, a specificity of 77% and an accuracy of 83%.



Conclusions

Good performance to discriminate benign and malignant lesions could be obtained using BOLD and DWI derived parameters with a LDA classification approach. However, these findings should be proven on larger and several dataset with different MR scanners.



Keywords

Breast cancerHypoxiaBOLD-MRIDW-MRIIVIMTissue discrimination