Standardisation and acceleration of quantitative analysis of dynamic MR mammographies via parametric images and automatised ROI definition

Purpose: To evaluate quantitative analysis of dynamic breast MRI studies using a dedicated PC based diagnosis system (DS) providing parametric images and automatic ROI definition as compared to the standard subtraction image, manual-ROI based procedure. Methods: We compared the diagnostic usefulness of parametric versus subtraction images in terms of visualisation of enhancement inhomogeneities and enhancement velocities. Quantitative analysis of enhancing lesions of 15 breast MRI studies was performed using both the DS and the system's console software (SC). We assessed the time needed for complete quantitative analysis and number of lesions evaluated. This was followed by assessment of within-reader and between-reader variability or within-case reproducibility of results of quantitative analysis. Results: Parametric images are superior to subtraction images in visualizing enhancement inhomogeneities or ring enhancement in breast cancers. Mean time needed for analysis at DS and SC was 4 (3-5) min. and 23 (8-39) min., respectively. During this time period, significantly more lesions were evaluated per case using the DS as compared to the SC (2-14 vs. 1-6). Mean within-reader variability of enhancement velocities of the same lesions was 0% and 25% (DS and SC); between-reader variability of enhancement values obtained in the same lesions was 11% (DS) versus 41% (SC). Conclusion: The DS significantly cuts down the time needed for quantitative analysis. It significantly improves reproducibility of quantitative enhancement values due to standardization of ROI analysis.