This poster is published under an
open license. Please read the
disclaimer for further details.
Keywords:
Multidisciplinary cancer care, Image registration, Imaging sequences, MR, Mammography, MR physics, Breast
Authors:
E. Cimino1, M. Marcon2, L. Tomkova1, M. Zanotel1, C. Zuiani1, M. Bazzocchi1; 1Udine/IT, 2Zurich/CH
DOI:
10.1594/ecr2016/C-1464
Methods and materials
In this retrospective study a total of 111 women (mean age 54.3 years; range 37-82 years) who underwent digital breast mammography and contrast-enhanced breast MRI in our institution were included.
Three radiologists in consensus (>4 years experience in breast imaging) scored MBD and BPEql according to the ACR classification (Fig.
1).
BPEql was evaluated by reviewing the second post-contrast images (90 s after contrast injection) using the subtracted images.
In order to perform the MRBD measurements,
free-hand regions-of-interest (ROIs) were drawn along the edges of both breast on every second slice of the axial water signal-only images and copied to the corresponding fat signal-only images of the same Dixon series (Fig.
2).
The software automatically provides mean signal intensity values of all pixels within the breast volume on both water (Iw) and fat signal-only (If) images.
The MRBD was calculated as follows:
MRBD[%] = 100*Iw/(Iw+If).
Mean signal intensity of pixels within the breast volume was measured on the images from the second post-contrast series to obtain the BPEqn.
The one-way analysis of variance was performed to identify significant mean MRBD and BPEqn value differences across the corresponding qualitative categories and each Bonferroni corrected p<0.05 was considered significant.
Statistical correlations between MBD and BPEql and between MRBD and BPEqn were calculated using the Spearman and Pearson coefficient respectively.
MRBD values of low (category A and B) versus high (C and D) breast density groups were compared using MRBD values using the receiver operating characteristic (ROC) curve.