Evaluation of the kinetic properties of background parenchymal enhancement throughout the phases of the menstrual cycle

Abstract

Purpose: To develop and apply a semiautomatic method of segmenting fibroglandular tissue to quantify magnetic resonance (MR) imaging contrast material-enhancement kinetics of breast background parenchyma (BP) and lesions throughout the phases of the menstrual cycle in women with benign and malignant lesions.

Materials and methods: The institutional review board approved this retrospective HIPAA-compliant study, and informed consent was waived. From December 2008 to August 2011, 58 premenopausal women who had undergone contrast material-enhanced MR imaging and MR imaging-guided biopsy were identified. The longest time from the start of the last known period was 34 days. One lesion per patient (37 benign and 21 malignant) was analyzed. The patient groups were stratified according to the week of the menstrual cycle when MR imaging was performed. A method based on principal component analysis (PCA) was applied for quantitative analysis of signal enhancement in the BP and lesions by using the percentage of slope and percentage of enhancement. Linear regression and the Mann-Whitney U test were used to assess the association between the kinetic parameters and the week of the menstrual cycle.

Results: In the women with benign lesions, percentages of slope and enhancement for both BP and lesions during week 2 were significantly (P < .05) lower than those in week 4. Percentage of enhancement in the lesion in week 2 was lower than that in week 3 (P < .05). The MR images of women with malignant lesions showed no significant difference between the weeks for any of the parameters. There was a strong positive correlation between lesion and BP percentage of slope (r = 0.72) and between lesion and BP percentage of enhancement (r = 0.67) in the benign group. There was also a significant (P = .03) difference in lesion percentage of slope between the benign and malignant groups at week 2.

Conclusion: The PCA-based method can quantify contrast enhancement kinetics of BP semiautomatically, and kinetics of BP and lesions vary according to the week of the menstrual cycle in benign but not in malignant lesions.

Figure 1:

Representative example of PCA-based analysis of DCE MR imaging data. A, Successive fat-suppressed sagittal T1-weighted MR images with contrast material injection during second image. Signal intensity range was kept constant to show contrast enhancement. B, Graphs show principal components obtained from PCA analysis. Plotted are eigenvectors (EV) scaled so that the sum of their squares is equal to the corresponding eigenvalues. C, Projection coefficient (Proj. Coeff.) maps of principal components in descending order of eigenvalues. Primary principal component (ie, with largest eigenvalue) usually captures the major portion of signal variance from the most enhancing tissue. Subsequent principal components correspond essentially to noise.

Figure 2:

Box and whisker plots of kinetic parameters grouped by menstrual cycle week for A–D, benign lesion (LS) group, and E–H, malignant lesion group. Lines inside boxes indicate median values, and boxes represent 25th and 75th percentile values. Whiskers show minimum and maximum values. P values are from Mann Whitney U tests and are shown if P is less than .05.

Figure 3:

Scatterplots of lesion (LS) and BP kinetic parameters for A, B, benign lesion group and, C, D, malignant lesion group. Dotted lines = linear regression lines.