Abstract P4-01-10: High-resolution diffusion weighted imaging for the separation of benign from malignant BI-RADS 4/5 lesions found on breast MRI at 3 Tesla

Abstract

Diffusion weighted imaging (DWI) has shown potential for separating malignant from benign lesions on breast MRI, theoretically improving the specificity of the breast MRI exam. However, standard DWI is hampered by multiple factors in the breast which limit utility in clinical practice. A readout segmented diffusion technique (RESOLVE)(1) permits the use of extremely short echo spacing and resamples uncorrectable data using a real-time navigator, reducing image distortion and potentially providing more accurate apparent diffusion coefficient (ADC) within lesions. In order to study the potentially utility of this method against standard diffusion, we compared ADC values from retrospectively-identified BI-RADS 4/5 (suspicious) lesions with both standard single shot-spin echo EPI (ss-EPI) and RESOLVE diffusion at 3 Tesla over a 10-month period. All patients subsequently went to image-directed biopsy. The imaging parameters were as follows: TR/TE=7500-10000/60 ms (ss-EPI) and 8000–12000/64 ms (RESOLVE), averages=8 (ss-EPI) and 1 (RESOLVE), readout segmentation factor 5 with echo spacing of 0.3 ms (RESOLVE), slices = 47–50, b-values 0, 800, resolution = 1.8×1.8×2.4 mm3, imaging times of approximately 5 minutes for both techniques. Freehand ROI's were drawn on each suspicious lesion based on b=800 maps in close reference to the post-contrast T1 series by a board-certified radiologist who was blinded to final pathology and sequence. Similar ROI's were drawn in normal tissue as a control. For each lesion, mean ADC values and signal intensities at b=800 were recorded. ADC values were averaged by technique and pathologic outcome (benign or malignant). Signal intensity was normalized by dividing mean signal intensity of the lesion ROI by that of the control. Significance was determined using Wilcoxon rank-sum test for ADC, and paired t-test for signal intensity measures. Of the final cohort of 38 lesions in 31 patients, 9 were malignant and the remainder were benign. Two lesions were no longer present at breast MRI biopsy, and hence deemed benign. The lesion-to-background signal intensity on diffusion was higher (p = 0.05) on RESOLVE (1.9±0.1) compared to standard diffusion (1.7±0.1). Statistically significant differences between benign and malignant lesion were observed for mean ADC obtained by both methods (Table 1; p < 0.001). Between sequences, there was excellent agreement between RESOLVE and standard EPI for control values obtained from normal tissue, and for mean ADC values of benign lesions. Among malignant lesions, however, there was a statistically significant decrease in mean ADC values measured by RESOLVE (p = 0.05), which effectively widened the separation between benign and malignant lesions. Our results suggest improved separation of benign from malignant lesions by RESOLVE compared to standard diffusion, as well as increased lesion-to-background contrast, suggesting that this diffusion method has particular promise as an adjunct to dynamic-contrast-enhanced breast MRI. (1) Porter DA, et al MRM 2009; 62:468–75. Citation Information: Cancer Res 2012;72(24 Suppl):Abstract nr P4-01-10.