UCSF

Objective: To evaluate fractional anisotropy (FA) in comparison to apparent diffusion coefficient (ADC) for discriminating between cancer and normal breast tissue using standard and high resolution diffusion tensor imaging (DTI).

Materials and Methods: Dynamic contrast enhanced MRI, standard DTI, and high-resolution DTI data were collected in ten patients with locally advanced breast cancer before the start of neoadjuvant treatment. Regions of interest were selected in tumor and ipsilateral normal tissue. ADC and FA values from both DTI sequences were calculated for tumor and normal tissue regions. Additional studies using an ice water phantom were performed to investigate the effects of off iso-center imaging location on quantitative diffusion measurements.

Results: ADC values computed from both standard and high-resolution DTI showed a significant difference between normal and tumor tissue (p < 0.0001). A statistically significant difference in FA value for normal and tumor tissue using high resolution DTI was also measured (p = 0.02). Standard DTI measurements of FA did not show a statistically significant difference. Estimates of ADC difference between normal and tumor tissue derived from high-resolution DTI were more accurate than those derived from standard DTI as reflected in narrower 95% confidence intervals. Phantom studies showed deviations of up to 16% (standard DTI) and 19% (high-resolution DTI) in ADC values and 44% (standard DTI) and 258% (high-resolution DTI) in FA values measured in left or right breast coil as off-center locations, as compared to measurements made using a head coil at the magnet iso-center.

Conclusions: Both standard and high-resolution DTI sequences found that tumor ADC is significantly lower than ADC of normal tissue. High-resolution DTI also showed that the FA value of tumor is significantly lower than normal tissue. High-resolution DTI sequence might be more sensitive and appears to give superior differentiation between normal tissue and cancer compared to standard DTI.