Comparing average breast fat content results from two different protocols at 1.5T and 3T: can the data be pooled?

PURPOSE

To compare the total breast fat content computed from two separate studies, performed on different scanners and with different protocols, with the goal of defining a relationship to allow pooling the data.

MATERIALS AND METHODS

Twelve healthy volunteer women were scanned with two different protocols on the same day. The protocols differed in four important aspects: vendors (GE vs. Philips), scanner main magnetic field strengths (1.5T vs. 3T), pulse sequences (2D fast spin-echo vs. 3D spoiled gradient-echo), and water/fat separation techniques. The resulting water and fat maps were processed with in-house software to extract breast tissue slice-wise. Percent fat content was calculated for each breast, per subject.

RESULTS

Total percent fat contents (averaged across both breasts) resulting from both protocols were plotted against each other, on a subject-by-subject basis, revealing a strong correlation (R(2)  > 0.99), with an overestimation of the fat content from Protocol 1 relative to Protocol 2. The proposed T2 TE-correction for Protocol 1 improves the correlation while decreasing the discrepancy between protocols.

CONCLUSION

Total breast fat content of healthy women resulting from the two protocols can be pooled using a linear relationship. The proposed T2 TE-corrected Protocol 1 is expected to yield accurate fat content.