As clinical breast MR increasingly moves from 1.5T to 3T, accurate knowledge of T1 and T2 relaxation times of breast tissue is important for accurate tissue characterization and MR sequence parameter optimization for both clinical and research applications. Sources reporting T1 and T2 relaxation times at 3T are limited¹. In addition, as breast tissue consists of a highly heterogeneous distribution of fat and fibroglandular tissue, partial volume effects are common and make it difficult to obtain independent relaxation rates for the two tissues. In this work, a previously developed, rapid, stimulated echo acquisition mode (STEAM)-based MR spectroscopy (MRS) technique² is applied to measure T1 and T2 relaxation times in breast fat and fibroglandular tissue in normal volunteers.

Six normal volunteers were recruited for this HIPAA compliant, IRB approved study. Volunteers underwent 3T breast MRI (Discovery 750w, GE Healthcare, Waukesha WI) using an 8-channel phased array breast coil (GE Healthcare, Waukesha WI).

The rapid multi-TR, multi-TE STEAM-MRS sequence developed by Hamilton et al.² provides simultaneous, accurate estimation of proton-density fat-fraction (PDFF), and T1 and T2 relaxation times of water and fat tissue within the acquisition voxel. MRS voxels were placed in each volunteer, one in primarily fatty tissue and the other in primarily fibroglandular tissue as shown in Figure 1. Voxel sizes were adjusted manually for each volunteer to keep the voxel primarily in one tissue type and ranged from approximately 10 x 10 x 10 mm to 20 x 20 x 20 mm. Thirty-two spectra with varying TRs (150-2000ms) and varying TEs (10ms-110ms) as described in Hamilton et al. were acquired in a 21 second acquisition for each voxel. Spectroscopic quantification was performed offline using an automated algorithm for joint quantification of all acquired spectra for a given voxel, including measurement of the T1 and T2 of fibroglandular tissue and fat, as well as voxel PDFF, based on the method described in Hernando et al.³

The measured T1 and T2 relaxation times are given in Table 1. Fibroglandular results from one volunteer were excluded as the breast was primarily fatty and the voxel contained less than 15% fibroglandular tissue, based on the estimated PDFF value.

A previous study¹ by Rakow-Penner et al based on inversion recovery imaging and spin echo imaging (with and without fat-water separation) had reported breast T1 and T2 values: T1_{fibroglandular}= 1445±93ms, T2_{fibroglandular}= 54±9ms, T1_fat= 367±8ms, and T2_fat= 53±2ms. These previous measurements are discrepant with our measurements. Specifically, Rakow-Penner et al reported that T2_fat ≈ T2_{fibroglandular} whereas we observed T2_fat > T2_{fibroglandular}. The source of this discrepancy may be related to technical differences in the relaxometry methods such as J-coupling effects or fat-water separation errors due to the lack of multi-peak spectral modeling of fat⁴, but currently remains unclear.

All voxels placed in the breast demonstrated the presence of both fat and water species contributing to the overall signal despite efforts to place voxels in areas appearing to consist of a single tissue type. This highlights the fact that fat and fibroglandular tissue are highly intermingled in the breast, making partial volume effects nearly impossible to avoid. Rakow-Penner et al.¹ utilized IDEAL with a single peak model of fat to provide fat-water separation, which can lead to inaccurate water-fat separation⁴ and ultimately the calculated T1 and T2 values. Spectroscopy is generally considered to be the most accurate noninvasive method to quantify the PDFF. The spectroscopic STEAM sequence used here allows for accurate separation of the fat and water components, enabling truly independent determination of T1 and T2 values of fat and fibroglandular tissue even in the presence of partial volume effects. STEAM also better mitigates the effects of J-coupling than PRESS spectroscopy methods⁵, allowing for more accurate measurement of T2 relaxation time. Further, the short, ~20 second scan time of the sequence allows it to easily be incorporated into clinical or research exams.

In this work, we used a rapid spectroscopic STEAM sequence for measurement of T1 and T2 of both fat and fibroglandular breast tissue at 3T. The measurements provided by this approach may enable improved tissue characterization and protocol optimization for a variety of breast imaging applications.