Authored date:2006-06-02

Introduction

Contrast enhanced* T1 3D-imaging is used routinely in our department in the pre-surgical planning of breast cancer. Covering the whole breast with the 3D volume and following its gadolinium uptake, gives a very sensitive imaging method for detection of lesions. However, in order to improve its specificity, both high spatial and temporal resolution are needed. Several tools have been used to obtain this. We routinely use a rectangular FOV (recFOV), which requires fewer phaseencoding steps to obtain the same resolution and hence reduce measurement time. A major drawback of recFOV in breast MRI is, however, that we need to acquire images in the coronal plane. In the transverse plane the phase encoding needs to be in the left-right direction to prevent heart pulsation artifacts from passing through the breast and as a consequence, no reduction of the FOV can be used.
When using the coronal orientation, we end up with an imaging sequence of the breast with a temporal resolution of
1:26 min, using 64 slices of 2 mm, resulting in a pixel size of 1.5 x 0.8 x 2 mm³ which has successfully been used for years (TE = 4.8 ms, in phase condition at 1.5T). It leaves us with high qualityMulti Planar Reconstructions (MPR) or Maximum Intensity Projections (MIP) of subtracted data, to look at morphology, location and extension of the lesion. Time behavior gives information on the vascularization of the lesion, depicted on the wash-in* and wash-out* maps which can be calculated automatically by the system (e.g. Inline Technology).
All the important criteria of the BIRADS scoring system can be evaluated in this way: Morphologic information on form (round – oval – linear – branching), margins (well defined – distinct) and enhancement (homogeneous – inhomogeneous – septated – ring enhancement). In addition, dynamic information concerning the initial (<50%, >100%) and late time behavior (continuous enhancement – step – wash-out) is available. But can we do better than this? Yes, we can.

Fig. 1  T1_fl3d pre- and 3 minutes post contrast in coronal direction using rec- FOV 50%, subtraction and thin MIP of an infiltrating ductal carcinoma.

How iPAT can improve visualization of the entire breast in the axial plane?

Since the introduction of integrated Parallel Acquisition Techniques (iPAT), we have used GRAPPA in order to improve our results. A prerequisite for using GRAPPA is the presence of phased array coils which is fulfilled with the commercially delivered bilateral breast coil. The improvement we see is the possibility to return to the axial orientation, which is similar to the cranio-caudal view of X-ray mammography. The axial plane is better suited than the coronal plane to establish the correlation between the region of enhancement and the conventional mammographic abnormalities found. The axial plane is also a good anatomic plane to explore the breast and the thoracic wall. The trajectory, inflammation, obstruction of the galactic ducts, and lesions in the sub-areolar region (Fig. 2) and the pectoral musculature is much better delineated and evaluated in the axial (and sagittal) plane (Fig. 3). An additional advantage is that we can acquire images extending sufficiently deep into the axillary region to depict the lymph nodes if sufficient RF coverage is present. Advantages of the axial over the sagittal orientation are the possibility to reduce the number of partitions, the capability to compare the left and right breast (as both are present in the image), and the certainty that the identification of left or right breast is correct, which is not straightforward in sagittal datasets. Using rectangular FOV in the coronal plane is now substituted by GRAPPA (iPAT2) imaging in the transverse plane. We obtain a resolution of typically 1.0 x 1.1 x 1.5 mm³, TE = 4.8 ms (in phase), giving us MPR and MIP reconstructions which reveal great detail. Temporal resolution with these parameter settings is kept just under 1 min for 88 partitions (Fig. 4).

Fig. 2  Intraductal, retro-arealar papilloma in a woman with nipple discharge: T2_tirm_trans, T1_fl3D_trans pre- and post-gadolinium and subtraction.

Fig. 3  Local recurrence of carcinoma with pectoral muscle invasion: T1_fl3D_trans pre- and post-gadolinium, subtraction and thin MIP.

Fig. 4A  Invasive ductal carcinoma: T1_fl3D_trans pre- and post- gadolinium, subtraction and thin MIP.

Fig. 4B  Mean Curve evaluation on wash-in and wash-out maps on the lesion and the fat.

Comparison of the measured dataset and the MIPs of the subtraction dataset of the same patient imaged in 2003 before surgery, not using iPAT and working in the coronal plane and on follow up after surgery (2004), imaged in the transverse plane using iPAT. Note the improved visualization of the pectoral and axillary region.

How iPAT can improve spatial resolution in the 3 orientations?

Suppression of fat can improve the in-plane resolution of these data. Fat suppression or water excitation pulses give us the chance of using a smaller TE. As the fat signal is no longer present and cannot oppose the enhanced water signal anymore, we can use the out of phase condition with TE< 4.8 ms and we can shorten TR, which shortens overall acquisition time. This is what is being used in VIEWS (Volume Imaging with Enhanced Water Signal) to obtain high isotropic resolution (160 partitions, spatial resolutions of 0.9 x 0.8 x 0.9 mm³, acquisition time of 4 min) (Fig. 5). When using iPAT, this high resolution imaging can be used for dynamic scanning every minute (0.9 x 0.6 x 1.2 mm³) (Fig. 6C, D). A drawback of VIEWS is the absence of fat signal. We are convinced that the presence of fat signal in non fat-suppressed images gives us very valuable information on morphology (intra-pectoral layers and lymph nodes). An additional disadvantage is that sequences acquired with water excitation pulses result in hyper-intense glandular tissue on pre-gadolinium images. The result is less inherent contrast with the post-gadolinium early enhancing lesions.

Fig. 5A/B  Pre- and 3 min post-gadolinium scan with a resolution of 0.9 x 0.8 x 0.9 mm³.

Fig. 5C/D  This high isotropic resolution is ideal for MIP and/or MPR reconstructions.

Fig. 6  Dynamic 1 minute scans with and without fat suppression. (Fig. A, B) Show the fl3d_trans, using iPAT, 0.9 x 0.9 x 2.0 mm³ resolution, 64 slices in 1:19 min. Fig. C, D: Show the VIEWS sequence, using iPAT, 0.9 x 0.6 x 1.2 mm³ resolution, 88 slices in 1 min. We compared the pregadolinium and the late enhanced 6 minute scan on the same patient. Note the hyper-intense glandular tissue in the fatsat scans (disadvantage of this technique).

How can iPAT improve the temporal resolution?

iPAT can help us improve the temporal resolution without losing sensitivity of the MR exam, as we are still able to cover both breasts simultaneously with an acceptable spatial resolution. When using GRAPPA in combination with a rectangular FOV in the coronal plane, we are able to image the entire breast in a timeframe of 5 sec. We obtain a voxel size of 
1.9 x 1.3 x 3.5 mm³ with a temporal resolution of 5 s using 3D VIBE. Early wash-in of lesions is observed 15 to 30 s post-gadolinium injection (Fig. 7). Gadolinium uptake is further followed and documented during the first 2 minutes post injection.
We are aware that by using this technique, we lose all detail in spatial resolution, earlier discussed. That is why we choose to combine both methods. We use a pre-gadolinium high spatial resolution scan, followed by the VIBE high temporal resolution scan during injection. We follow signal enhancement with VIBE for 2 minutes, immediately followed by the high spatial resolution scans at later time points. Important is that we catch the 3 minute signal in high spatial resolution data, in order not to lose the information on morphologic detail. For this purpose, we use a non-fatsat fl3D_trans of
2:46 min with an in plane resolution of 1 mm x 0.8 mm and 64 slices of 2.5 to 3 mm. We measure in the coronal plane to compare with the VIBE images. As a recFOV of 50% is used, no iPAT is applied. The VIBE gives us information on early wash-in, 3 minutes and later scans on the wash-out behavior and morphologic detail is looked at in the high spatial subtraction dataset.
The only disadvantage at this moment is that both datasets cannot be plotted on the same time graph as they have different spatial and temporal resolution. Mean Curve evaluation doesn’t accept the complete measured data in one plot. Results on the eventual improvement of specificity due to the information provided by the contrast enhancement behavior are still under study.

Fig. 7  Dynamic 5 s scans using VIBE. Pre-gadolinium, 15 s post- and 30 s post-gadolinium scan plus thin MIP on the lesion patient. In plane resolution of 1.9 x 1.3 x 3.5 mm³, coronal orientation, use of iPAT2 and recFOV 50%.

Fig. 8  Mean Curve evaluation on the VIBE dataset, following Gadolinium uptake with 25 scans of
5 s, in total 2 min after Gadolinium injection. ROI is placed on the wash-in map.

Fig. 9  Mean Curve evaluation on the high resolution FLASH 3D-scan, 1.0 x 0.8 x 3 mm³ resolution, acquisition time 2:46 min, done at 3 minutes and 6 minutes post Gadolinium. ROI was placed on the subtraction (3 min post-pre) or can be set on the wash-out map. A clear wash-out is detected in this patient at later times. Absolute enhancement of 355 (arbitrary units) reflects a 265% relative enhancement compared to the pre-gadolinium high resolution scan.

Conclusion

Use of iPAT largely improved the quality of our contrast enhanced MR imaging of the breast. Datasets are now combining the axial plane with the high spatial resolution VIEWS technique, revealing high morphologic detail. Combination with the VIBE technique for evaluation of the time behavior of contrast enhancement, with high temporal resolution, and covering the entire breast, is feasible now. This results in high diagnostic sensitivity – thanks to the high resolution – and higher specificity thanks to the dynamic information.

* This information about this product is preliminary. The product is under development and not commercially available in the US, and its future availability cannot be ensured. This article discusses clinical uses which are not commercially available in the US.