Editorial for “Effect of Phase Encoding Direction on Image Quality in Single‐Shot EPI Diffusion‐Weighted Imaging of the Breast”

Journal of Magnetic Resonance Imaging, vol. 60, pp. 1350–1351

Abstract

Diffusion-weighted imaging (DWI) of the breast is a promising technique to help lesion detection and characterization, prognosis, and the assessment of treatment response.1 However, it comes with technical challenges such as limited spatial resolution and susceptibility to off-resonance distortion artifacts. These factors still impede the clinical application of DWI for breast imaging, and more so, its potential use as a standalone diagnostic tool. Therefore, any technical improvements for breast DWI are welcome. In their current article, Rodriguez-Soto et al put a focus on the choice of phase encoding direction and its effects on image quality of single-shot echo planar imaging (EPI) sequences of the breast.2 The European Society of Breast Imaging recommends EPI sequences for breast DWI.3 Single-shot EPI sequences acquire all k-space data of a slice after one excitation, using rapidly switched readout- and phase-encoding gradients. Consequently, they are characterized by a long signal readout, which explains their rather modest matrix sizes and their susceptibility to off-resonance artifacts. Why is breast imaging particularly challenging? First, axial bilateral breast images require a large field of view to avoid backfolding. Second, the female breast anatomy presents with many air-tissue interfaces, and thus intrinsically high susceptibility variations. These are difficult to shim, especially in bilateral axial breast imaging, which impairs stable fat suppression albeit needed for DWI. Third, the direct vicinity of the heart can lead to pulsation artifacts. Both off-resonance distortions and cardiac pulsation artifacts will, naturally, be affected by the choice of phase (and frequency) encoding direction.4 In the past, different improvements over single-shot EPI have been proposed in the context of breast imaging. One technique to shorten the EPI readout duration beyond parallel imaging is multi-shot EPI,5, 6 enabling higher resolution and less distortion artifacts; yet, at the expense of increased scan time and susceptibility to inter-shot motion. Simultaneous multi-slice imaging can address the scan time requirements of multi-shot EPI,7 while the use of navigators can mitigate inter-shot motion. Reduced-field-of-view EPI may be employed to increase the resolution of breast DWI.8 And, of course, artificial intelligence is on the horizon to further “beautify” diffusion-weighted images of the breast. Deep learning algorithms will certainly open new doors, eg., for imaging with higher b-values or higher resolution while maintaining signal-to-noise ratios.9 Rodriguez-Soto et al took a step backward and investigated the choice of phase encoding direction as a basic influence factor on geometric distortions and image quality in single shot EPI–DWI of the breast at 3 T. They compared the four possible in-plane encoding directions, i.e., left-to-right (LR), right-to-left (RL), anterior-to-posterior (AP), and posterior-to-anterior (PA). Geometric distortions were first quantified in a commercially available breast phantom, which revealed most important distortions for phase encoding along PA (P < 0.05). Then, two breast radiologists stated their preferred phase-encoding direction and assessed the presence of cardiac artifacts in a cohort of 11 healthy volunteers, of which one had implants. Of note, visual assessment was performed based on the b = 0 s/mm2 images. The authors report a clear preference of the radiologists for the PA direction, which may seem surprising in comparison to the phantom results, where PA resulted in highest distortions. This finding could indicate that not only the absolute magnitude of the distortion, but also its direction (stretching vs. compression) and symmetry between both breasts affect the suitability of diffusion-weighted images for radiologic assessment. Interestingly, Zhao et al found reduced deformations for phase encoding along PA as compared to AP in vivo.10 In the present study, no physiological artifacts were observed in any of the volunteer datasets. In addition to the volunteer study, the authors present an example patient case with invasive ductal carcinoma. The b = 800 s/mm2 images showcase a strong impact of the phase encoding direction on the lesion shape and motivate the need for future clinical investigations. Indeed, acquiring data including breast lesions of different size, shape, and location for different breast densities would be a valuable next step. In summary, the application of DWI in the breast faces some well-known technical challenges. Although the selection of phase encoding direction has an impact on image quality, it is—unfortunately—not the only required optimization. Yet, a combination of many promising technical improvements, both from the acquisition and the reconstruction side, will likely move breast DWI further forward.

Authors 1

1. Teresa Lemainque corresponding Aachen Department of Diagnostic and Interventional Radiology Medical Faculty

   RWTH Aachen University

   Affiliation as printed

   Department of Diagnostic and Interventional Radiology, Medical Faculty RWTH Aachen University Aachen Germany

   Department of Diagnostic and Interventional Radiology, Medical Faculty, RWTH Aachen University, Aachen, Germany