Institutional review board approval was obtained for this prospective, monocentric study. All study procedures were conducted in accordance with the ethical standards as laid down in the 1964 Declaration of Helsinki and its later amendments.
Acceleration strategies for DL accelerated MRI
With DL reconstruction providing higher signal-to-noise ratio (SNR) and allowing for higher acceleration with conventional sampling patterns, the data acquisition can be tailored.
For TSE acquisitions modifying the original acquisition’s image contrast is usually not a desirable goal for tailored acquisitions. Rather, DL reconstruction in MRI can be used to improve on a combination of image resolution, acquisition time, and SNR while maintaining the original contrast.
In contrast to that, some applications like the Half-Fourier single-shot turbo spin echo (HASTE) sequence provide fast and robust acquisitions at the cost of image contrast as compared to the TSE sequence. Here, the improved DL reconstruction enables acquisition protocols providing improved image contrast. Specifically, with a higher acceleration factor the duration of the echo train can be shortened, and therefore, the effect of T2 decay can be reduced. As an additional benefit, the specific absorption rate (SAR) is reduced with the number of required refocusing pulses. This allows for further sequence optimizations in the form of larger gaps between consecutively acquired slices and reduced repetition times.
Besides the data acquisition for the actual image data, calibration data for the coil-sensitivity estimation needs to be acquired. For the TSE sequence, these data which cover the center of k-space (typically about 16 phase-encoding lines) are acquired as part of the imaging scan.
A conventional under-sampling pattern as known from parallel imaging is used. As shown in earlier works (8, 9), these provide the same performance when reconstructed with DL-based methods as incoherent sampling patterns favored by CS. They have the important advantages of being clinically established and are highly flexible regarding adaptations of resolution and signal evolution during sampling. Furthermore, their artifact behavior with regard to motion, reduced field-of-view, and aliasing is well-known and potentially even improved by the DL-based reconstruction. Also, as a part of conventional under-sampling patterns, a fraction of the k-space’s periphery is often not acquired in order to reduce acquisition time. This effectively reduces the resolution in the phase-encoding direction and is referred to as phase resolution. It describes the fraction of acquired data in the phase-encoding direction in percent neglecting the regular parallel imaging type of under-sampling. For illustration, an exemplary sampling pattern of acceleration factor 2 and a phase resolution < 100% is shown in Figure 1. A calibration region around the k-space center is fully sampled and used for the estimation of coil-sensitivity maps.
Deep Learning Image Reconstruction
For all discussed applications, the prototype image reconstruction comprises a fixed iterative reconstruction scheme or variational network (9, 10). The fixed unrolled algorithm for accelerated MR image reconstruction consists of multiple cascades, each made up from a data consistency using a trainable Nesterov Momentum followed by a CNN-based regularization. The regularization model's architecture is based on a novel hierarchical design of an iterative network that repeatedly decreases and increases the resolution of the feature maps, allowing for a more memory-efficient model than conventional CNNs. In addition to the input under-sampled k-space data, coil-sensitivity maps are also provided, which are estimated from the calibration data as a pre-processing step. Also, a bias-field is extracted from a separate adjustment acquisition for image homogenization. The architecture of the reconstruction network is illustrated in Figure 2. During the training phase, the bias-field is inserted into the image reconstruction as a final correction step.
For the image reconstruction k-space data, bias-field correction and coil-sensitivity maps are inserted into the variational network. Compared to previous works and the previously described cascades, the variational network also utilizes two additional types of cascades, namely, pre- and post-cascades. Like regular cascades, pre-cascades employ trainable extrapolation; however, no regularization is applied, allowing the network to focus on parallel imaging. Such design is motivated by the empirical finding that initial steps in the variational network focus on the signal recovery of missing data near the k-space center. This approach supports acquisitions without integrated calibration and flexible k-space sampling. Finally, post-cascades employing non-trainable extrapolation are also utilized for further guarantees on the data consistency, which minimizes the risk of hallucination when adversarial training is applied. The network is first trained to minimize the combined L1 and a multi-scale version of the structural similarity (SSIM) content losses between network prediction and ground truth images. A semi-supervised refinement is applied in a subsequent training step where an adversarial loss based on Wasserstein Generative Adversarial Networks (WGAN) is also added (11).
The reconstruction was trained on volunteer acquisitions using conventional TSE protocols. About 10,000 slices were acquired on volunteers using clinical 1.5T and 3T scanners (MAGNETOM scanners, Siemens Healthcare, Erlangen, Germany). Fully sampled acquisitions with high resolution were performed in head, pelvis, and knee using representative contrasts for the respective body regions. The training data therefore included a wide range of image contrasts, orientations, body regions, and resolutions. For both sequence types, the input to the reconstruction network was retrospectively down-sampled to an acceleration factor of 4. The training was implemented in PyTorch and performed on a GPU cluster NVIDIA Tesla V100 (32GB of memory) GPU.
Implementation of DL image reconstruction in clinical workflow
For deployment in the scanner reconstruction pipeline, the obtained network was converted to a C++ implemented inference framework. For the CPU-only reconstruction on a clinical MRI scanner, inference needed about 2 seconds per slice for the used protocol settings. The reconstruction was triggered after the end of the acquisition, which resulted in a perceived reconstruction time of 2-3 minutes including additional pre- and post-processing. GPU-based reconstruction brings the duration down to the order of 10 seconds for a complete dataset but was not available in the local setting of this study.
Diagnostic image evaluation
Accelerated TSEDL sequences were prospectively acquired along with standard TSE sequences at clinical 3T MRI scanner (MAGNETOM, Siemens Healthcare, Erlangen, Germany), and a final sample of 30 examinations in 15 volunteers were included in this analysis, see Table 1.
TABLE 1. Demographics of included individuals
|
Variables
|
|
Total (male/female), n
|
30 (20/10)
|
|
knee: 10 (7/3)
|
|
shoulder: 10 (6/4)
|
|
lumbar spine: 10 (7/3)
|
Age, mean ± SD (range), y
|
total: 28 ± 7 (20 - 54)
|
|
knee: 27 ± 4 (20 - 31)
|
|
shoulder: 30 ± 10 (20 – 54)
lumbar spine: 28+7 (20 – 43)
|
SD indicates standard deviation; y, years; n, number.
|
Two radiologists with 9 years and 3 years of experience in MRI independently rated both TSE and TSEDL by using a random order.
Overall image quality, artifacts, noise, sharpness, subjective SNR, as well as diagnostic confidence and image impression ratings were performed on an ordinal 4-point Likert scale (1 = non-diagnostic, with major streak artifacts; 2 = non-diagnostic, moderate artifacts with low image quality; 3 = minor artifacts with good image quality; 4 = no artifacts with excellent image quality; image impression: 1 = very unrealistic, 2 = unrealistic, 3 = realistic, and 4 = very realistic). Reading scores were considered sufficient when reaching ≥ 3.
Image analysis was performed on a PACS workstation (GE Healthcare Centricity™ PACS RA1000, Milwaukee, WI, USA).
Statistical analyses were performed using SPSS version 26 (IBM Corp, Armonk, NY, USA). Besides descriptive statistics, compromising median and interquartile range (IQR), the reading score of the qualitative image analysis of the TSE sequences were compared using a paired Wilcoxon signed-rank test. Significance was assumed at a level of P < 0.05.
Inter-rater agreement was calculated through Cohen’s kappa. Kappa values were interpreted as follows: 0–0.20 = poor agreement, 0.21–0.40 = fair agreement, 0.41–0.60 = moderate agreement, 0.61–0.80 = substantial agreement, 0.81–1 = (almost) perfect agreement.
Technologists’ assessment
Five MRI technologists were included in this survey. On a 4-point Likert scale, the required time for reconstruction process (1 = long reconstruction time, 2 = moderate, 3 = low, 4 = very low reconstruction time), the required time and effort for implementation in workflow (1 = high, 2 = moderate, 3 = low, 4 = very low), as well as the technical stability (1 = low technical stability, 2 = moderate technical stability, 3 = good technical stability, 4 = excellent technical stability), and acceptance of DL sequences in clinical workflow (1 = low acceptance, 2 = moderate acceptance, 3 = good acceptance 4 = high acceptance) was evaluated.