Numerical and Clinical Evaluation of the Robustness of Open-source Networks for Parallel MR Imaging Reconstruction.

PURPOSE: Deep neural networks (DNNs) for MRI reconstruction often require large datasets for training. Still, in clinical settings, the domains of datasets are diverse, and how robust DNNs are to domain differences between training and testing datasets has been an open question. Here, we numerically and clinically evaluate the generalization of the reconstruction networks across various domains under clinically practical conditions and provide practical guidance on what points to consider when selecting models for clinical application. METHODS: We compare the reconstruction performance between four network models: U-Net, the deep cascade of convolutional neural networks (DC-CNNs), Hybrid Cascade, and variational network (VarNet). We used the public multicoil dataset fastMRI for training and testing and performed a single-domain test, where the domains of the dataset used for training and testing were the same, and cross-domain tests, where the source and target domains were different. We conducted a single-domain test (Experiment 1) and cross-domain tests (Experiments 2-4), focusing on six factors (the number of images, sampling pattern, acceleration factor, noise level, contrast, and anatomical structure) both numerically and clinically. RESULTS: U-Net had lower performance than the three model-based networks and was less robust to domain shifts between training and testing datasets. VarNet had the highest performance and robustness among the three model-based networks, followed by Hybrid Cascade and DC-CNN. Especially, VarNet showed high performance even with a limited number of training images (200 images/10 cases). U-Net was more robust to domain shifts concerning noise level than the other model-based networks. Hybrid Cascade showed slightly better performance and robustness than DC-CNN, except for robustness to noise-level domain shifts. The results of the clinical evaluations generally agreed with the results of the quantitative metrics. CONCLUSION: In this study, we numerically and clinically evaluated the robustness of the publicly available networks using the multicoil data. Therefore, this study provided practical guidance for clinical applications.

Three-dimensional Multi-parameter Mapping of Relaxation Times and Susceptibility Using Partially RF-spoiled Gradient Echo.

PURPOSE: MR parameter mapping is a technique that obtains distributions of parameters such as relaxation time and proton density (PD) and is starting to be used for disease quantification in clinical diagnoses. Quantitative susceptibility mapping is also promising for the early diagnosis of brain disorders such as degenerative neurological disorders. Therefore, we developed an MR quantitative parameter mapping (QPM) method to map four tissue-related parameters (T1, T2*, PD, and susceptibility) and B1 simultaneously by using a 3D partially RF-spoiled gradient echo (pRSGE). We verified the accuracy and repeatability of QPM in phantom and volunteer experiments. METHODS: Tissue-related parameters are estimated by varying four scan parameters of the 3D pRSGE: flip angle, RF-pulse phase increment, TR and TE, performing multiple image scans, and finding a least-squares fit for an intensity function (which expresses the relationship between the scan parameters and intensity values). The intensity function is analytically complex, but by using a Bloch simulation to create it numerically, the least-squares fitting can be used to estimate the quantitative values. This has the advantage of shortening the image-reconstruction processing time needed to estimate the quantitative values than with methods using pattern matching. RESULTS: A 1.1-mm isotropic resolution scan covering the whole brain was completed with a scan time of approximately 12 minutes, and the reconstruction time using a GPU was approximately 1 minute. The phantom experiments confirmed that both the accuracy and repeatability of the quantitative values were high. The volunteer scans also confirmed that the accuracy of the quantitative values was comparable to that of conventional methods. CONCLUSION: The proposed QPM method can map T1, T2*, PD, susceptibility, and B1 simultaneously within a scan time that can be applied to human subjects.

Simultaneous Arterial and Venous Imaging Using 3D Quantitative Parameter Mapping.

PURPOSE: To increase the number of images that can be acquired in MR examinations using quantitative parameters, we developed a method for obtaining arterial and venous images with mapping of proton density (PD), RF inhomogeneity (B1), longitudinal relaxation time (T1), apparent transverse relaxation time (T2*), and magnetic susceptibility through calculation, all with the same spatial resolution. METHODS: The proposed method uses partially RF-spoiled gradient echo sequences to obtain 3D images of a subject with multiple scan parameters. The PD, B1, T1, T2*, and magnetic susceptibility maps are estimated using the quantification method we previously developed. Arterial images are obtained by adding images using optimized weights to emphasize the arteries. A morphology filter is used to obtain venous images from the magnetic susceptibility maps. For evaluation, images obtained from four out of five healthy volunteers were used to optimize the weights used in the arterial-image calculation, and the optimized weights were applied to the images from the fifth volunteer to obtain an arterial image. Arterial images of the five volunteers were calculated using the leave-one-out method, and the contrast between the arterial and background regions defined using the reference time-of-flight (TOF) method was evaluated using the area under the receiver operation characteristic curve (AUC). The contrast between venous and background regions defined by a reference quantitative susceptibility mapping (QSM) method was also evaluated for the venous image. RESULTS: The AUC to discriminate blood vessels and background using the proposed method was 0.905 for the arterial image and 0.920 for the venous image. CONCLUSION: The results indicate that the arterial images and venous images have high signal intensity at the same region as determined from the reference TOF and QSM methods, demonstrating the possibility of acquiring vasculature images with quantitative parameter mapping through calculation in an integrated manner.

Differentiation Between Multiple System Atrophy and Other Spinocerebellar Degenerations Using Diffusion Kurtosis Imaging.

RATIONALE AND OBJECTIVE: Differentiation between multiple system atrophy (MSA) and other spinocerebellar degenerations showing cerebellar ataxia is often difficult. Hence, we investigated whether magnetic resonance diffusion kurtosis imaging (DKI) could detect pathological changes that occur in these patients and be used for differential diagnosis. METHODS: Thirty-six subjects (12 patients with MSA accompanied by predominant cerebellar ataxia [MSA-C], 10 patients with spinocerebellar ataxias [SCAs] or sporadic adult-onset ataxia of unknown etiology [SAOA], and 14 healthy controls) were examined using 1.5- or 3-T magnetic resonance scanners. From the DKI data, the mean kurtosis, fractional anisotropy, and mean diffusivity values of the pontine crossing tract (PCT), middle cerebellar peduncle, and cerebellum were automatically measured, and the ratios against the values of the corpus callosum were calculated. RESULTS: We found significant decreases in mean kurtosis and fractional anisotropy ratios in the PCT and middle cerebellar peduncle, and a significant increase in the mean diffusivity ratio in the PCT in the MSA-C group, as compared with the SCA/SAOA and control groups (p < 0.027-0.001). Among these metrics, there were no significant differences in the diagnostic performance. By contrast, the ratios in the cerebellum showed no significant differences between the MSA-C and SCA/SAOA groups but were significantly altered when compared with the controls (p < 0.001). CONCLUSION: Quantitative DKI analyses can be used to differentiate between patients with MSA-C and those with SCA/SAOA.

The Relationship between Neurite Density Measured with Confocal Microscopy in a Cleared Mouse Brain and Metrics Obtained from Diffusion Tensor and Diffusion Kurtosis Imaging.

PURPOSE: Diffusional kurtosis imaging (DKI) enables sensitive measurement of tissue microstructure by quantifying the non-Gaussian diffusion of water. Although DKI is widely applied in many situations, histological correlation with DKI analysis is lacking. The purpose of this study was to determine the relationship between DKI metrics and neurite density measured using confocal microscopy of a cleared mouse brain. METHODS: One thy-1 yellow fluorescent protein 16 mouse was deeply anesthetized and perfusion fixation was performed. The brain was carefully dissected out and whole-brain MRI was performed using a 7T animal MRI system. DKI and diffusion tensor imaging (DTI) data were obtained. After the MRI scan, brain sections were prepared and then cleared using aminoalcohols (CUBIC). Confocal microscopy was performed using a two-photon confocal microscope with a laser. Forty-eight ROIs were set on the caudate putamen, seven ROIs on the anterior commissure, and seven ROIs on the ventral hippocampal commissure on the confocal microscopic image and a corresponding MR image. In each ROI, histological neurite density and the metrics of DKI and DTI were calculated. The correlations between diffusion metrics and neurite density were analyzed using Pearson correlation coefficient analysis. RESULTS: Mean kurtosis (MK) (P = 5.2 × 10-9, r = 0.73) and radial kurtosis (P = 2.3 × 10-9, r = 0.74) strongly correlated with neurite density in the caudate putamen. The correlation between fractional anisotropy (FA) and neurite density was moderate (P = 0.0030, r = 0.42). In the anterior commissure and the ventral hippocampal commissure, neurite density and FA are very strongly correlated (P = 1.3 × 10-5, r = 0.90). MK in these areas were very high value and showed no significant correlation (P = 0.48). CONCLUSION: DKI accurately reflected neurite density in the area with crossing fibers, potentially allowing evaluation of complex microstructures.

Differential diagnosis of parkinsonism by a combined use of diffusion kurtosis imaging and quantitative susceptibility mapping.

PURPOSE: We investigated whether diffusion kurtosis imaging (DKI) and quantitative susceptibility mapping (QSM) could detect pathological changes that occur in Parkinson's disease (PD), multiple system atrophy with predominant parkinsonism (MSA-P) or predominant cerebellar ataxia (MSA-C), and progressive supranuclear palsy syndrome (PSPS) and thus be used for differential diagnosis that is often difficult. METHODS: Seventy patients (41 with PD, 6 with MSA-P, 7 with MSA-C, 16 with PSPS) and 20 healthy controls were examined using a 3.0 T MRI scanner. From DKI and QSM data, we automatically obtained mean kurtosis (MK), fractional anisotropy (FA), and mean diffusivity (MD) values of the midbrain tegmentum (MBT), pontine crossing tract (PCT), and superior/middle cerebellar peduncles (CPs), which were used to calculate diffusion MBT/PCT ratios (dMPRs) and diffusion superior/middle CP ratios (dCPRs), as well as MS (magnetic susceptibility) values of the anterior/posterior putamen (PUa and PUp) and globus pallidus (GP). RESULTS: dMPRs of MK were significantly decreased in PSPS and increased in MSA-C compared with the other groups, while dCPRs of MK showed significant differences only between MSA-C and PD, PSPS, or control. MS values were significantly increased in the PUp of MSA-P and in the PUa and GP of PSPS compared with those in PD. The combined use of MK-dMPR and MS-PUp showed sensitivities of 83-100% and specificities of 81-100% for discriminating among the disease groups, respectively. CONCLUSION: A quantitative assessment using DKI and QSM analyses, particularly MK-dMPR and MS-PUp values, can readily identify patients with parkinsonism.

Detection of changes in the periaqueductal gray matter of patients with episodic migraine using quantitative diffusion kurtosis imaging: preliminary findings.

INTRODUCTION: The periaqueductal gray matter (PAG) is considered to play an important role in generating migraine, but findings from imaging studies remain unclear. Therefore, we investigated whether diffusion kurtosis imaging (DKI) can detect changes in the PAG of migraine patients. METHODS: We obtained source images for DKI from 20 patients with episodic migraine and 20 healthy controls using a 3 T magnetic resonance imaging scanner. Mean kurtosis (MK), fractional anisotropy (FA), and mean diffusivity (MD) maps were generated, and the values of the PAG and other deep gray and white matter structures were automatically measured using an atlas-based region-of-interest analysis. The metrics of these structures were compared between the patients and controls. RESULTS: The MK and MD values of the PAG were significantly increased in the migraine patients compared with the controls (p < 0.05). The FA values were not significantly different. There were no significant differences in the metrics of the other structures between the patients and controls. The MK values of the PAG were significantly positively correlated with both age and the untreated period in the patient group under univariate analysis (r = 0.53 and 0.56, respectively; p < 0.05) but not multivariate analysis. CONCLUSIONS: DKI detected significant increases in the MK and MD values of the PAG in patients with migraine, which suggests that structural changes in the PAG are associated with the pathophysiological mechanisms of migraine.

Optimization of Scan Parameters to Reduce Acquisition Time for Diffusion Kurtosis Imaging at 1.5T.

PURPOSE: To shorten acquisition of diffusion kurtosis imaging (DKI) in 1.5-tesla magnetic resonance (MR) imaging, we investigated the effects of the number of b-values, diffusion direction, and number of signal averages (NSA) on the accuracy of DKI metrics. METHODS: We obtained 2 image datasets with 30 gradient directions, 6 b-values up to 2500 s/mm(2), and 2 signal averages from 5 healthy volunteers and generated DKI metrics, i.e., mean, axial, and radial kurtosis (MK, K∥, and K⊥) maps, from various combinations of the datasets. These maps were estimated by using the intraclass correlation coefficient (ICC) with those from the full datasets. RESULTS: The MK and K⊥ maps generated from the datasets including only the b-value of 2500 s/mm(2) showed excellent agreement (ICC, 0.96 to 0.99). Under the same acquisition time and diffusion directions, agreement was better of MK, K∥, and K⊥ maps obtained with 3 b-values (0, 1000, and 2500 s/mm(2)) and 4 signal averages than maps obtained with any other combination of numbers of b-value and varied NSA. Good agreement (ICC > 0.6) required at least 20 diffusion directions in all the metrics. CONCLUSION: MK and K⊥ maps with ICC greater than 0.95 can be obtained at 1.5T within 10 min (b-value = 0, 1000, and 2500 s/mm(2); 20 diffusion directions; 4 signal averages; slice thickness, 6 mm with no interslice gap; number of slices, 12).

See-through Brains and Diffusion Tensor MRI Clarified Fiber Connections: A Preliminary Microstructural Study in a Mouse with Callosal Agenesis.

Clearing methods that render the brain optically transparent allow high-resolution three-dimensional (3D) imaging of neural networks. We used diffusion tensor imaging (DTI) and two-photon imaging of cleared brains to analyze white matter in BTBR mice. We confirmed corpus callosum agenesis and identified an abnormal commissure close to the third ventricle. DTI and cleared-brain two-photon imaging revealed that these commissural fibers constituted a frontal clustering of the ventral hippocampal commissure and provided a detailed assessment of white matter structure in mice.

Differentiation among parkinsonisms using quantitative diffusion kurtosis imaging.

Differential diagnoses among Parkinson's disease (PD), multiple system atrophy (MSA), and progressive supranuclear palsy syndrome (PSPS) are often difficult. Hence, we investigated whether diffusion kurtosis imaging (DKI) could detect pathological changes that occur in these disorders and be used to differentiate between such patients. Fourteen patients (five with PD, four MSA, and five PSPS) and six healthy controls were examined using a 1.5-T scanner. Mean kurtosis (MK), fractional anisotropy, and mean diffusivity maps were generated, and these values of the midbrain tegmentum (MBT) and pontine crossing tract (PCT), as well as MBT/PCT ratios, were obtained. We found no significant differences in MBT and PCT values on DKI maps among the groups. In contrast, MBT/PCT ratios from MK maps were significantly increased in the MSA group and decreased in the PSPS group compared with the other groups. MBT/PCT ratios from mean diffusivity maps showed a significant increase in the PSPS group. Therefore, quantitative DKI analyses, particularly the MBT/PCT ratio from MK maps, can differentiate patients with parkinsonisms.