Motion robust coronary MR angiography using zigzag centric ky-kz trajectory and high-resolution deep learning reconstruction.

PURPOSE: To develop a new MR coronary angiography (MRCA) technique by employing a zigzag fan-shaped centric ky-kz k-space trajectory combined with high-resolution deep learning reconstruction (HR-DLR). METHODS: All imaging data were acquired from 12 healthy subjects and 2 patients using two clinical 3-T MR imagers, with institutional review board approval. Ten healthy subjects underwent both standard 3D fast gradient echo (sFGE) and centric ky-kz k-space trajectory FGE (cFGE) acquisitions to compare the scan time and image quality. Quantitative measures were also performed for signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) as well as sharpness of the vessel. Furthermore, the feasibility of the proposed cFGE sequence was assessed in two patients. For assessing the feasibility of the centric ky-kz trajectory, the navigator-echo window of a 30-mm threshold was applied in cFGE, whereas sFGE was applied using a standard 5-mm threshold. Image quality of MRCA using cFGE with HR-DLR and sFGE without HR-DLR was scored in a 5-point scale (non-diagnostic = 1, fair = 2, moderate = 3, good = 4, and excellent = 5). Image evaluation of cFGE, applying HR-DLR, was compared with sFGE without HR-DLR. Friedman test, Wilcoxon signed-rank test, or paired t tests were performed for the comparison of related variables. RESULTS: The actual MRCA scan time of cFGE with a 30-mm threshold was acquired in less than 5 min, achieving nearly 100% efficiency, showcasing its expeditious and robustness. In contrast, sFGE was acquired with a 5-mm threshold and had an average scan time of approximately 15 min. Overall image quality for MRCA was scored 3.3 for sFGE and 2.7 for cFGE without HR-DLR but increased to 3.6 for cFGE with HR-DLR and (p < 0.05). The clinical result of patients obtained within 5 min showed good quality images in both patients, even with a stent, without artifacts. Quantitative measures of SNR, CNR, and sharpness of vessel presented higher in cFGE with HR-DLR. CONCLUSION: Our findings demonstrate a robust, time-efficient solution for high-quality MRCA, enhancing patient comfort and increasing clinical throughput.

Effect of deep learning-based reconstruction on high-resolution three-dimensional T2-weighted fast asymmetric spin-echo imaging in the preoperative evaluation of cerebellopontine angle tumors.

PURPOSE: We aimed to evaluate the effect of deep learning-based reconstruction (DLR) on high-spatial-resolution three-dimensional T2-weighted fast asymmetric spin-echo (HR-3D T2-FASE) imaging in the preoperative evaluation of cerebellopontine angle (CPA) tumors. METHODS: This study included 13 consecutive patients who underwent preoperative HR-3D T2-FASE imaging using a 3 T MRI scanner. The reconstruction voxel size of HR-3D T2-FASE imaging was 0.23 × 0.23 × 0.5 mm. The contrast-to-noise ratios (CNRs) of the structures were compared between HR-3D T2-FASE images with and without DLR. The observers' preferences based on four categories on the tumor side on HR-3D T2-FASE images were evaluated. The facial nerve in relation to the tumor on HR-3D T2-FASE images was assessed with reference to intraoperative findings. RESULTS: The mean CNR between the tumor and trigeminal nerve and between the cerebrospinal fluid and trigeminal nerve was significantly higher for DLR images than non-DLR-based images (14.3 ± 8.9 vs. 12.0 ± 7.6, and 66.4 ± 12.0 vs. 53.9 ± 8.5, P < 0.001, respectively). The observer's preference for the depiction and delineation of the tumor, cranial nerves, vessels, and location relation on DLR HR-3D T2FASE images was superior to that on non-DLR HR-3D T2FASE images in 7 (54%), 6 (46%), 6 (46%), and 6 (46%) of 13 cases, respectively. The facial nerves around the tumor on HR-3D T2-FASE images were visualized accurately in five (38%) cases with DLR and in four (31%) without DLR. CONCLUSION: DLR HR-3D T2-FASE imaging is useful for the preoperative assessment of CPA tumors.

Optimizing High-Resolution MR Angiography: The Synergistic Effects of 3D Wheel Sampling and Deep Learning-Based Reconstruction.

OBJECTIVE: The aim of this study was to assess the utility of the combined use of 3D wheel sampling and deep learning-based reconstruction (DLR) for intracranial high-resolution (HR)-time-of-flight (TOF)-magnetic resonance angiography (MRA) at 3 T. METHODS: This prospective study enrolled 20 patients who underwent head MRI at 3 T, including TOF-MRA. We used 3D wheel sampling called "fast 3D" and DLR for HR-TOF-MRA (spatial resolution, 0.39 × 0.59 × 0.5 mm3) in addition to conventional MRA (spatial resolution, 0.39 × 0.89 × 1 mm3). We compared contrast and contrast-to-noise ratio between the blood vessels (basilar artery and anterior cerebral artery) and brain parenchyma, full width at half maximum in the P3 segment of the posterior cerebral artery among 3 protocols. Two board-certified radiologists evaluated noise, contrast, sharpness, artifact, and overall image quality of 3 protocols. RESULTS: The contrast and contrast-to-noise ratio of fast 3D-HR-MRA with DLR are comparable or higher than those of conventional MRA and fast 3D-HR-MRA without DLR. The full width at half maximum was significantly lower in fast 3D-MRA with and without DLR than in conventional MRA (P = 0.006, P < 0.001). In qualitative evaluation, fast 3D-MRA with DLR had significantly higher sharpness and overall image quality than conventional MRA and fast 3D-MRA without DLR (sharpness: P = 0.021, P = 0.001; overall image quality: P = 0.029, P < 0.001). CONCLUSIONS: The combination of 3D wheel sampling and DLR can improve visualization of arteries in intracranial TOF-MRA.

Emergence of number sense through the integration of multimodal information: developmental learning insights from neural network models.

Introduction: Associating multimodal information is essential for human cognitive abilities including mathematical skills. Multimodal learning has also attracted attention in the field of machine learning, and it has been suggested that the acquisition of better latent representation plays an important role in enhancing task performance. This study aimed to explore the impact of multimodal learning on representation, and to understand the relationship between multimodal representation and the development of mathematical skills. Methods: We employed a multimodal deep neural network as the computational model for multimodal associations in the brain. We compared the representations of numerical information, that is, handwritten digits and images containing a variable number of geometric figures learned through single- and multimodal methods. Next, we evaluated whether these representations were beneficial for downstream arithmetic tasks. Results: Multimodal training produced better latent representation in terms of clustering quality, which is consistent with previous findings on multimodal learning in deep neural networks. Moreover, the representations learned using multimodal information exhibited superior performance in arithmetic tasks. Discussion: Our novel findings experimentally demonstrate that changes in acquired latent representations through multimodal association learning are directly related to cognitive functions, including mathematical skills. This supports the possibility that multimodal learning using deep neural network models may offer novel insights into higher cognitive functions.

A multimodal deep learning approach for the prediction of cognitive decline and its effectiveness in clinical trials for Alzheimer's disease.

Alzheimer's disease is one of the most important health-care challenges in the world. For decades, numerous efforts have been made to develop therapeutics for Alzheimer's disease, but most clinical trials have failed to show significant treatment effects on slowing or halting cognitive decline. Among several challenges in such trials, one recently noticed but unsolved is biased allocation of fast and slow cognitive decliners to treatment and placebo groups during randomization caused by the large individual variation in the speed of cognitive decline. This allocation bias directly results in either over- or underestimation of the treatment effect from the outcome of the trial. In this study, we propose a stratified randomization method using the degree of cognitive decline predicted by an artificial intelligence model as a stratification index to suppress the allocation bias in randomization and evaluate its effectiveness by simulation using ADNI data set.

Elucidating multifinal and equifinal pathways to developmental disorders by constructing real-world neurorobotic models.

Vigorous research has been conducted to accumulate biological and theoretical knowledge about neurodevelopmental disorders, including molecular, neural, computational, and behavioral characteristics; however, these findings remain fragmentary and do not elucidate integrated mechanisms. An obstacle is the heterogeneity of developmental pathways causing clinical phenotypes. Additionally, in symptom formations, the primary causes and consequences of developmental learning processes are often indistinguishable. Herein, we review developmental neurorobotic experiments tackling problems related to the dynamic and complex properties of neurodevelopmental disorders. Specifically, we focus on neurorobotic models under predictive processing lens for the study of developmental disorders. By constructing neurorobotic models with predictive processing mechanisms of learning, perception, and action, we can simulate formations of integrated causal relationships among neurodynamical, computational, and behavioral characteristics in the robot agents while considering developmental learning processes. This framework has the potential to bind neurobiological hypotheses (excitation-inhibition imbalance and functional disconnection), computational accounts (unusual encoding of uncertainty), and clinical symptoms. Developmental neurorobotic approaches may serve as a complementary research framework for integrating fragmented knowledge and overcoming the heterogeneity of neurodevelopmental disorders.

Exploring the impact of super-resolution deep learning on MR angiography image quality.

PURPOSE: The aim of this study is to assess the effect of super-resolution deep learning-based reconstruction (SR-DLR), which uses k-space properties, on image quality of intracranial time-of-flight (TOF) magnetic resonance angiography (MRA) at 3 T. METHODS: This retrospective study involved 35 patients who underwent intracranial TOF-MRA using a 3-T MRI system with SR-DLR based on k-space properties in October and November 2022. We reconstructed MRA with SR-DLR (matrix = 1008 × 1008) and MRA without SR-DLR (matrix = 336 × 336). We measured the signal-to-noise ratio (SNR), contrast, and contrast-to-noise ratio (CNR) in the basilar artery (BA) and the anterior cerebral artery (ACA) and the sharpness of the posterior cerebral artery (PCA) using the slope of the signal intensity profile curve at the half-peak points. Two radiologists evaluated image noise, artifacts, contrast, sharpness, and overall image quality of the two image types using a 4-point scale. We compared quantitative and qualitative scores between images with and without SR-DLR using the Wilcoxon signed-rank test. RESULTS: The SNRs, contrasts, and CNRs were all significantly higher in images with SR-DLR than those without SR-DLR (p < 0.001). The slope was significantly greater in images with SR-DLR than those without SR-DLR (p < 0.001). The qualitative scores in MRAs with SR-DLR were all significantly higher than MRAs without SR-DLR (p < 0.001). CONCLUSION: SR-DLR with k-space properties can offer the benefits of increased spatial resolution without the associated drawbacks of longer scan times and reduced SNR and CNR in intracranial MRA.

Aberrant sense of agency induced by delayed prediction signals in schizophrenia: a computational modeling study.

Aberrant sense of agency (SoA, a feeling of control over one's own actions and their subsequent events) has been considered key to understanding the pathology of schizophrenia. Behavioral studies have demonstrated that a bidirectional (i.e., excessive and diminished) SoA is observed in schizophrenia. Several neurophysiological and theoretical studies have suggested that aberrancy may be due to temporal delays (TDs) in sensory-motor prediction signals. Here, we examined this hypothesis via computational modeling using a recurrent neural network (RNN) expressing the sensory-motor prediction process. The proposed model successfully reproduced the behavioral features of SoA in healthy controls. In addition, simulation of delayed prediction signals reproduced the bidirectional schizophrenia-pattern SoA, whereas three control experiments (random noise addition, TDs in outputs, and TDs in inputs) demonstrated no schizophrenia-pattern SoA. These results support the TD hypothesis and provide a mechanistic understanding of the pathology underlying aberrant SoA in schizophrenia.

Feasibility study of super-resolution deep learning-based reconstruction using k-space data in brain diffusion-weighted images.

PURPOSE: The purpose of this study is to evaluate the influence of super-resolution deep learning-based reconstruction (SR-DLR), which utilizes k-space data, on the quality of images and the quantitation of the apparent diffusion coefficient (ADC) for diffusion-weighted images (DWI) in brain magnetic resonance imaging (MRI). METHODS: A retrospective analysis was performed on 34 patients who had undergone DWI using a 3 T MRI system with SR-DLR reconstruction based on k-space data in August 2022. DWI was reconstructed with SR-DLR (Matrix = 684 × 684) and without SR-DLR (Matrix = 228 × 228). Measurements were made of the signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR) in white matter (WM) and grey matter (GM), and the full width at half maximum (FWHM) of the septum pellucidum. Two radiologists assessed image noise, contrast, artifacts, blur, and the overall quality of three image types using a four-point scale. Quantitative and qualitative scores between images with and without SR-DLR were compared using the Wilcoxon signed-rank test. RESULTS: Images with SR-DLR showed significantly higher SNRs and CNRs than those without SR-DLR (p < 0.001). No statistically significant variances were found in the apparent diffusion coefficients (ADCs) in WM and GM between images with and without SR-DLR (ADC in WM, p = 0.945; ADC in GM, p = 0.235). Moreover, the FWHM without SR-DLR was notably lower compared to that with SR-DLR (p < 0.001). CONCLUSION: SR-DLR has the potential to augment the quality of DWI in DL MRI scans without significantly impacting ADC quantitation.

Intracranial aneurysms treated with stent-assisted coil embolization: evaluation with four-dimensional ultrashort-TE MR angiography.

OBJECTIVES: As a novel follow-up method for intracranial aneurysms treated with stent-assisted coil embolization (SACE), we developed four-dimensional magnetic resonance angiography (MRA) with minimized acoustic noise utilizing ultrashort-echo time (4D mUTE-MRA). We aimed to assess whether 4D mUTE-MRA is useful for the evaluation of intracranial aneurysms treated with SACE. METHODS: This study included 31 consecutive patients with intracranial aneurysm treated with SACE who underwent 4D mUTE-MRA at 3 T and digital subtraction angiography (DSA). For 4D mUTE-MRA, five dynamic MRA images with a spatial resolution of 0.5 × 0.5 × 0.5 mm3 were obtained every 200 ms. Two readers independently reviewed the 4D mUTE-MRA images to evaluate the aneurysm occlusion status (total occlusion, residual neck, and residual aneurysm) and the flow in the stent using a 4-point scale (from 1 [not visible] to 4 [excellent]). The interobserver and intermodality agreement was assessed using κ statistics. RESULTS: On DSA images, 10 aneurysms were classified as total occlusion, 14 as residual neck, and 7 as residual aneurysm. In terms of aneurysm occlusion status, the intermodality and interobserver agreement was excellent (κ = 0.92 and κ = 0.96, respectively). For the flow in the stents on 4D mUTE-MRA, the mean score was significantly higher for single stents than multiple stents (p < .001) and for open-cell type stents than closed-cell type (p < .01). CONCLUSIONS: 4D mUTE-MRA is a useful tool with a high spatial and temporal resolution for the evaluation of intracranial aneurysms treated with SACE. KEY POINTS: • In the evaluation of intracranial aneurysms treated with SACE on 4D mUTE-MRA and DSA, the intermodality and interobserver agreement in aneurysm occlusion status was excellent. • 4D mUTE-MRA shows good to excellent visualization of flow in the stents, especially for cases treated with a single or open-cell stent. • 4D mUTE-MRA can provide hemodynamic information related to embolized aneurysms and the distal arteries to stented parent arteries.

Simulating developmental diversity: Impact of neural stochasticity on atypical flexibility and hierarchy.

Introduction: Investigating the pathological mechanisms of developmental disorders is a challenge because the symptoms are a result of complex and dynamic factors such as neural networks, cognitive behavior, environment, and developmental learning. Recently, computational methods have started to provide a unified framework for understanding developmental disorders, enabling us to describe the interactions among those multiple factors underlying symptoms. However, this approach is still limited because most studies to date have focused on cross-sectional task performance and lacked the perspectives of developmental learning. Here, we proposed a new research method for understanding the mechanisms of the acquisition and its failures in hierarchical Bayesian representations using a state-of-the-art computational model, referred to as in silico neurodevelopment framework for atypical representation learning. Methods: Simple simulation experiments were conducted using the proposed framework to examine whether manipulating the neural stochasticity and noise levels in external environments during the learning process can lead to the altered acquisition of hierarchical Bayesian representation and reduced flexibility. Results: Networks with normal neural stochasticity acquired hierarchical representations that reflected the underlying probabilistic structures in the environment, including higher-order representation, and exhibited good behavioral and cognitive flexibility. When the neural stochasticity was high during learning, top-down generation using higher-order representation became atypical, although the flexibility did not differ from that of the normal stochasticity settings. However, when the neural stochasticity was low in the learning process, the networks demonstrated reduced flexibility and altered hierarchical representation. Notably, this altered acquisition of higher-order representation and flexibility was ameliorated by increasing the level of noises in external stimuli. Discussion: These results demonstrated that the proposed method assists in modeling developmental disorders by bridging between multiple factors, such as the inherent characteristics of neural dynamics, acquisitions of hierarchical representation, flexible behavior, and external environment.

Interaction between Functional Connectivity and Neural Excitability in Autism: A Novel Framework for Computational Modeling and Application to Biological Data.

Functional connectivity (FC) and neural excitability may interact to affect symptoms of autism spectrum disorder (ASD). We tested this hypothesis with neural network simulations, and applied it with functional magnetic resonance imaging (fMRI). A hierarchical recurrent neural network embodying predictive processing theory was subjected to a facial emotion recognition task. Neural network simulations examined the effects of FC and neural excitability on changes in neural representations by developmental learning, and eventually on ASD-like performance. Next, by mapping each neural network condition to subject subgroups on the basis of fMRI parameters, the association between ASD-like performance in the simulation and ASD diagnosis in the corresponding subject subgroup was examined. In the neural network simulation, the more homogeneous the neural excitability of the lower-level network, the more ASD-like the performance (reduced generalization and emotion recognition capability). In addition, in homogeneous networks, the higher the FC, the more ASD-like performance, while in heterogeneous networks, the higher the FC, the less ASD-like performance, demonstrating that FC and neural excitability interact. As an underlying mechanism, neural excitability determines the generalization capability of top-down prediction, and FC determines whether the model's information processing will be top-down prediction-dependent or bottom-up sensory-input dependent. In fMRI datasets, ASD was actually more prevalent in subject subgroups corresponding to the network condition showing ASD-like performance. The current study suggests an interaction between FC and neural excitability, and presents a novel framework for computational modeling and biological application of a developmental learning process underlying cognitive alterations in ASD.

Ultrafast cervcial spine MRI protocol using deep learning-based reconstruction: Diagnostic equivalence to a conventional protocol.

PURPOSE: A major drawback of magnetic resonance imaging (MRI) is its limited imaging speed. This study proposed an ultrafast cervical spine MRI protocol (2 min 57 s) using deep learning-based reconstruction (DLR) and compared the diagnostic results to those of conventional MRI protocols (12 min 54 s). METHODS: Fifty patients who underwent cervical spine MRI using both conventional and ultrafast protocols, including sagittal T1-weighted, T2-weighted, short-TI inversion recovery, and axial T2*-weighted imaging were included in this study. The ultrafast protocol shortened the acquisition time to approximately-one-fourth of that of the conventional protocol by reducing the phase matrix, oversampling rate, and number of excitations, and by applying compressed sensing. To compensate for the decreased signal-to-noise ratio caused by acceleration, noise reduction using DLR was performed. For image interpretation, three neuroradiologists graded or classified degenerative changes, including central canal stenosis, foraminal stenosis, endplate degeneration, disc degeneration, and disc hernia. The presence of other pathologies was also recorded. Given the absence of a reference standard, we tested the interchangeability of the two protocols by calculating the 95% confidence interval (CI) of the individual equivalence index. We also assessed the inter-protocol intra-reader agreement using kappa statistics. RESULTS: Except for endplate degeneration, the 95 % CI of the individual equivalence index for all variables did not exceed 5 %, indicating interchangeability between the two protocols. The kappa values ranged from 0.600 to 0.977, indicating substantial to almost perfect agreement. CONCLUSIONS: The proposed ultrafast MRI protocol yielded almost equivalent diagnostic results compared as the conventional protocol.

Emergence of sensory attenuation based upon the free-energy principle.

The brain attenuates its responses to self-produced exteroceptions (e.g., we cannot tickle ourselves). Is this phenomenon, known as sensory attenuation, enabled innately, or acquired through learning? Here, our simulation study using a multimodal hierarchical recurrent neural network model, based on variational free-energy minimization, shows that a mechanism for sensory attenuation can develop through learning of two distinct types of sensorimotor experience, involving self-produced or externally produced exteroceptions. For each sensorimotor context, a particular free-energy state emerged through interaction between top-down prediction with precision and bottom-up sensory prediction error from each sensory area. The executive area in the network served as an information hub. Consequently, shifts between the two sensorimotor contexts triggered transitions from one free-energy state to another in the network via executive control, which caused shifts between attenuating and amplifying prediction-error-induced responses in the sensory areas. This study situates emergence of sensory attenuation (or self-other distinction) in development of distinct free-energy states in the dynamic hierarchical neural system.

Computational Mechanisms of Osmoregulation: A Reinforcement Learning Model for Sodium Appetite.

Homeostatic control with oral nutrient intake is a vital complex system involving the orderly interactions between the external and internal senses, behavioral control, reward learning, and decision-making. Sodium appetite is a representative system and has been intensively investigated in animal models of homeostatic systems and oral nutrient intake. However, the system-level mechanisms for regulating sodium intake behavior and homeostatic control remain unclear. In the current study, we attempted to provide a mechanistic understanding of sodium appetite behavior by using a computational model, the homeostatic reinforcement learning model, in which homeostatic behaviors are interpreted as reinforcement learning processes. Through simulation experiments, we confirmed that our homeostatic reinforcement learning model successfully reproduced homeostatic behaviors by regulating sodium appetite. These behaviors include the approach and avoidance behaviors to sodium according to the internal states of individuals. In addition, based on the assumption that the sense of taste is a predictor of changes in the internal state, the homeostatic reinforcement learning model successfully reproduced the previous paradoxical observations of the intragastric infusion test, which cannot be explained by the classical drive reduction theory. Moreover, we extended the homeostatic reinforcement learning model to multimodal data, and successfully reproduced the behavioral tests in which water and sodium appetite were mediated by each other. Finally, through an experimental simulation of chemical manipulation in a specific neural population in the brain stem, we proposed a testable hypothesis for the function of neural circuits involving sodium appetite behavior. The study results support the idea that osmoregulation via sodium appetitive behavior can be understood as a reinforcement learning process, and provide a mechanistic explanation for the underlying neural mechanisms of decision-making related to sodium appetite and homeostatic behavior.

Schizophrenia-Mimicking Layers Outperform Conventional Neural Network Layers.

We have reported nanometer-scale three-dimensional studies of brain networks of schizophrenia cases and found that their neurites are thin and tortuous when compared to healthy controls. This suggests that connections between distal neurons are suppressed in microcircuits of schizophrenia cases. In this study, we applied these biological findings to the design of a schizophrenia-mimicking artificial neural network to simulate the observed connection alteration in the disorder. Neural networks that have a "schizophrenia connection layer" in place of a fully connected layer were subjected to image classification tasks using the MNIST and CIFAR-10 datasets. The results revealed that the schizophrenia connection layer is tolerant to overfitting and outperforms a fully connected layer. The outperformance was observed only for networks using band matrices as weight windows, indicating that the shape of the weight matrix is relevant to the network performance. A schizophrenia convolution layer was also tested using the VGG configuration, showing that 60% of the kernel weights of the last three convolution layers can be eliminated without loss of accuracy. The schizophrenia layers can be used instead of conventional layers without any change in the network configuration and training procedures; hence, neural networks can easily take advantage of these layers. The results of this study suggest that the connection alteration found in schizophrenia is not a burden to the brain, but has functional roles in brain performance.

A high-generalizability machine learning framework for predicting the progression of Alzheimer's disease using limited data.

Alzheimer's disease is a neurodegenerative disease that imposes a substantial financial burden on society. A number of machine learning studies have been conducted to predict the speed of its progression, which varies widely among different individuals, for recruiting fast progressors in future clinical trials. However, because the data in this field are very limited, two problems have yet to be solved: the first is that models built on limited data tend to induce overfitting and have low generalizability, and the second is that no cross-cohort evaluations have been done. Here, to suppress the overfitting caused by limited data, we propose a hybrid machine learning framework consisting of multiple convolutional neural networks that automatically extract image features from the point of view of brain segments, which are relevant to cognitive decline according to clinical findings, and a linear support vector classifier that uses extracted image features together with non-image information to make robust final predictions. The experimental results indicate that our model achieves superior performance (accuracy: 0.88, area under the curve [AUC]: 0.95) compared with other state-of-the-art methods. Moreover, our framework demonstrates high generalizability as a result of evaluations using a completely different cohort dataset (accuracy: 0.84, AUC: 0.91) collected from a different population than that used for training.

Deep Feature Extraction for Resting-State Functional MRI by Self-Supervised Learning and Application to Schizophrenia Diagnosis.

In this study, we propose a deep-learning technique for functional MRI analysis. We introduced a novel self-supervised learning scheme that is particularly useful for functional MRI wherein the subject identity is used as the teacher signal of a neural network. The neural network is trained solely based on functional MRI-scans, and the training does not require any explicit labels. The proposed method demonstrated that each temporal volume of resting state functional MRI contains enough information to identify the subject. The network learned a feature space in which the features were clustered per subject for the test data as well as for the training data; this is unlike the features extracted by conventional methods including region of interests (ROIs) pooling signals and principal component analysis. In addition, applying a simple linear classifier to the per-subject mean of the features (namely "identity feature"), we demonstrated that the extracted features could contribute to schizophrenia diagnosis. The classification accuracy of our identity features was comparable to that of the conventional functional connectivity. Our results suggested that our proposed training scheme of the neural network captured brain functioning related to the diagnosis of psychiatric disorders as well as the identity of the subject. Our results together highlight the validity of our proposed technique as a design for self-supervised learning.

Applicability of deep learning-based reconstruction trained by brain and knee 3T MRI to lumbar 1.5T MRI.

BACKGROUND: Several deep learning-based methods have been proposed for addressing the long scanning time of magnetic resonance imaging. Most are trained using brain 3T magnetic resonance images, but is unclear whether performance is affected when applying these methods to different anatomical sites and at different field strengths. PURPOSE: To validate the denoising performance of deep learning-based reconstruction method trained by brain and knee 3T magnetic resonance images when applied to lumbar 1.5T magnetic resonance images. MATERIAL AND METHODS: Using a 1.5T scanner, we obtained lumber T2-weighted sequences in 10 volunteers using three different scanning times: 228 s (standard), 119 s (double-fast), and 68 s (triple-fast). We compared the images obtained by the standard sequence with those obtained by the deep learning-based reconstruction-applied faster sequences. RESULTS: Signal-to-noise ratio values were significantly higher for deep learning-based reconstruction-double-fast than for standard and did not differ significantly between deep learning-based reconstruction-triple-fast and standard. Contrast-to-noise ratio values also did not differ significantly between deep learning-based reconstruction-triple-fast and standard. Qualitative scores for perceived signal-to-noise ratio and overall image quality were significantly higher for deep learning-based reconstruction-double fast and deep learning-based reconstruction-triple-fast than for standard. Average scores for sharpness, contrast, and structure visibility were equal to or higher for deep learning-based reconstruction-double-fast and deep learning-based reconstruction-triple-fast than for standard, but the differences were not statistically significant. The average scores for artifact were lower for deep learning-based reconstruction-double-fast and deep learning-based reconstruction-triple-fast than for standard, but the differences were not statistically significant. CONCLUSION: The deep learning-based reconstruction method trained by 3T brain and knee images may reduce the scanning time of 1.5T lumbar magnetic resonance images by one-third without sacrificing image quality.