Domain-Randomized Deep Learning for Neuroimage Analysis: Selecting Training Strategies, Navigating Challenges, and Maximizing Benefits.

A variety of methods have been applied to the architectural configuration and learning or training of artificial deep neural networks (DNN). These methods play a crucial role in the success or failure of the DNN for most problems and applications. Evolutionary algorithms (EAs) are gaining momentum as a computationally feasible method for the automated optimization of DNNs. Neuroevolution is a term, which describes these processes of automated configuration and training of DNNs using EAs. While many works exist in the literature, no comprehensive surveys currently exist focusing exclusively on the strengths and limitations of using neuroevolution approaches in DNNs. Absence of such surveys can lead to a disjointed and fragmented field preventing DNNs researchers potentially adopting neuroevolutionary methods in their own research, resulting in lost opportunities for wider application within real-world deep learning problems. This article presents a comprehensive survey, discussion, and evaluation of the state-of-the-art in using EAs for architectural configuration and training of DNNs. This article highlights the most pertinent current issues and challenges in neuroevolution and identifies multiple promising future research directions. <p xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><i>Impact Statement—</i>The concept of deep learning originated from the study of artificial neural networks (ANNs). ANNs have achieved extraordinary results in a variety of diverse application areas. Numerous methods have been applied to the architectural configuration and learning or training of artificial DNN and these methods play a crucial role in the success or failure of the DNN for most problems and applications. Recently, EAs have been gaining momentum as a computationally feasible method (called neuroevolution) for the automated configuration and learning or training of DNNs. This article reviews over 170 recent scientific papers describing how major EAs paradigms are being applied by researchers to the configuration and optimization of multiple DNNs. By articulating a clear understanding of the context, state-of-the-art, and feasibility of Neuroevolution, researchers in AI, EAs, and DNN will benefit from this article. The impact of this article comes from contributing toward enhancing research capacity, knowledge, and skills for researchers currently working in neuroevolution and actively engaging those considering becoming involved in this area.

Deep Learning (DL) has become a prominent machine learning technique due to the availability of efficient computational resources in the form of Graphics Processing Units (GPUs), large-scale datasets and a variety of models. The newer generation of GPUs are being designed with special emphasis on optimizing performance for DL applications. Also, the availability of easy-to-use DL frameworks—like PyTorch and TensorFlow— has enhanced productivity of domain experts to work on their custom DL applications from diverse domains. However, existing Deep Neural Network (DNN) training approaches may not fully utilize the newly emerging powerful GPUs like the NVIDIA A100—this is the primary issue that we address in this paper. Our motivating analyses show that the GPU utilization on NVIDIA A100 can be as low as 43% using traditional DNN training approaches for small-to-medium DL models and input data size. This paper proposes AccDP—a data-parallel distributed DNN training approach—to accelerate GPU-based DL applications. AccDP exploits the Message Passing Interface (MPI) communication library coupled with the NVIDIA’s Multi-Process Service (MPS) to increase the amount of work assigned to parallel GPUs resulting in higher utilization of compute resources. We evaluate our proposed design on different small-to-medium DL models and input sizes on the state-of-the-art HPC clusters. By injecting more parallelism into DNN training using our approach, the evaluation shows up to 58% improvement in training performance on a single GPU and up to 62% on 16 GPUs compared to regular DNN training. Furthermore, we conduct an in-depth characterization to determine the impact of several DNN training factors and best practices—including the batch size and the number of data loading workers— to optimally utilize GPU devices. To the best of our knowledge, this is the first work that explores the use of MPS and MPI to maximize the utilization of GPUs in distributed DNN training.

Stroke is one of the leading causes of adult disability throughout the world and, even though neural mechanisms of loss of function have been extensively studied, many aspects of poststroke cerebral responses remain poorly understood. Of particular interest is the mounting evidence of the capacity of the adult brain to reorganize after injury, which is believed to contribute to limitation of the extent of neural dysfunction and to restoration of affected neural functions. Processes such as neuronal plasticity, glial proliferation, and neovascularization may be essential for preservation or recovery of function after stroke and may conceivably go hand-in-hand at nearby and remote sites of active tissue reorganization.1,2 Specifically, the acute initiation of neurovascular remodeling, stimulating the proliferation and growth of existing arteries/arterioles (ie, arteriogenesis) or new capillaries (ie, angiogenesis), may be critical to facilitate the survival and restructuring of neural tissue, which ultimately may contribute to functional recovery at later stages.3 A variety of vascular imaging strategies is available for in vivo detection, characterization, and quantification of these processes, which can significantly aid in elucidation of the role of neurovascular remodeling after stroke, as discussed in this review. Because most of the described imaging modalities are present in experimental and clinical settings, this raises significant opportunities for translational studies. Growth and remodeling of existing and nascent vascular networks in health and disease involve a number of critical steps that have been comprehensively described by Risau4 and Carmeliet.5 Arteriogenesis involves adaptive growth and proliferation of preexisting (collateral) arteries and arterioles in response to increase in intravascular shear forces. The increased shear stress leads to upregulation of cell adhesion molecules, followed by accumulation of monocytes and other leukocytes that release cytokines and growth factors around the proliferating and maturating arteries.6 Although arteriogenic growth of collateral …

Deep learning (DL) used for discriminative tasks in ophthalmology, such as diagnosing diabetic retinopathy or age-related macular degeneration (AMD), requires large image data sets graded by human experts to train deep convolutional neural networks (DCNNs). In contrast, generative DL techniques could synthesize large new data sets of artificial retina images with different stages of AMD. Such images could enhance existing data sets of common and rare ophthalmic diseases without concern for personally identifying information to assist medical education of students, residents, and retinal specialists, as well as for training new DL diagnostic models for which extensive data sets from large clinical trials of expertly graded images may not exist. To develop DL techniques for synthesizing high-resolution realistic fundus images serving as proxy data sets for use by retinal specialists and DL machines. Generative adversarial networks were trained on 133 821 color fundus images from 4613 study participants from the Age-Related Eye Disease Study (AREDS), generating synthetic fundus images with and without AMD. We compared retinal specialists' ability to diagnose AMD on both real and synthetic images, asking them to assess image gradability and testing their ability to discern real from synthetic images. The performance of AMD diagnostic DCNNs (referable vs not referable AMD) trained on either all-real vs all-synthetic data sets was compared. Accuracy of 2 retinal specialists (T.Y.A.L. and K.D.P.) for diagnosing and distinguishing AMD on real vs synthetic images and diagnostic performance (area under the curve) of DL algorithms trained on synthetic vs real images. The diagnostic accuracy of 2 retinal specialists on real vs synthetic images was similar. The accuracy of diagnosis as referable vs nonreferable AMD compared with certified human graders for retinal specialist 1 was 84.54% (error margin, 4.06%) on real images vs 84.12% (error margin, 4.16%) on synthetic images and for retinal specialist 2 was 89.47% (error margin, 3.45%) on real images vs 89.19% (error margin, 3.54%) on synthetic images. Retinal specialists could not distinguish real from synthetic images, with an accuracy of 59.50% (error margin, 3.93%) for retinal specialist 1 and 53.67% (error margin, 3.99%) for retinal specialist 2. The DCNNs trained on real data showed an area under the curve of 0.9706 (error margin, 0.0029), and those trained on synthetic data showed an area under the curve of 0.9235 (error margin, 0.0045). Deep learning-synthesized images appeared to be realistic to retinal specialists, and DCNNs achieved diagnostic performance on synthetic data close to that for real images, suggesting that DL generative techniques hold promise for training humans and machines.

Cardiovascular magnetic resonance (MR) is emerging as a multipurpose imaging modality for the assessment of cardiovascular disease in general and ischemic heart disease in particular. Currently, the pace of innovation is rapid, and the modality is changing from one that is used primarily as a research tool to one that is increasingly used in routine clinical practice. The process of innovation includes not only improvements in scanner hardware, such as coil and gradient technology, and the development of new contrast agents but also the development of novel pulse sequences. The concept of the pulse sequence, in which programming changes at the scanner can lead to fundamental changes in activating tissue, is unique to MR and gives this modality the potential to assess a vast number of biological parameters. Cardiovascular MR promises to play an important clinical and investigational role in both vascular and cardiac systems. Current and potential future applications of cardiovascular MR will be discussed with a particular focus on ischemic heart disease. Multidetector-row computed tomography, another promising and complementary noninvasive imaging technology, will be discussed briefly in relation to cardiovascular MR for the assessment of atherothrombotic disease. ### Nomenclature and Evolving Imaging Assessment Atherothrombosis is a systemic or multiterritory arterial disease that primarily affects the large- and medium systemic arteries, including the aorta and the carotid, coronary, and peripheral arteries. Although the epicardial coronary arteries appear to be the most susceptible to atherothrombosis,1,2 intramyocardial arteries are relatively resistant. The concept of multiterritory atherothrombosis has been addressed in 2 large studies of symptomatic patients that showed that at entry into the studies, 3% to 8% had symptomatic atherothrombotic disease in all 3 main arterial districts and 23% to 32% had disease in 2 districts.3,4 From a structural point of view, the 4 main components of the atherothrombotic plaques are as follows: (1) fibrocellular, …

Coronary Artery Aneurysms After Kawasaki Disease: Understanding the Pathology

Imaging inflammation in large intracranial artery pathology may play an important role in the diagnosis of and risk stratification for a variety of cerebrovascular diseases. Looking beyond the lumen has already generated widespread excitement in the stroke community, and the potential to unveil molecular processes in the vessel wall is a natural evolution to develop a more comprehensive understanding of the pathogenesis of diseases, such as ICAD and brain aneurysms.

Further characterizing the physiological process of posterior globe flattening in spaceflight associated neuro-ocular syndrome with generative adversarial networks.

Bystander Killing of Malignant Glioma by Bone Marrow–derived Tumor-Infiltrating Progenitor Cells Expressing a Suicide Gene

Limited amino acid availability for positron emission tomography (PET) imaging hinders therapeutic decision-making for gliomas without typical high-grade imaging features. To address this gap, we evaluated a generative artificial intelligence (AI) approach for creating synthetic O-(2-18F-fluoroethyl)-l-tyrosine ([18F]FET)-PET and predicting high [18F]FET uptake from magnetic resonance imaging (MRI). We trained a deep learning (DL)-based model to segment tumors in MRI, extracted radiomic features using the PythonPyRadiomics package, andutilized a Random Forest classifier to predict high[18F]FET uptake. To generate [18F]FET-PET images, we employed a generative adversarial network framework and utilized a split-input fusion module for processing different MRI sequences through feature extraction, concatenation, and self-attention. We included magnetic resonance imaging (MRI) and PET images from 215 studies for the hotspot classification and 211 studies for the synthetic PET generation task. The top-performing radiomic features achieved 80% accuracy for hotspot prediction. From the synthetic [18F]FET-PET, 85% were classified as clinically useful by senior physicians. Peak signal-to-noise ratio analysis indicated high signal fidelity with a peak at 40 dB, while structural similarity index values showed structural congruence. Root mean square error analysis demonstrated lower values below 5.6. Most visual information fidelity scores ranged between 0.6 and 0.7. This indicates that synthetic PET images retain the essential information required for clinical assessment and diagnosis. For the first time, we demonstrate that predicting high [18F]FET uptake and generating synthetic PET images from preoperative MRI in lower-grade and high-grade glioma are feasible. Advanced MRI modalities and other generative AI models will be used to improve the algorithm further in future studies.

Focus issue: Artificial intelligence in medical physics.

Positron emission tomography (PET) of the brain is an important problem-solving tool in pediatric neuroimaging, neurology, and neurosurgery. Fluorine 18 fluorodeoxyglucose (FDG) PET or dual-modality PET and computed tomographic (CT) imaging (PET/CT), with magnetic resonance (MR) imaging correlation, can be used to evaluate childhood epilepsy and pediatric brain tumors, areas in which PET adds value in patient management. FDG PET has been widely used in pediatric temporal lobe epilepsy, most commonly manifesting as mesial temporal sclerosis, which demonstrates hypometabolism at interictal PET and hypermetabolism during seizures. Recently, FDG PET has shown added value for patients with extratemporal epilepsy, in whom FDG PET can help identify cortical foci of interictal hypometabolism that are undetectable or difficult to detect with MR imaging. These findings can then guide additional investigations and surgery. FDG PET also enhances medical decision making in children with brain tumors, in whom FDG PET can be used to (a) improve the diagnostic yield of stereotactic biopsies by detecting metabolically active areas of tumor, (b) help guide the surgeon in achieving total tumor resection, and (c) increase detection of residual or recurrent tumor. Technologic advances in the past decade have allowed fusion of PET and MR images, combining the high resolution of MR imaging with the low-resolution functional capability of PET. As dual-modality integrated PET/MR imaging systems become available, CT coregistration for PET can be eliminated, thus reducing patient radiation exposure. Increasing familiarity with normal and abnormal appearances of FDG PET brain images correlated with MR images can enhance diagnostic yield and improve the care of children with epilepsy and brain tumors.

Reciprocal assistance of intravascular imaging in three-dimensional stent reconstruction: Using cross-modal translation based on disentanglement representation.

Recently deep-learning (DL) techniques have been widely adopted in side-channel power analysis. A DL-assisted SCA generally consists of two phases: a deep neural network (DNN) training phase and a follow-on attack phase using the trained DNN. However, currently the two phases are not well aligned, as there is no conclusion on what metric used in the training can result in the most effective attack in the second phase. When traditional loss functions such as negative log-likelihood (NLL) are used in training a DNN, the trained model does not yield optimal follow-on attack. Recently some information theoretical SCA leakage metrics are proposed, either as the validation metric to stop the DNN training with traditional loss functions, or as both the validation metric and the training loss function. None of those proposed metrics, however, directly measures the SCA effectiveness. We propose to conduct DNN training directly with a common SCA effectiveness metric, Guessing Entropy (GE). We overcome the prior practical difficulty of using GE in DNN training by utilizing the GEEA estimation algorithm introduced in CHES 2020. We show that using GEEA as either the validation metric or the loss function produces DNN models that lead to much more effective follow-on attacks. Our work consolidates the DL-assisted SCA framework with a consistent metric, which shows great potential to be adopted as the universal SCA-oriented DNN training framework.

Hierarchical feature representation and multimodal fusion with deep learning for AD/MCI diagnosis