The new era of artificial intelligence in neuroradiology: current research and promising tools.

Radiology has a number of characteristics that make it an especially suitable medical discipline for early artificial intelligence (AI) adoption. These include having a well-established digital workflow, standardized protocols for image storage, and numerous well-defined interpretive activities. The more than 200 commercial radiologic AI-based products recently approved by the Food and Drug Administration (FDA) to assist radiologists in a number of narrow image-analysis tasks such as image enhancement, workflow triage, and quantification, corroborate this observation. However, in order to leverage AI to boost efficacy and efficiency, and to overcome substantial obstacles to widespread successful clinical use of these products, radiologists should become familiarized with the emerging applications in their particular areas of expertise. In light of this, in this article we survey the existing literature on the application of AI-based techniques in neuroradiology, focusing on conditions such as vascular diseases, epilepsy, and demyelinating and neurodegenerative conditions. We also introduce some of the algorithms behind the applications, briefly discuss a few of the challenges of generalization in the use of AI models in neuroradiology, and skate over the most relevant commercially available solutions adopted in clinical practice. If well designed, AI algorithms have the potential to radically improve radiology, strengthening image analysis, enhancing the value of quantitative imaging techniques, and mitigating diagnostic errors.

A radiologia tem uma série de características que a torna uma disciplina médica especialmente adequada à adoção precoce da inteligência artificial (IA), incluindo um fluxo de trabalho digital bem estabelecido, protocolos padronizados para armazenamento de imagens e inúmeras atividades interpretativas bem definidas. Tal adequação é corroborada pelos mais de 200 produtos radiológicos comerciais baseados em IA recentemente aprovados pelo Food and Drug Administration (FDA) para auxiliar os radiologistas em uma série de tarefas restritas de análise de imagens, como quantificação, triagem de fluxo de trabalho e aprimoramento da qualidade das imagens. Entretanto, para o aumento da eficácia e eficiência da IA, além de uma utilização clínica bem-sucedida dos produtos que utilizam essa tecnologia, os radiologistas devem estar atualizados com as aplicações em suas áreas específicas de atuação. Assim, neste artigo, pesquisamos na literatura existente aplicações baseadas em IA em neurorradiologia, mais especificamente em condições como doenças vasculares, epilepsia, condições desmielinizantes e neurodegenerativas. Também abordamos os principais algoritmos por trás de tais aplicações, discutimos alguns dos desafios na generalização no uso desses modelos e introduzimos as soluções comercialmente disponíveis mais relevantes adotadas na prática clínica. Se cautelosamente desenvolvidos, os algoritmos de IA têm o potencial de melhorar radicalmente a radiologia, aperfeiçoando a análise de imagens, aumentando o valor das técnicas de imagem quantitativas e mitigando erros de diagnóstico.

Point-of-care AI-assisted stepwise ultrasound pneumothorax diagnosis.

Objective. Ultrasound is extensively utilized as a convenient and cost-effective method in emergency situations. Unfortunately, the limited availability of skilled clinicians in emergency hinders the wider adoption of point-of-care ultrasound. To overcome this challenge, this paper aims to aid less experienced healthcare providers in emergency lung ultrasound scans.Approach. To assist healthcare providers, it is important to have a comprehensive model that can automatically guide the entire process of lung ultrasound based on the clinician's workflow. In this paper, we propose a framework for diagnosing pneumothorax using artificial intelligence (AI) assistance. Specifically, the proposed framework for lung ultrasound scan follows the steps taken by skilled physicians. It begins with finding the appropriate transducer position on the chest to locate the pleural line accurately in B-mode. The next step involves acquiring temporal M-mode data to determine the presence of lung sliding, a crucial indicator for pneumothorax. To mimic the sequential process of clinicians, two DL models were developed. The first model focuses on quality assurance (QA) and regression of the pleural line region-of-interest, while the second model classifies lung sliding. To achieve the inference on a mobile device, a size of EfficientNet-Lite0 model was further reduced to have fewer than 3 million parameters.Main results. The results showed that both the QA and lung sliding classification models achieved over 95% in area under the receiver operating characteristic (AUC), while the ROI performance reached 89% in the dice similarity coefficient. The entire stepwise pipeline was simulated using retrospective data, yielding an AUC of 89%.Significance. The step-wise AI framework for the pneumothorax diagnosis with QA offers an intelligible guide for each clinical workflow, which achieved significantly high precision and real-time inferences.

Quantification of the Thoracic Aorta and Detection of Aneurysm at CT: Development and Validation of a Fully Automatic Methodology.

Purpose: To develop and validate a deep learning-based system that predicts the largest ascending and descending aortic diameters at chest CT through automatic thoracic aortic segmentation and identifies aneurysms in each segment. Materials and Methods: In this retrospective study conducted from July 2019 to February 2021, a U-Net and a postprocessing algorithm for thoracic aortic segmentation and measurement were developed by using a dataset (dataset A) that included 315 CT studies split into training, hyperparameter-tuning, and testing sets. The U-Net and postprocessing algorithm were associated with a Digital Imaging and Communications in Medicine series filter and visualization interface and were further validated by using a dataset (dataset B) that included 1400 routine CT studies. In dataset B, system-predicted measurements were compared with annotations made by two independent readers as well as radiology reports to evaluate system performance. Results: In dataset B, the mean absolute error between the automatic and reader-measured diameters was equal to or less than 0.27 cm for both the ascending aorta and the descending aorta. The intraclass correlation coefficients (ICCs) were greater than 0.80 for the ascending aorta and equal to or greater than 0.70 for the descending aorta, and the ICCs between readers were 0.91 (95% CI: 0.90, 0.92) and 0.82 (95% CI: 0.80, 0.84), respectively. Aneurysm detection accuracy was 88% (95% CI: 86, 90) and 81% (95% CI: 79, 83) compared with reader 1 and 90% (95% CI: 88, 91) and 82% (95% CI: 80, 84) compared with reader 2 for the ascending aorta and descending aorta, respectively. Conclusion: Thoracic aortic aneurysms were accurately predicted at CT by using deep learning.Keywords: Aorta, Convolutional Neural Network, Machine Learning, CT, Thorax, AneurysmsSupplemental material is available for this article.© RSNA, 2022.

Longitudinal whole-brain analysis of multi-subject diffusion data in diffuse axonal injury.

BACKGROUND: Diffuse axonal injury occurs with high acceleration and deceleration forces in traumatic brain injury (TBI). This lesion leads to disarrangement of the neuronal network, which can result in some degree of deficiency. The Extended Glasgow Outcome Scale (GOS-E) is the primary outcome instrument for the evaluation of TBI victims. Diffusion tensor imaging (DTI) assesses white matter (WM) microstructure based on the displacement distribution of water molecules. OBJECTIVE: To investigate WM microstructure within the first year after TBI using DTI, the patient's clinical outcomes, and associations. METHODS: We scanned 20 moderate and severe TBI victims at 2 months and 1 year after the event. Imaging processing was done with the FMRIB software library; we used the tract-based spatial statistics software yielding fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD) for statistical analyses. We computed the average difference between the two measures across subjects and performed a one-sample t-test and threshold-free cluster enhancement, using a corrected p-value < 0.05. Clinical outcomes were evaluated with the GOS-E. We tested for associations between outcome measures and significant mean FA clusters. RESULTS: Significant clusters of altered FA were identified anatomically using the JHU WM atlas. We found increasing spotted areas of FA with time in the right brain hemisphere and left cerebellum. Extensive regions of increased MD, RD, and AD were observed. Patients presented an excellent overall recovery. CONCLUSIONS: There were no associations between FA and outcome scores, but we cannot exclude the existence of a small to moderate association.

Development and clinical application of a deep learning model to identify acute infarct on magnetic resonance imaging.

Stroke is a leading cause of death and disability. The ability to quickly identify the presence of acute infarct and quantify the volume on magnetic resonance imaging (MRI) has important treatment implications. We developed a machine learning model that used the apparent diffusion coefficient and diffusion weighted imaging series. It was trained on 6,657 MRI studies from Massachusetts General Hospital (MGH; Boston, USA). All studies were labelled positive or negative for infarct (classification annotation) with 377 having the region of interest outlined (segmentation annotation). The different annotation types facilitated training on more studies while not requiring the extensive time to manually segment every study. We initially validated the model on studies sequestered from the training set. We then tested the model on studies from three clinical scenarios: consecutive stroke team activations for 6-months at MGH, consecutive stroke team activations for 6-months at a hospital that did not provide training data (Brigham and Women's Hospital [BWH]; Boston, USA), and an international site (Diagnósticos da América SA [DASA]; Brazil). The model results were compared to radiologist ground truth interpretations. The model performed better when trained on classification and segmentation annotations (area under the receiver operating curve [AUROC] 0.995 [95% CI 0.992-0.998] and median Dice coefficient for segmentation overlap of 0.797 [IQR 0.642-0.861]) compared to segmentation annotations alone (AUROC 0.982 [95% CI 0.972-0.990] and Dice coefficient 0.776 [IQR 0.584-0.857]). The model accurately identified infarcts for MGH stroke team activations (AUROC 0.964 [95% CI 0.943-0.982], 381 studies), BWH stroke team activations (AUROC 0.981 [95% CI 0.966-0.993], 247 studies), and at DASA (AUROC 0.998 [95% CI 0.993-1.000], 171 studies). The model accurately segmented infarcts with Pearson correlation comparing model output and ground truth volumes between 0.968 and 0.986 for the three scenarios. Acute infarct can be accurately detected and segmented on MRI in real-world clinical scenarios using a machine learning model.

Longitudinal assessment of magnetization transfer ratio, brain volume, and cognitive functions in diffuse axonal injury.

BACKGROUND: Diffuse axonal injury (DAI) is a frequent mechanism of traumatic brain injury (TBI) that triggers a sequence of parenchymal changes that progresses from focal axonal shear injuries up to inflammatory response and delayed axonal disconnection. OBJECTIVE: The main purpose of this study is to evaluate changes in the axonal/myelinic content and the brain volume up to 12 months after TBI and to correlate these changes with neuropsychological results. METHODS: Patients with DAI (n = 25) were scanned at three time points after trauma (2, 6, and 12 months), and the total brain volume (TBV), gray matter volume, and white matter volume (WMV) were calculated in each time point. The magnetization transfer ratio (MTR) for the total brain (TB MTR), gray matter (GM MTR), and white matter (WM MTR) was also quantified. In addition, Hopkins verbal learning test (HVLT), Trail Making Test (TMT), and Rey-Osterrieth Complex Figure test were performed at 6 and 12 months after the trauma. RESULTS: There was a significant reduction in the mean TBV, WMV, TB MTR, GM MTR, and WM MTR between time points 1 and 3 (p < .05). There was also a significant difference in HVLT-immediate, TMT-A, and TMT-B scores between time points 2 and 3. The MTR decline correlated more with the cognitive dysfunction than the volume reduction. CONCLUSION: A progressive axonal/myelinic rarefaction and volume loss were characterized, especially in the white matter (WM) up to 1 year after the trauma. Despite that, specific neuropsychological tests revealed that patients' episodic verbal memory, attention, and executive function improved during the study. The current findings may be valuable in developing long-term TBI rehabilitation management programs.

DTI-derived parameters differ between moderate and severe traumatic brain injury and its association with psychiatric scores.

BACKGROUND AND AIM: Diffusion tensor imaging (DTI) parameters in the corpus callosum have been suggested to be a biomarker for prognostic outcomes in individuals with diffuse axonal injury (DAI). However, differences between the DTI parameters on moderate and severe trauma in DAI over time are still unclear. A secondary goal was to study the association between the changes in the DTI parameters, anxiety, and depressive scores in DAI over time. METHODS: Twenty subjects were recruited from a neurological outpatient clinic and evaluated at 2, 6, and 12 months after the brain injury and compared to matched age and sex healthy controls regarding the DTI parameters in the corpus callosum. State-Trace Anxiety Inventory and Beck Depression Inventory were used to assess psychiatric outcomes in the TBI group over time. RESULTS: Differences were observed in the fractional anisotropy and mean diffusivity of the genu, body, and splenium of the corpus callosum between DAI and controls (p < 0.02). Differences in both parameters in the genu of the corpus callosum were also detected between patients with moderate and severe DAI (p < 0.05). There was an increase in the mean diffusivity values and the fractional anisotropy decrease in the DAI group over time (p < 0.02). There was no significant correlation between changes in the fractional anisotropy and mean diffusivity across the study and psychiatric outcomes in DAI. CONCLUSION: DTI parameters, specifically the mean diffusivity in the corpus callosum, may provide reliable characterization and quantification of differences determined by the brain injury severity. No correlation was observed with DAI parameters and the psychiatric outcome scores.

A Deep Learning-based Model for Detecting Abnormalities on Brain MR Images for Triaging: Preliminary Results from a Multisite Experience.

PURPOSE: To develop a deep learning model for detecting brain abnormalities on MR images. MATERIALS AND METHODS: In this retrospective study, a deep learning approach using T2-weighted fluid-attenuated inversion recovery images was developed to classify brain MRI findings as "likely normal" or "likely abnormal." A convolutional neural network model was trained on a large, heterogeneous dataset collected from two different continents and covering a broad panel of pathologic conditions, including neoplasms, hemorrhages, infarcts, and others. Three datasets were used. Dataset A consisted of 2839 patients, dataset B consisted of 6442 patients, and dataset C consisted of 1489 patients and was only used for testing. Datasets A and B were split into training, validation, and test sets. A total of three models were trained: model A (using only dataset A), model B (using only dataset B), and model A + B (using training datasets from A and B). All three models were tested on subsets from dataset A, dataset B, and dataset C separately. The evaluation was performed by using annotations based on the images, as well as labels based on the radiology reports. RESULTS: Model A trained on dataset A from one institution and tested on dataset C from another institution reached an F1 score of 0.72 (95% CI: 0.70, 0.74) and an area under the receiver operating characteristic curve of 0.78 (95% CI: 0.75, 0.80) when compared with findings from the radiology reports. CONCLUSION: The model shows relatively good performance for differentiating between likely normal and likely abnormal brain examination findings by using data from different institutions.Keywords: MR-Imaging, Head/Neck, Computer Applications-General (Informatics), Convolutional Neural Network (CNN), Deep Learning Algorithms, Machine Learning Algorithms© RSNA, 2021Supplemental material is available for this article.

Misconceptions in the health technology industry that are delaying the translation of artificial intelligence technology into relevant clinical applications.

There is great optimism that artificial intelligence (AI), as it disrupts the medical world, will provide considerable improvements in all areas of health care, from diagnosis to treatment. In addition, there is considerable evidence that AI algorithms have surpassed human performance in various tasks, such as analyzing medical images, as well as correlating symptoms and biomarkers with the diagnosis and prognosis of diseases. However, the mismatch between the performance of AI-based software and its clinical usefulness is still a major obstacle to its widespread acceptance and use by the medical community. In this article, three fundamental concepts observed in the health technology industry are highlighted as possible causative factors for this gap and might serve as a starting point for further evaluation of the structure of AI companies and of the status quo.

Há uma grande expectativa de que a inteligência artificial (IA), ao transformar a medicina, determine melhoras relevantes em todas as áreas da assistência médica, desde o diagnóstico até o tratamento. Simultaneamente, há evidências de que algoritmos baseados em IA já ultrapassaram o desempenho do ser humano em diversas atividades, como, por exemplo, na análise de imagens médicas ou na associação entre sintomas e biomarcadores com o diagnóstico e prognóstico de doenças. No entanto, a defasagem entre o potencial de desempenho das ferramentas ou aplicativos médicos que utilizam IA e sua relevância clínica prejudica bastante a utilização em larga escala desses programas de computadores. Neste artigo, três conceitos básicos da indústria de tecnologia da saúde são sugeridos como possíveis fatores causais para essa dissincronia entre desempenho e utilidade. Tal discussão pode servir como ponto de partida para uma avaliação mais profunda sobre a estrutura e status quo da indústria médica tecnológica atual.

Misconceptions in the health technology industry that are delaying the translation of artificial intelligence technology into relevant clinical applications / Conceitos enviesados na indústria de tecnologia da saúde que retardam a tradução da inteligência artificial em ferramentas clínicas relevantes

Abstract There is great optimism that artificial intelligence (AI), as it disrupts the medical world, will provide considerable improvements in all areas of health care, from diagnosis to treatment. In addition, there is considerable evidence that AI algorithms have surpassed human performance in various tasks, such as analyzing medical images, as well as correlating symptoms and biomarkers with the diagnosis and prognosis of diseases. However, the mismatch between the performance of AI-based software and its clinical usefulness is still a major obstacle to its widespread acceptance and use by the medical community. In this article, three fundamental concepts observed in the health technology industry are highlighted as possible causative factors for this gap and might serve as a starting point for further evaluation of the structure of AI companies and of the status quo.

Resumo Há uma grande expectativa de que a inteligência artificial (IA), ao transformar a medicina, determine melhoras relevantes em todas as áreas da assistência médica, desde o diagnóstico até o tratamento. Simultaneamente, há evidências de que algoritmos baseados em IA já ultrapassaram o desempenho do ser humano em diversas atividades, como, por exemplo, na análise de imagens médicas ou na associação entre sintomas e biomarcadores com o diagnóstico e prognóstico de doenças. No entanto, a defasagem entre o potencial de desempenho das ferramentas ou aplicativos médicos que utilizam IA e sua relevância clínica prejudica bastante a utilização em larga escala desses programas de computadores. Neste artigo, três conceitos básicos da indústria de tecnologia da saúde são sugeridos como possíveis fatores causais para essa dissincronia entre desempenho e utilidade. Tal discussão pode servir como ponto de partida para uma avaliação mais profunda sobre a estrutura e status quo da indústria médica tecnológica atual.

Dynamic changes in white matter following traumatic brain injury and how diffuse axonal injury relates to cognitive domain.

Objective: The goal is to evaluate longitudinally with diffusion tensor imaging (DTI) the integrity of cerebral white matter in patients with moderate and severe DAI and to correlate the DTI findings with cognitive deficits.Methods: Patients with DAI (n = 20) were scanned at three timepoints (2, 6 and 12 months) after trauma. A healthy control group (n = 20) was evaluated once with the same high-field MRI scanner. The corpus callosum (CC) and the bilateral superior longitudinal fascicles (SLFs) were assessed by deterministic tractography with ExploreDTI. A neuropschychological evaluation was also performed.Results: The CC and both SLFs demonstrated various microstructural abnormalities in between-groups comparisons. All DTI parameters demonstrated changes across time in the body of the CC, while FA (fractional anisotropy) increases were seen on both SLFs. In the splenium of the CC, progressive changes in the mean diffusivity (MD) and axial diffusivity (AD) were also observed. There was an improvement in attention and memory along time. Remarkably, DTI parameters demonstrated several correlations with the cognitive domains.Conclusions: Our findings suggest that microstructural changes in the white matter are dynamic and may be detectable by DTI throughout the first year after trauma. Likewise, patients also demonstrated improvement in some cognitive skills.

Viscoelasticity of children and adolescent brains through MR elastography.

Magnetic Resonance Elastography (MRE) is an elasticity imaging technique that allows a safe, fast, and non-invasive evaluation of the mechanical properties of biological tissues in vivo. Since mechanical properties reflect a tissue's composition and arrangement, MRE is a powerful tool for the investigation of the microstructural changes that take place in the brain during childhood and adolescence. The goal of this study was to evaluate the viscoelastic properties of the brain in a population of healthy children and adolescents in order to identify potential age and sex dependencies. We hypothesize that because of myelination, age dependent changes in the mechanical properties of the brain will occur during childhood and adolescence. Our sample consisted of 26 healthy individuals (13 M, 13 F) with age that ranged from 7-17 years (mean: 11.9 years). We performed multifrequency MRE at 40, 60, and 80 Hz actuation frequencies to acquire the complex-valued shear modulus G = G' + iG″ with the fundamental MRE parameters being the storage modulus (G'), the loss modulus (G″), and the magnitude of complex-valued shear modulus (|G|). We fitted a springpot model to these frequency-dependent MRE parameters in order to obtain the parameter α, which is related to tissue's microstructure, and the elasticity parameter k, which was converted to a shear modulus parameter (µ) through viscosity (η). We observed no statistically significant variation in the parameter µ, but a significant increase of the microstructural parameter α of the white matter with increasing age (p < 0.05). Therefore, our MRE results suggest that subtle microstructural changes such as neural tissue's enhanced alignment and geometrical reorganization during childhood and adolescence could result in significant biomechanical changes. In line with previously reported MRE data for adults, we also report significantly higher shear modulus (µ) for female brains when compared to males (p < 0.05). The data presented here can serve as a clinical baseline in the analysis of the pediatric and adolescent brain's viscoelasticity over this age span, as well as extending our understanding of the biomechanics of brain development.

Ultra-Low-Dose 18F-Florbetaben Amyloid PET Imaging Using Deep Learning with Multi-Contrast MRI Inputs.

Purpose To reduce radiotracer requirements for amyloid PET/MRI without sacrificing diagnostic quality by using deep learning methods. Materials and Methods Forty data sets from 39 patients (mean age ± standard deviation [SD], 67 years ± 8), including 16 male patients and 23 female patients (mean age, 66 years ± 6 and 68 years ± 9, respectively), who underwent simultaneous amyloid (fluorine 18 [18F]-florbetaben) PET/MRI examinations were acquired from March 2016 through October 2017 and retrospectively analyzed. One hundredth of the raw list-mode PET data were randomly chosen to simulate a low-dose (1%) acquisition. Convolutional neural networks were implemented with low-dose PET and multiple MR images (PET-plus-MR model) or with low-dose PET alone (PET-only) as inputs to predict full-dose PET images. Quality of the synthesized images was evaluated while Bland-Altman plots assessed the agreement of regional standard uptake value ratios (SUVRs) between image types. Two readers scored image quality on a five-point scale (5 = excellent) and determined amyloid status (positive or negative). Statistical analyses were carried out to assess the difference of image quality metrics and reader agreement and to determine confidence intervals (CIs) for reading results. Results The synthesized images (especially from the PET-plus-MR model) showed marked improvement on all quality metrics compared with the low-dose image. All PET-plus-MR images scored 3 or higher, with proportions of images rated greater than 3 similar to those for the full-dose images (-10% difference [eight of 80 readings], 95% CI: -15%, -5%). Accuracy for amyloid status was high (71 of 80 readings [89%]) and similar to intrareader reproducibility of full-dose images (73 of 80 [91%]). The PET-plus-MR model also had the smallest mean and variance for SUVR difference to full-dose images. Conclusion Simultaneously acquired MRI and ultra-low-dose PET data can be used to synthesize full-dose-like amyloid PET images. © RSNA, 2018 Online supplemental material is available for this article. See also the editorial by Catana in this issue.

Longitudinal changes in brain volumetry and cognitive functions after moderate and severe diffuse axonal injury.

BACKGROUND AND OBJECTIVE: Diffuse axonal injury (DAI) induces a long-term process of brain atrophy and cognitive deficits. The goal of this study was to determine whether there are correlations between brain volume loss, microhaemorrhage load (MHL) and neuropsychological performance during the first year after DAI. METHODS: Twenty-four patients with moderate or severe DAI were evaluated at 2, 6 and 12 months post-injury. MHL was evaluated at 3 months, and brain volumetry was evaluated at 3, 6 and 12 months. The trail making test (TMT) was used to evaluate executive function (EF), and the Hopkins verbal learning test (HVLT) was used to evaluate episodic verbal memory (EVM) at 6 and 12 months. RESULTS: There were significant white matter volume (WMV), subcortical grey matter volume and total brain volume (TBV) reductions during the study period (p < 0.05). MHL was correlated only with WMV reduction. EF and EVM were not correlated with MHL but were, in part, correlated with WMV and TBV reductions. CONCLUSIONS: Our findings suggest that MHL may be a predictor of WMV reduction but cannot predict EF or EVM in DAI. Brain atrophy progresses over time, but patients showed better EF and EVM in some of the tests, which could be due to neuroplasticity.