Artificial intelligence for cardiovascular disease risk assessment in personalised framework: a scoping review.

Background: The field of precision medicine endeavors to transform the healthcare industry by advancing individualised strategies for diagnosis, treatment modalities, and predictive assessments. This is achieved by utilizing extensive multidimensional biological datasets encompassing diverse components, such as an individual's genetic makeup, functional attributes, and environmental influences. Artificial intelligence (AI) systems, namely machine learning (ML) and deep learning (DL), have exhibited remarkable efficacy in predicting the potential occurrence of specific cancers and cardiovascular diseases (CVD). Methods: We conducted a comprehensive scoping review guided by the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) framework. Our search strategy involved combining key terms related to CVD and AI using the Boolean operator AND. In August 2023, we conducted an extensive search across reputable scholarly databases including Google Scholar, PubMed, IEEE Xplore, ScienceDirect, Web of Science, and arXiv to gather relevant academic literature on personalised medicine for CVD. Subsequently, in January 2024, we extended our search to include internet search engines such as Google and various CVD websites. These searches were further updated in March 2024. Additionally, we reviewed the reference lists of the final selected research articles to identify any additional relevant literature. Findings: A total of 2307 records were identified during the process of conducting the study, consisting of 564 entries from external sites like arXiv and 1743 records found through database searching. After 430 duplicate articles were eliminated, 1877 items that remained were screened for relevancy. In this stage, 1241 articles remained for additional review after 158 irrelevant articles and 478 articles with insufficient data were removed. 355 articles were eliminated for being inaccessible, 726 for being written in a language other than English, and 281 for not having undergone peer review. Consequently, 121 studies were deemed suitable for inclusion in the qualitative synthesis. At the intersection of CVD, AI, and precision medicine, we found important scientific findings in our scoping review. Intricate pattern extraction from large, complicated genetic datasets is a skill that AI algorithms excel at, allowing for accurate disease diagnosis and CVD risk prediction. Furthermore, these investigations have uncovered unique genetic biomarkers linked to CVD, providing insight into the workings of the disease and possible treatment avenues. The construction of more precise predictive models and personalised treatment plans based on the genetic profiles of individual patients has been made possible by the revolutionary advancement of CVD risk assessment through the integration of AI and genomics. Interpretation: The systematic methodology employed ensured the thorough examination of available literature and the inclusion of relevant studies, contributing to the robustness and reliability of the study's findings. Our analysis stresses a crucial point in terms of the adaptability and versatility of AI solutions. AI algorithms designed in non-CVD domains such as in oncology, often include ideas and tactics that might be modified to address cardiovascular problems. Funding: No funding received.

Functional Connectivity Changes on Resting-State fMRI after Mild Traumatic Brain Injury: A Systematic Review.

BACKGROUND: Mild traumatic brain injury is theorized to cause widespread functional changes to the brain. Resting-state fMRI may be able to measure functional connectivity changes after traumatic brain injury, but resting-state fMRI studies are heterogeneous, using numerous techniques to study ROIs across various resting-state networks. PURPOSE: We systematically reviewed the literature to ascertain whether adult patients who have experienced mild traumatic brain injury show consistent functional connectivity changes on resting-state -fMRI, compared with healthy patients. DATA SOURCES: We used 5 databases (PubMed, EMBASE, Cochrane Central, Scopus, Web of Science). STUDY SELECTION: Five databases (PubMed, EMBASE, Cochrane Central, Scopus, and Web of Science) were searched for research published since 2010. Search strategies used keywords of "functional MR imaging" and "mild traumatic brain injury" as well as related terms. All results were screened at the abstract and title levels by 4 reviewers according to predefined inclusion and exclusion criteria. For full-text inclusion, each study was evaluated independently by 2 reviewers, with discordant screening settled by consensus. DATA ANALYSIS: Data regarding article characteristics, cohort demographics, fMRI scan parameters, data analysis processing software, atlas used, data characteristics, and statistical analysis information were extracted. DATA SYNTHESIS: Across 66 studies, 80 areas were analyzed 239 times for at least 1 time point, most commonly using independent component analysis. The most analyzed areas and networks were the whole brain, the default mode network, and the salience network. Reported functional connectivity changes varied, though there may be a slight trend toward decreased whole-brain functional connectivity within 1 month of traumatic brain injury and there may be differences based on the time since injury. LIMITATIONS: Studies of military, sports-related traumatic brain injury, and pediatric patients were excluded. Due to the high number of relevant studies and data heterogeneity, we could not be as granular in the analysis as we would have liked. CONCLUSIONS: Reported functional connectivity changes varied, even within the same region and network, at least partially reflecting differences in technical parameters, preprocessing software, and analysis methods as well as probable differences in individual injury. There is a need for novel rs-fMRI techniques that better capture subject-specific functional connectivity changes.

A Comprehensive and Broad Approach to Resting-State Functional Connectivity in Adult Patients with Mild Traumatic Brain Injury.

BACKGROUND AND PURPOSE: Several recent works using resting-state fMRI suggest possible alterations of resting-state functional connectivity after mild traumatic brain injury. However, the literature is plagued by various analysis approaches and small study cohorts, resulting in an inconsistent array of reported findings. In this study, we aimed to investigate differences in whole-brain resting-state functional connectivity between adult patients with mild traumatic brain injury within 1 month of injury and healthy control subjects using several comprehensive resting-state functional connectivity measurement methods and analyses. MATERIALS AND METHODS: A total of 123 subjects (72 patients with mild traumatic brain injury and 51 healthy controls) were included. A standard fMRI preprocessing pipeline was used. ROI/seed-based analyses were conducted using 4 standard brain parcellation methods, and the independent component analysis method was applied to measure resting-state functional connectivity. The fractional amplitude of low-frequency fluctuations was also measured. Group comparisons were performed on all measurements with appropriate whole-brain multilevel statistical analysis and correction. RESULTS: There were no significant differences in age, sex, education, and hand preference between groups as well as no significant correlation between all measurements and these potential confounders. We found that each resting-state functional connectivity measurement revealed various regions or connections that were different between groups. However, after we corrected for multiple comparisons, the results showed no statistically significant differences between groups in terms of resting-state functional connectivity across methods and analyses. CONCLUSIONS: Although previous studies point to multiple regions and networks as possible mild traumatic brain injury biomarkers, this study shows that the effect of mild injury on brain resting-state functional connectivity has not survived after rigorous statistical correction. A further study using subject-level connectivity analyses may be necessary due to both subtle and variable effects of mild traumatic brain injury on brain functional connectivity across individuals.

Predicting Clinician Fixations on Glaucoma OCT Reports via CNN-Based Saliency Prediction Methods.

Goal: To predict physician fixations specifically on ophthalmology optical coherence tomography (OCT) reports from eye tracking data using CNN based saliency prediction methods in order to aid in the education of ophthalmologists and ophthalmologists-in-training. Methods: Fifteen ophthalmologists were recruited to each examine 20 randomly selected OCT reports and evaluate the likelihood of glaucoma for each report on a scale of 0-100. Eye movements were collected using a Pupil Labs Core eye-tracker. Fixation heat maps were generated using fixation data. Results: A model trained with traditional saliency mapping resulted in a correlation coefficient (CC) value of 0.208, a Normalized Scanpath Saliency (NSS) value of 0.8172, a Kullback-Leibler (KLD) value of 2.573, and a Structural Similarity Index (SSIM) of 0.169. Conclusions: The TranSalNet model was able to predict fixations within certain regions of the OCT report with reasonable accuracy, but more data is needed to improve model accuracy. Future steps include increasing data collection, improving quality of data, and modifying the model architecture.

Airway tree caliber heterogeneity and airflow obstruction among older adults.

Smaller mean airway tree caliber is associated with airflow obstruction and chronic obstructive pulmonary disease (COPD). We investigated whether airway tree caliber heterogeneity was associated with airflow obstruction and COPD. Two community-based cohorts (MESA Lung, CanCOLD) and a longitudinal case-control study of COPD (SPIROMICS) performed spirometry and computed tomography measurements of airway lumen diameters at standard anatomical locations (trachea-to-subsegments) and total lung volume. Percent-predicted airway lumen diameters were calculated using sex-specific reference equations accounting for age, height, and lung volume. The association of airway tree caliber heterogeneity, quantified as the standard deviation (SD) of percent-predicted airway lumen diameters, with baseline forced expired volume in 1-second (FEV1), FEV1/forced vital capacity (FEV1/FVC) and COPD, as well as longitudinal spirometry, were assessed using regression models adjusted for age, sex, height, race-ethnicity, and mean airway tree caliber. Among 2,505 MESA Lung participants (means ± SD age: 69 ± 9 yr; 53% female, mean airway tree caliber: 99 ± 10% predicted, airway tree caliber heterogeneity: 14 ± 5%; median follow-up: 6.1 yr), participants in the highest quartile of airway tree caliber heterogeneity exhibited lower FEV1 (adjusted mean difference: -125 mL, 95%CI: -171,-79), lower FEV1/FVC (adjusted mean difference: -0.01, 95%CI: -0.02,-0.01), and higher odds of COPD (adjusted odds ratio: 1.42, 95%CI: 1.01-2.02) when compared with the lowest quartile, whereas longitudinal changes in FEV1 and FEV1/FVC did not differ significantly. Observations in CanCOLD and SPIROMICS were consistent. Among older adults, airway tree caliber heterogeneity was associated with airflow obstruction and COPD at baseline but was not associated with longitudinal changes in spirometry.NEW & NOTEWORTHY In this study, by leveraging two community-based samples and a case-control study of heavy smokers, we show that among older adults, airway tree caliber heterogeneity quantified by CT is associated with airflow obstruction and COPD independent of age, sex, height, race-ethnicity, and dysanapsis. These observations suggest that airway tree caliber heterogeneity is a structural trait associated with low baseline lung function and normal decline trajectory that is relevant to COPD.

Pulmonary emphysema subtypes defined by unsupervised machine learning on CT scans.

BACKGROUND: Treatment and preventative advances for chronic obstructive pulmonary disease (COPD) have been slow due, in part, to limited subphenotypes. We tested if unsupervised machine learning on CT images would discover CT emphysema subtypes with distinct characteristics, prognoses and genetic associations. METHODS: New CT emphysema subtypes were identified by unsupervised machine learning on only the texture and location of emphysematous regions on CT scans from 2853 participants in the Subpopulations and Intermediate Outcome Measures in COPD Study (SPIROMICS), a COPD case-control study, followed by data reduction. Subtypes were compared with symptoms and physiology among 2949 participants in the population-based Multi-Ethnic Study of Atherosclerosis (MESA) Lung Study and with prognosis among 6658 MESA participants. Associations with genome-wide single-nucleotide-polymorphisms were examined. RESULTS: The algorithm discovered six reproducible (interlearner intraclass correlation coefficient, 0.91-1.00) CT emphysema subtypes. The most common subtype in SPIROMICS, the combined bronchitis-apical subtype, was associated with chronic bronchitis, accelerated lung function decline, hospitalisations, deaths, incident airflow limitation and a gene variant near DRD1, which is implicated in mucin hypersecretion (p=1.1 ×10-8). The second, the diffuse subtype was associated with lower weight, respiratory hospitalisations and deaths, and incident airflow limitation. The third was associated with age only. The fourth and fifth visually resembled combined pulmonary fibrosis emphysema and had distinct symptoms, physiology, prognosis and genetic associations. The sixth visually resembled vanishing lung syndrome. CONCLUSION: Large-scale unsupervised machine learning on CT scans defined six reproducible, familiar CT emphysema subtypes that suggest paths to specific diagnosis and personalised therapies in COPD and pre-COPD.

Engineered bacterial swarm patterns as spatial records of environmental inputs.

A diverse array of bacteria species naturally self-organize into durable macroscale patterns on solid surfaces via swarming motility-a highly coordinated and rapid movement of bacteria powered by flagella. Engineering swarming is an untapped opportunity to increase the scale and robustness of coordinated synthetic microbial systems. Here we engineer Proteus mirabilis, which natively forms centimeter-scale bullseye swarm patterns, to 'write' external inputs into visible spatial records. Specifically, we engineer tunable expression of swarming-related genes that modify pattern features, and we develop quantitative approaches to decoding. Next, we develop a dual-input system that modulates two swarming-related genes simultaneously, and we separately show that growing colonies can record dynamic environmental changes. We decode the resulting multicondition patterns with deep classification and segmentation models. Finally, we engineer a strain that records the presence of aqueous copper. This work creates an approach for building macroscale bacterial recorders, expanding the framework for engineering emergent microbial behaviors.

Cardiac Adipose Tissue Segmentation via Image-Level Annotations.

Automatically identifying the structural substrates underlying cardiac abnormalities can potentially provide real-time guidance for interventional procedures. With the knowledge of cardiac tissue substrates, the treatment of complex arrhythmias such as atrial fibrillation and ventricular tachycardia can be further optimized by detecting arrhythmia substrates to target for treatment (i.e., adipose) and identifying critical structures to avoid. Optical coherence tomography (OCT) is a real-time imaging modality that aids in addressing this need. Existing approaches for cardiac image analysis mainly rely on fully supervised learning techniques, which suffer from the drawback of workload on labor-intensive annotation process of pixel-wise labeling. To lessen the need for pixel-wise labeling, we develop a two-stage deep learning framework for cardiac adipose tissue segmentation using image-level annotations on OCT images of human cardiac substrates. In particular, we integrate class activation mapping with superpixel segmentation to solve the sparse tissue seed challenge raised in cardiac tissue segmentation. Our study bridges the gap between the demand on automatic tissue analysis and the lack of high-quality pixel-wise annotations. To the best of our knowledge, this is the first study that attempts to address cardiac tissue segmentation on OCT images via weakly supervised learning techniques. Within an in-vitro human cardiac OCT dataset, we demonstrate that our weakly supervised approach on image-level annotations achieves comparable performance as fully supervised methods trained on pixel-wise annotations.

Obstructive Sleep Apnea and Longitudinal Changes in Interstitial Lung Imaging and Lung Function: The MESA Study.

Rationale: Obstructive sleep apnea (OSA) has been hypothesized to be a risk factor in interstitial lung disease (ILD) and is associated with radiological markers that may represent the earlier stages of ILD. Prior studies have been limited by their cross-sectional design and potential confounding by body habitus. Objectives: To test the hypothesis that OSA severity is associated with more high-attenuation areas (HAAs) on computed tomography and worse lung function over time among older community-dwelling adults. Methods: We used data from participants in the MESA (Multi-Ethnic Study of Atherosclerosis) who had apnea-hypopnea index (AHI) measured from polysomnography (2010-2013), high attenuation areas (HAAs, -600 to -250 Hounsfield units, n = 784), assessments from exams 5 (2010-2012) and 6 (2016-2018) full-lung computed tomography scans, and spirometry assessments (n = 677). Linear mixed-effects models with random intercept were used to examine associations of OSA severity (i.e., AHI and hypoxic burden) with changes in HAAs, total lung volumes, and forced vital capacity (FVC) between exams 5 and 6. Potential confounders were adjusted for in the model, including age, sex, smoking history, height, and weight. Results: Among those with a higher AHI there were more men and a higher body mass index. Participants with AHI ⩾ 15 events/h and in the highest hypoxic burden quartile each had increases in HAAs of 11.30% (95% confidence interval [CI], 3.74-19.35%) and 9.85% (95% CI, 1.40-19.01%) per 10 years, respectively. There was a more rapid decline in total lung volumes imaged and FVC among those with AHI ⩾ 15 events/h of 220.2 ml (95% CI, 47.8-392.5 ml) and 3.63% (95% CI, 0.43-6.83%) per 10 years, respectively. Conclusions: A greater burden of hypoxia related to obstructive events during sleep was associated with increased lung densities over time and a more rapid decline in lung volumes regardless of body habitus. Our findings suggest OSA may be a contributing factor in the early stages of ILD.

Deep Learning Enables Reduced Gadolinium Dose for Contrast-Enhanced Blood-Brain Barrier Opening.

Focused ultrasound (FUS) can be used to open the blood-brain barrier (BBB), and MRI with contrast agents can detect that opening. However, repeated use of gadolinium-based contrast agents (GBCAs) presents safety concerns to patients. This study is the first to propose the idea of modeling a volume transfer constant (Ktrans) through deep learning to reduce the dosage of contrast agents. The goal of the study is not only to reconstruct artificial intelligence (AI) derived Ktrans images but to also enhance the intensity with low dosage contrast agent T1 weighted MRI scans. We successfully validated this idea through a previous state-of-the-art temporal network algorithm, which focused on extracting time domain features at the voxel level. Then we used a Spatiotemporal Network (ST-Net), composed of a spatiotemporal convolutional neural network (CNN)-based deep learning architecture with the addition of a three-dimensional CNN encoder, to improve the model performance. We tested the ST-Net model on ten datasets of FUS-induced BBB-openings aquired from different sides of the mouse brain. ST-Net successfully detected and enhanced BBB-opening signals without sacrificing spatial domain information. ST-Net was shown to be a promising method of reducing the need of contrast agents for modeling BBB-opening K-trans maps from time-series Dynamic Contrast-Enhanced Magnetic Resonance Imaging (DCE-MRI) scans.

MUC5B, telomere length and longitudinal quantitative interstitial lung changes: the MESA Lung Study.

BACKGROUND: The MUC5B promoter variant (rs35705950) and telomere length are linked to pulmonary fibrosis and CT-based qualitative assessments of interstitial abnormalities, but their associations with longitudinal quantitative changes of the lung interstitium among community-dwelling adults are unknown. METHODS: We used data from participants in the Multi-Ethnic Study of Atherosclerosis with high-attenuation areas (HAAs, Examinations 1-6 (2000-2018)) and MUC5B genotype (n=4552) and telomere length (n=4488) assessments. HAA was defined as the per cent of imaged lung with attenuation of -600 to -250 Hounsfield units. We used linear mixed-effects models to examine associations of MUC5B risk allele (T) and telomere length with longitudinal changes in HAAs. Joint models were used to examine associations of longitudinal changes in HAAs with death and interstitial lung disease (ILD). RESULTS: The MUC5B risk allele (T) was associated with an absolute change in HAAs of 2.60% (95% CI 0.36% to 4.86%) per 10 years overall. This association was stronger among those with a telomere length below an age-adjusted percentile of 5% (p value for interaction=0.008). A 1% increase in HAAs per year was associated with 7% increase in mortality risk (rate ratio (RR)=1.07, 95% CI 1.02 to 1.12) for overall death and 34% increase in ILD (RR=1.34, 95% CI 1.20 to 1.50). Longer baseline telomere length was cross-sectionally associated with less HAAs from baseline scans, but not with longitudinal changes in HAAs. CONCLUSIONS: Longitudinal increases in HAAs were associated with the MUC5B risk allele and a higher risk of death and ILD.

Segmentation with Residual Attention U-Net and an Edge-Enhancement Approach Preserves Cell Shape Features.

The ability to extrapolate gene expression dynamics in living single cells requires robust cell segmentation, and one of the challenges is the amorphous or irregularly shaped cell boundaries. To address this issue, we modified the U-Net architecture to segment cells in fluorescence widefield microscopy images and quantitatively evaluated its performance. We also proposed a novel loss function approach that emphasizes the segmentation accuracy on cell boundaries and encourages shape feature preservation. With a 97% sensitivity, 93% specificity, 91% Jaccard similarity, and 95% Dice coefficient, our proposed method called Residual Attention U-Net with edge-enhancement surpassed the state-of-the-art U-Net in segmentation performance as evaluated by the traditional metrics. More remarkably, the same proposed candidate also performed the best in terms of the preservation of valuable shape features, namely area, eccentricity, major axis length, solidity and orientation. These improvements on shape feature preservation can serve as useful assets for downstream cell tracking and quantification of changes in cell statistics or features over time.

Deep learning of MRI contrast enhancement for mapping cerebral blood volume from single-modal non-contrast scans of aging and Alzheimer's disease brains.

While MRI contrast agents such as those based on Gadolinium are needed for high-resolution mapping of brain metabolism, these contrast agents require intravenous administration, and there are rising concerns over their safety and invasiveness. Furthermore, non-contrast MRI scans are more commonly performed than those with contrast agents and are readily available for analysis in public databases such as the Alzheimer's Disease Neuroimaging Initiative (ADNI). In this article, we hypothesize that a deep learning model, trained using quantitative steady-state contrast-enhanced structural MRI datasets, in mice and humans, can generate contrast-equivalent information from a single non-contrast MRI scan. The model was first trained, optimized, and validated in mice, and was then transferred and adapted to humans. We observe that the model can substitute for Gadolinium-based contrast agents in approximating cerebral blood volume, a quantitative representation of brain activity, at sub-millimeter granularity. Furthermore, we validate the use of our deep-learned prediction maps to identify functional abnormalities in the aging brain using locally obtained MRI scans, and in the brain of patients with Alzheimer's disease using publicly available MRI scans from ADNI. Since it is derived from a commonly-acquired MRI protocol, this framework has the potential for broad clinical utility and can also be applied retrospectively to research scans across a host of neurological/functional diseases.

Quantification of lung ventilation defects on hyperpolarized MRI: The Multi-Ethnic Study of Atherosclerosis (MESA) COPD study.

PURPOSE: To develop an end-to-end deep learning (DL) framework to segment ventilation defects on pulmonary hyperpolarized MRI. MATERIALS AND METHODS: The Multi-Ethnic Study of Atherosclerosis Chronic Obstructive Pulmonary Disease (COPD) study is a nested longitudinal case-control study in older smokers. Between February 2016 and July 2017, 56 participants (age, mean ± SD, 74 ± 8 years; 34 men) underwent same breath-hold proton (1H) and helium (3He) MRI, which were annotated for non-ventilated, hypo-ventilated, and normal-ventilated lungs. In this retrospective DL study, 820 1H and 3He slices from 42/56 (75%) participants were randomly selected for training, with the remaining 14/56 (25%) for test. Full lung masks were segmented using a traditional U-Net on 1H MRI and were imported into a cascaded U-Net, which were used to segment ventilation defects on 3He MRI. Models were trained with conventional data augmentation (DA) and generative adversarial networks (GAN)-DA. RESULTS: Conventional-DA improved 1H and 3He MRI segmentation over the non-DA model (P = 0.007 to 0.03) but GAN-DA did not yield further improvement. The cascaded U-Net improved non-ventilated lung segmentation (P < 0.005). Dice similarity coefficients (DSC) between manually and DL-segmented full lung, non-ventilated, hypo-ventilated, and normal-ventilated regions were 0.965 ± 0.010, 0.840 ± 0.057, 0.715 ± 0.175, and 0.883 ± 0.060, respectively. We observed no statistically significant difference in DCSs between participants with and without COPD (P = 0.41, 0.06, and 0.18 for non-ventilated, hypo-ventilated, and normal-ventilated regions, respectively). CONCLUSION: The proposed cascaded U-Net framework generated fully-automated segmentation of ventilation defects on 3He MRI among older smokers with and without COPD that is consistent with our reference method.

ID-Seg: an infant deep learning-based segmentation framework to improve limbic structure estimates.

Infant brain magnetic resonance imaging (MRI) is a promising approach for studying early neurodevelopment. However, segmenting small regions such as limbic structures is challenging due to their low inter-regional contrast and high curvature. MRI studies of the adult brain have successfully applied deep learning techniques to segment limbic structures, and similar deep learning models are being leveraged for infant studies. However, these deep learning-based infant MRI segmentation models have generally been derived from small datasets, and may suffer from generalization problems. Moreover, the accuracy of segmentations derived from these deep learning models relative to more standard Expectation-Maximization approaches has not been characterized. To address these challenges, we leveraged a large, public infant MRI dataset (n = 473) and the transfer-learning technique to first pre-train a deep convolutional neural network model on two limbic structures: amygdala and hippocampus. Then we used a leave-one-out cross-validation strategy to fine-tune the pre-trained model and evaluated it separately on two independent datasets with manual labels. We term this new approach the Infant Deep learning SEGmentation Framework (ID-Seg). ID-Seg performed well on both datasets with a mean dice similarity score (DSC) of 0.87, a mean intra-class correlation (ICC) of 0.93, and a mean average surface distance (ASD) of 0.31 mm. Compared to the Developmental Human Connectome pipeline (dHCP) pipeline, ID-Seg significantly improved segmentation accuracy. In a third infant MRI dataset (n = 50), we used ID-Seg and dHCP separately to estimate amygdala and hippocampus volumes and shapes. The estimates derived from ID-seg, relative to those from the dHCP, showed stronger associations with behavioral problems assessed in these infants at age 2. In sum, ID-Seg consistently performed well on two different datasets with an 0.87 DSC, however, multi-site testing and extension for brain regions beyond the amygdala and hippocampus are still needed.

PTNet3D: A 3D High-Resolution Longitudinal Infant Brain MRI Synthesizer Based on Transformers.

An increased interest in longitudinal neurodevelopment during the first few years after birth has emerged in recent years. Noninvasive magnetic resonance imaging (MRI) can provide crucial information about the development of brain structures in the early months of life. Despite the success of MRI collections and analysis for adults, it remains a challenge for researchers to collect high-quality multimodal MRIs from developing infant brains because of their irregular sleep pattern, limited attention, inability to follow instructions to stay still during scanning. In addition, there are limited analytic approaches available. These challenges often lead to a significant reduction of usable MRI scans and pose a problem for modeling neurodevelopmental trajectories. Researchers have explored solving this problem by synthesizing realistic MRIs to replace corrupted ones. Among synthesis methods, the convolutional neural network-based (CNN-based) generative adversarial networks (GANs) have demonstrated promising performance. In this study, we introduced a novel 3D MRI synthesis framework- pyramid transformer network (PTNet3D)- which relies on attention mechanisms through transformer and performer layers. We conducted extensive experiments on high-resolution Developing Human Connectome Project (dHCP) and longitudinal Baby Connectome Project (BCP) datasets. Compared with CNN-based GANs, PTNet3D consistently shows superior synthesis accuracy and superior generalization on two independent, large-scale infant brain MRI datasets. Notably, we demonstrate that PTNet3D synthesized more realistic scans than CNN-based models when the input is from multi-age subjects. Potential applications of PTNet3D include synthesizing corrupted or missing images. By replacing corrupted scans with synthesized ones, we observed significant improvement in infant whole brain segmentation.

Automated Multiclass Artifact Detection in Diffusion MRI Volumes via 3D Residual Squeeze-and-Excitation Convolutional Neural Networks.

Diffusion MRI (dMRI) is widely used to investigate neuronal and structural development of brain. dMRI data is often contaminated with various types of artifacts. Hence, artifact type identification in dMRI volumes is an essential pre-processing step prior to carrying out any further analysis. Manual artifact identification amongst a large pool of dMRI data is a highly labor-intensive task. Previous attempts at automating this process are often limited to a binary classification ("poor" vs. "good" quality) of the dMRI volumes or focus on detecting a single type of artifact (e.g., motion, Eddy currents, etc.). In this work, we propose a deep learning-based automated multiclass artifact classifier for dMRI volumes. Our proposed framework operates in 2 steps. In the first step, the model predicts labels associated with 3D mutually exclusive collectively exhaustive (MECE) sub-volumes or "slabs" extracted from whole dMRI volumes. In the second step, through a voting process, the model outputs the artifact class present in the whole volume under investigation. We used two different datasets for training and evaluating our model. Specifically, we utilized 2,494 poor-quality dMRI volumes from the Adolescent Brain Cognitive Development (ABCD) and 4,226 from the Healthy Brain Network (HBN) dataset. Our results demonstrate accurate multiclass volume-level main artifact type prediction with 96.61 and 97.52% average accuracies on the ABCD and HBN test sets, respectively. Finally, in order to demonstrate the effectiveness of the proposed framework in dMRI pre-processing pipelines, we conducted a proof-of-concept dMRI analysis exploring the relationship between whole-brain fractional anisotropy (FA) and participant age, to test whether the use of our model improves the brain-age association.

Improving across-dataset brain tissue segmentation for MRI imaging using transformer.

Brain tissue segmentation has demonstrated great utility in quantifying MRI data by serving as a precursor to further post-processing analysis. However, manual segmentation is highly labor-intensive, and automated approaches, including convolutional neural networks (CNNs), have struggled to generalize well due to properties inherent to MRI acquisition, leaving a great need for an effective segmentation tool. This study introduces a novel CNN-Transformer hybrid architecture designed to improve brain tissue segmentation by taking advantage of the increased performance and generality conferred by Transformers for 3D medical image segmentation tasks. We first demonstrate the superior performance of our model on various T1w MRI datasets. Then, we rigorously validate our model's generality applied across four multi-site T1w MRI datasets, covering different vendors, field strengths, scan parameters, and neuropsychiatric conditions. Finally, we highlight the reliability of our model on test-retest scans taken in different time points. In all situations, our model achieved the greatest generality and reliability compared to the benchmarks. As such, our method is inherently robust and can serve as a valuable tool for brain related T1w MRI studies. The code for the TABS network is available at: https://github.com/raovish6/TABS.