Unsupervised Cross-Modality Adaptation via Dual Structural-Oriented Guidance for 3D Medical Image Segmentation.

Deep convolutional neural networks (CNNs) have achieved impressive performance in medical image segmentation; however, their performance could degrade significantly when being deployed to unseen data with heterogeneous characteristics. Unsupervised domain adaptation (UDA) is a promising solution to tackle this problem. In this work, we present a novel UDA method, named dual adaptation-guiding network (DAG-Net), which incorporates two highly effective and complementary structural-oriented guidance in training to collaboratively adapt a segmentation model from a labelled source domain to an unlabeled target domain. Specifically, our DAG-Net consists of two core modules: 1) Fourier-based contrastive style augmentation (FCSA) which implicitly guides the segmentation network to focus on learning modality-insensitive and structural-relevant features, and 2) residual space alignment (RSA) which provides explicit guidance to enhance the geometric continuity of the prediction in the target modality based on a 3D prior of inter-slice correlation. We have extensively evaluated our method with cardiac substructure and abdominal multi-organ segmentation for bidirectional cross-modality adaptation between MRI and CT images. Experimental results on two different tasks demonstrate that our DAG-Net greatly outperforms the state-of-the-art UDA approaches for 3D medical image segmentation on unlabeled target images.

Deep Learning for Automatic Upper Airway Obstruction Detection by Analysis of Flow-Volume Curve.

BACKGROUND: Due to the similar symptoms of upper airway obstruction to asthma, misdiagnosis is common. Spirometry is a cost-effective screening test for upper airway obstruction and its characteristic patterns involving fixed, variable intrathoracic and extrathoracic lesions. We aimed to develop a deep learning model to detect upper airway obstruction patterns and compared its performance with that of lung function clinicians. METHODS: Spirometry records were reviewed to detect the possible condition of airway stenosis. Then they were confirmed by the gold standard (e.g., computed tomography, endoscopy, or clinic diagnosis of upper airway obstruction). Images and indices derived from flow-volume curves were used for training and testing the model. Clinicians determined cases using spirometry records from the test set. The deep learning model evaluated the same data. RESULTS: Of 45,831 patients' spirometry records, 564 subjects with curves suggesting upper airway obstruction, after verified by the gold standard, 351 patients were confirmed. These cases and another 200 cases without airway stenosis were used as the training and testing sets. 432 clinicians evaluated 20 cases of each of the three patterns and 20 no airway stenosis cases (n = 80). They assigned an accuracy of 41.2% (±15.4) (interquartile range: 27.5-52.5%), with poor agreements (κ = 0.12). For the same cases, the model generated a correct detection of 81.3% (p < 0.0001). CONCLUSIONS: Deep learning could detect upper airway obstruction patterns from other classic patterns of ventilatory defects with high accuracy, whereas clinicians presented marked errors and variabilities. The model may serve as a support tool to enhance clinicians' correct diagnosis of upper airway obstruction using spirometry.

Helping Roles of Artificial Intelligence (AI) in the Screening and Evaluation of COVID-19 Based on the CT Images.

OBJECTIVE: The aim of this study was to explore the role of the AI system which was designed and developed based on the characteristics of COVID-19 CT images in the screening and evaluation of COVID-19. METHODS: The research team adopted an improved U-shaped neural network to segment lungs and pneumonia lesions in CT images through multilayer convolution iterations. Then the appropriate 159 cases were selected to establish and train the model, and Dice loss function and Adam optimizer were used for network training with the initial learning rate of 0.001. Finally, 39 cases (29 positive and 10 negative) were selected for the comparative test. Experimental group: an attending physician a and an associate chief physician a read the CT images to diagnose COVID-19 with the help of the AI system. Control group: an attending physician b and an associate chief physician b did the diagnosis only by their experience, without the help of the AI system. The time spent by each doctor in the diagnosis and their diagnostic results were recorded. Paired t-test, univariate ANOVA, chi-squared test, receiver operating characteristic curves, and logistic regression analysis were used for the statistical analysis. RESULTS: There was statistical significance in the time spent in the diagnosis of different groups (P<0.05). For the group with the optimal diagnostic results, univariate and multivariate analyses both suggested no significant correlation for all variables, and thus it might be the assistance of the AI system, the epidemiological history and other factors that played an important role. CONCLUSION: The AI system developed by us, which was created due to COVID-19, had certain clinical practicability and was worth popularizing.

A slice classification model-facilitated 3D encoder-decoder network for segmenting organs at risk in head and neck cancer.

For deep learning networks used to segment organs at risk (OARs) in head and neck (H&N) cancers, the class-imbalance problem between small volume OARs and whole computed tomography (CT) images results in delineation with serious false-positives on irrelevant slices and unnecessary time-consuming calculations. To alleviate this problem, a slice classification model-facilitated 3D encoder-decoder network was developed and validated. In the developed two-step segmentation model, a slice classification model was firstly utilized to classify CT slices into six categories in the craniocaudal direction. Then the target categories for different OARs were pushed to the different 3D encoder-decoder segmentation networks, respectively. All the patients were divided into training (n = 120), validation (n = 30) and testing (n = 20) datasets. The average accuracy of the slice classification model was 95.99%. The Dice similarity coefficient and 95% Hausdorff distance, respectively, for each OAR were as follows: right eye (0.88 ± 0.03 and 1.57 ± 0.92 mm), left eye (0.89 ± 0.03 and 1.35 ± 0.43 mm), right optic nerve (0.72 ± 0.09 and 1.79 ± 1.01 mm), left optic nerve (0.73 ± 0.09 and 1.60 ± 0.71 mm), brainstem (0.87 ± 0.04 and 2.28 ± 0.99 mm), right temporal lobe (0.81 ± 0.12 and 3.28 ± 2.27 mm), left temporal lobe (0.82 ± 0.09 and 3.73 ± 2.08 mm), right temporomandibular joint (0.70 ± 0.13 and 1.79 ± 0.79 mm), left temporomandibular joint (0.70 ± 0.16 and 1.98 ± 1.48 mm), mandible (0.89 ± 0.02 and 1.66 ± 0.51 mm), right parotid (0.77 ± 0.07 and 7.30 ± 4.19 mm) and left parotid (0.71 ± 0.12 and 8.41 ± 4.84 mm). The total segmentation time was 40.13 s. The 3D encoder-decoder network facilitated by the slice classification model demonstrated superior performance in accuracy and efficiency in segmenting OARs in H&N CT images. This may significantly reduce the workload for radiation oncologists.

Retinotopic Changes in the Gray Matter Volume and Cerebral Blood Flow in the Primary Visual Cortex of Patients With Primary Open-Angle Glaucoma.

PURPOSE: To assess the cortical structure and cerebral blood flow changes in the brain of patients with primary open-angle glaucoma (POAG). METHODS: High-resolution anatomical magnetic resonance imaging (MRI) and arterial spin labeling (ASL)-MRI were performed in 23 POAG patients and 29 controls. Patients were further divided into early-moderate and advanced groups based on mean deviation (MD) cutoff of 12 dB. A baseline scan was obtained and repeated during visual stimulation to the central preserved visual field in the more affected eye of POAG patients and a randomly selected eye of controls. Gray matter volume (GMV) and cerebral blood flow (CBF) throughout the whole brain were compared between patients and controls. RESULTS: Compared to controls, a region with significant reduction of GMV was detected in the anterior calcarine fissure of advanced POAG patients (P < 0.001, voxels = 503, 1698 mm3). Patients with early-moderate POAG had resting CBF similar to that of controls. However, a region with marked CBF decrease was detected in the anterior calcarine fissure of advanced POAG patients (P < 0.001, voxels = 1687, 13,496 mm3). The region with CBF reduction in advanced POAG showed good colocalization with the region with GMV decrease in this group. Following visual stimulation, patients with advanced POAG showed significantly lower increase in CBF in the occipital lobes (P < 0.001, voxels = 112, 896 mm3) as compared to controls (P < 0.001, voxels = 1880, 15,040 mm3) and early-moderate POAG (P < 0.001, voxels = 2233, 17,864 mm3). CONCLUSIONS: Primary open-angle glaucoma patients demonstrate a disease severity-dependent retinotopic pattern of cortical atrophy and CBF abnormalities in the visual cortex. Cerebral blood flow may be a potential biomarker for the brain involvement in glaucoma.

Resting spontaneous activity in the default mode network predicts performance decline during prolonged attention workload.

After continuous and prolonged cognitive workload, people typically show reduced behavioral performance and increased feelings of fatigue, which are known as "time-on-task (TOT) effects". Although TOT effects are pervasive in modern life, their underlying neural mechanisms remain elusive. In this study, we induced TOT effects by administering a 20-min continuous psychomotor vigilance test (PVT) to a group of 16 healthy adults and used resting-state blood oxygen level-dependent (BOLD) functional magnetic resonance imaging (fMRI) to examine spontaneous brain activity changes associated with fatigue and performance. Behaviorally, subjects displayed robust TOT effects, as reflected by increasingly slower reaction times as the test progressed and higher self-reported mental fatigue ratings after the 20-min PVT. Compared to pre-test measurements, subjects exhibited reduced amplitudes of low-frequency fluctuation (ALFF) in the default mode network (DMN) and increased ALFF in the thalamus after the test. Subjects also exhibited reduced anti-correlations between the posterior cingulate cortex (PCC) and right middle prefrontal cortex after the test. Moreover, pre-test resting ALFF in the PCC and medial prefrontal cortex (MePFC) predicted subjects' subsequent performance decline; individuals with higher ALFF in these regions exhibited more stable reaction times throughout the 20-min PVT. These results support the important role of both task-positive and task-negative networks in mediating TOT effects and suggest that spontaneous activity measured by resting-state BOLD fMRI may be a marker of mental fatigue.

Altered salience network connectivity predicts macronutrient intake after sleep deprivation.

Although insufficient sleep is a well-recognized risk factor for overeating and weight gain, the neural mechanisms underlying increased caloric (particularly fat) intake after sleep deprivation remain unclear. Here we used resting-state functional magnetic resonance imaging and examined brain connectivity changes associated with macronutrient intake after one night of total sleep deprivation (TSD). Compared to the day following baseline sleep, healthy adults consumed a greater percentage of calories from fat and a lower percentage of calories from carbohydrates during the day following TSD. Subjects also exhibited increased brain connectivity in the salience network from the dorsal anterior cingulate cortex (dACC) to bilateral putamen and bilateral anterior insula (aINS) after TSD. Moreover, dACC-putamen and dACC-aINS connectivity correlated with increased fat and decreased carbohydrate intake during the day following TSD, but not during the day following baseline sleep. These findings provide a potential neural mechanism by which sleep loss leads to increased fat intake.

Gender differences in cerebral regional homogeneity of adult healthy volunteers: a resting-state FMRI study.

OBJECTIVE: We sought to use the regional homogeneity (ReHo) approach as an index in the resting-state functional MRI to investigate the gender differences of spontaneous brain activity within cerebral cortex and resting-state networks (RSNs) in young adult healthy volunteers. METHODS: One hundred and twelve healthy volunteers (56 males, 56 females) participated in the resting-state fMRI scan. The ReHo mappings in the cerebral cortex and twelve RSNs of the male and female groups were compared. RESULTS: We found statistically significant gender differences in the primary visual network (PVN) (P < 0.004, with Bonferroni correction) and left attention network (LAtN), default mode network (DMN), sensorimotor network (SMN), executive network (EN), and dorsal medial prefrontal network (DMPFC) as well (P < 0.05, uncorrected). The male group showed higher ReHo in the left precuneus, while the female group showed higher ReHo in the right middle cingulate gyrus, fusiform gyrus, left inferior parietal lobule, precentral gyrus, supramarginal gyrus, and postcentral gyrus. CONCLUSIONS: Our results suggested that men and women had regional specific differences during the resting-state. The findings may improve our understanding of the gender differences in behavior and cognition from the perspective of resting-state brain function.

Serotonin transporter genotype modulates functional connectivity between amygdala and PCC/PCu during mood recovery.

The short (S) allele of the serotonin transporter-linked polymorphic region (5-HTTLPR) has been associated with increased susceptibility to depression. Previous neuroimaging studies have consistently showed increased amygdala activity during the presentation of negative stimuli or regulation of negative emotion in the homozygous short allele carriers, suggesting the key role of amygdala response in mediating increased risk for depression. The brain default mode network (DMN) has also been shown to modulate amygdala activity. However, it remains unclear whether 5-HTTLPR genetic variation modulates functional connectivity (FC) between the amygdala and regions of DMN. In this study, we re-analyzed our previous imaging dataset and examined the effects of 5-HTTLPR genetic variation on amygdala connectivity. A total of 15 homozygous short (S/S) and 15 homozygous long individuals (L/L) were scanned in functional magnetic resonance imaging (fMRI) during four blocks: baseline, sad mood, mood recovery, and return to baseline. The S/S and L/L groups showed a similar pattern of FC and no differences were found between the two groups during baseline and sad mood scans. However, during mood recovery, the S/S group showed significantly reduced anti-correlation between amygdala and posterior cingulate cortex/precuneus (PCC/PCu) compared to the L/L group. Moreover, PCC/PCu-amygdala connectivity correlated with amygdala activity in the S/S group but not the L/L group. These results suggest that 5-HTTLPR genetic variation modulates amygdala connectivity which subsequently affects its activity during mood regulation, providing an additional mechanism by which the S allele confers depression risk.

Assessment of risk-taking and impulsive behaviors: A comparison between three tasks.

The balloon analogue risk task (BART), the delay discounting task (DDT), and the Iowa gambling task (IGT) are increasingly used for the assessment of risk-taking and impulsive behaviors. This study examined the reliability of and relationships between these three tasks in healthy Chinese subjects. The BART and DDT showed moderate to high test-retest reliability across three test sessions. However, the IGT showed low reliability for the first two sessions but high reliability for the last two sessions. Between tasks, only the BART and IGT showed significant correlations at the last two sessions, while no other correlations were found. These findings support the view that impulsivity is a complex construct with no single personality trait underlying the disposition for impulsive behaviors.

Resting state brain function analysis using concurrent BOLD in ASL perfusion fMRI.

The past decade has seen astounding discoveries about resting-state brain activity patterns in normal brain as well as their alterations in brain diseases. While the vast majority of resting-state studies are based on the blood-oxygen-level-dependent (BOLD) functional MRI (fMRI), arterial spin labeling (ASL) perfusion fMRI can simultaneously capture BOLD and cerebral blood flow (CBF) signals, providing a unique opportunity for assessing resting brain functions with concurrent BOLD (ccBOLD) and CBF signals. Before taking that benefit, it is necessary to validate the utility of ccBOLD signal for resting-state analysis using conventional BOLD (cvBOLD) signal acquired without ASL modulations. To address this technical issue, resting cvBOLD and ASL perfusion MRI were acquired from a large cohort (n = 89) of healthy subjects. Four widely used resting-state brain function analyses were conducted and compared between the two types of BOLD signal, including the posterior cingulate cortex (PCC) seed-based functional connectivity (FC) analysis, independent component analysis (ICA), analysis of amplitude of low frequency fluctuation (ALFF), and analysis of regional homogeneity (ReHo). Consistent default mode network (DMN) as well as other resting-state networks (RSNs) were observed from cvBOLD and ccBOLD using PCC-FC analysis and ICA. ALFF from both modalities were the same for most of brain regions but were different in peripheral regions suffering from the susceptibility gradients induced signal drop. ReHo showed difference in many brain regions, likely reflecting the SNR and resolution differences between the two BOLD modalities. The DMN and auditory networks showed highest CBF values among all RSNs. These results demonstrated the feasibility of ASL perfusion MRI for assessing resting brain functions using its concurrent BOLD in addition to CBF signal, which provides a potentially useful way to maximize the utility of ASL perfusion MRI.

Changes in cerebral blood flow and anxiety associated with an 8-week mindfulness programme in women with breast cancer.

This study employed functional magnetic resonance imaging to evaluate changes in cerebral blood flow (CBF) associated with the Mindfulness-based Art Therapy (MBAT) programme and correlate such changes to stress and anxiety in women with breast cancer. Eighteen breast cancer patients were randomized to the MBAT or education control group. The patients received the diagnosis of breast cancer between 6 months and 3 years prior to enrollment and were not in active treatment. The age of participants ranged from 52 to 77 years. A voxel-based analysis was performed to assess differences at rest, during meditation and during a stress task. The anxiety sub-scale of the Symptoms Checklist-90-Revised was compared with changes in resting CBF before and after the programmes. Subjects in the MBAT arm demonstrated significant increases in CBF at rest and during meditation in multiple limbic regions, including the left insula, right amygdala, right hippocampus and bilateral caudate. Patients in the MBAT programme also had a significant correlation between increased CBF in the left caudate and decreased anxiety scores. In the MBAT group, responses to a stressful cue resulted in reduced activation of the posterior cingulate. The results demonstrate that the MBAT programme was associated with significant changes in CBF, which correlated with decreased anxiety over an 8-week period.

Altered resting brain function and structure in professional badminton players.

Neuroimaging studies of professional athletic or musical training have demonstrated considerable practice-dependent plasticity in various brain structures, which may reflect distinct training demands. In the present study, structural and functional brain alterations were examined in professional badminton players and compared with healthy controls using magnetic resonance imaging (MRI) and resting-state functional MRI. Gray matter concentration (GMC) was assessed using voxel-based morphometry (VBM), and resting-brain functions were measured by amplitude of low-frequency fluctuation (ALFF) and seed-based functional connectivity. Results showed that the athlete group had greater GMC and ALFF in the right and medial cerebellar regions, respectively. The athlete group also demonstrated smaller ALFF in the left superior parietal lobule and altered functional connectivity between the left superior parietal and frontal regions. These findings indicate that badminton expertise is associated with not only plastic structural changes in terms of enlarged gray matter density in the cerebellum, but also functional alterations in fronto-parietal connectivity. Such structural and functional alterations may reflect specific experiences of badminton training and practice, including high-capacity visuo-spatial processing and hand-eye coordination in addition to refined motor skills.