Structural Magnetic Resonance Imaging Research Articles

Identification of multimodal brain imaging biomarkers in first-episode drugs-naive major depressive disorder through a multi-site large-scale MRI consortium data

BackgroundMajor depressive disorder (MDD) is a severe and common mental illness. The first-episode drugs-naive MDD (FEDN-MDD) patients, who have not undergone medication intervention, contribute to understanding the biological basis of MDD. Multimodal Magnetic Resonance Imaging can provide a comprehensive understanding of brain functional and structural abnormalities in MDD. However, most MDD studies use single-modal, small-scale MRI data. And several multimodal studies of MDD are limited to simple linear combinations of functional and structural features. MethodsWe screened a large sample of FEDN-MDD patients and healthy controlsmultimodal MRI data. Extracting the fractional amplitude of low-frequency fluctuations (fALFF) feature from functional magnetic resonance imaging and the gray matter volume (GMV) feature from structural magnetic resonance imaging. The mCCA-jICA method was used to integrate these two modal features to investigate the functional-structural co-variation abnormalities in MDD. To validate the stability of the extracted functional-structural covariant abnormalities features, we apply them to identify FEDN-MDD patients. ResultsThe results show that compared to healthy controls, FEDN-MDD patients exhibit joint group-discriminative independent component and modality-specific group-discriminative independent component, suggesting functional-structural covariant abnormalities in MDD patients. Using lightGBM classifier, we achieve a classification accuracy of 99.84 %. LimitationWe use GMV and fALFF for multimodal fusion shows promise, but requires further validation with other datasets and exploration of additional multimodal features. ConclusionsThis may indicate that multimodal fusion features can effectively explore information between different modalities and can accurately identify FEDN-MDD patients, suggesting their potential as multimodal brain imaging biomarkers for MDD.

Beyond IQ: executive function deficits and their relation to functional, clinical, and neuroimaging outcomes in 3q29 deletion syndrome.

3q29 deletion syndrome (3q29del) is a rare (~1:30 000) genomic disorder associated with a wide array of neurodevelopmental and psychiatric phenotypes. Prior work by our team identified clinically significant executive function (EF) deficits in 47% of individuals with 3q29del; however, the nuances of EF in this population have not been described. We used the Behavior Rating Inventory of Executive Function (BRIEF) to perform the first in-depth assessment of real-world EF in a cohort of 32 individuals with 3q29del (62.5% male, mean age = 14.5 ± 8.3 years). All participants were also evaluated with gold-standard neuropsychiatric and cognitive assessments. High-resolution structural magnetic resonance imaging was performed on a subset of participants (n = 24). We found global deficits in EF; individuals with 3q29del scored higher than the population mean on the BRIEF global executive composite (GEC) and all subscales. In total, 81.3% of study subjects (n = 26) scored in the clinical range on at least one BRIEF subscale. BRIEF GEC T scores were higher among 3q29del participants with a diagnosis of attention deficit/hyperactivity disorder (ADHD), and BRIEF GEC T scores were associated with schizophrenia spectrum symptoms as measured by the Structured Interview for Psychosis-Risk Syndromes. BRIEF GEC T scores were not associated with cognitive ability. The BRIEF-2 ADHD form accurately (sensitivity = 86.7%) classified individuals with 3q29del based on ADHD diagnosis status. BRIEF GEC T scores were correlated with cerebellar white matter and subregional cerebellar cortex volumes. Together, these data expand our understanding of the phenotypic spectrum of 3q29del and identify EF as a core feature linked to both psychiatric and neuroanatomical features of the syndrome.

Uncovering neural substrates across Alzheimer's disease stages using contrastive variational autoencoder.

Alzheimer's disease is the most common major neurocognitive disorder. Although currently, no cure exists, understanding the neurobiological substrate underlying Alzheimer's disease progression will facilitate early diagnosis and treatment, slow disease progression, and improve prognosis. In this study, we aimed to understand the morphological changes underlying Alzheimer's disease progression using structural magnetic resonance imaging data from cognitively normal individuals, individuals with mild cognitive impairment, and Alzheimer's disease via a contrastive variational autoencoder model. We used contrastive variational autoencoder to generate synthetic data to boost the downstream classification performance. Due to the ability to parse out the nonclinical factors such as age and gender, contrastive variational autoencoder facilitated a purer comparison between different Alzheimer's disease stages to identify the pathological changes specific to Alzheimer's disease progression. We showed that brain morphological changes across Alzheimer's disease stages were significantly associated with individuals' neurofilament light chain concentration, a potential biomarker for Alzheimer's disease, highlighting the biological plausibility of our results.

Disentangling normal and pathological brain atrophy for the diagnosis of mild cognitive impairment and Alzheimer’s disease

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder. Mild cognitive impairment (MCI) is the prodromal stage of AD. Accurate identification of these conditions is crucial for the early diagnosis of these diseases. Brain tissue atrophies, observable in regions such as the hippocampus and cortices, serve as an essential biomarker for MCI and AD. However, similar atrophies are also present in elderly individuals with normal cognitive function, albeit to a lesser extent, and can be found in other non-biomarker brain tissues. To address this challenge, we introduce an atrophy disentanglement network (AD-Net), designed to decouple age-related normal atrophies and disease-specific pathological atrophies in structural magnetic resonance imaging images. Specifically, we first design a dual-task prediction module, guiding the model to differentiate normal and pathological atrophies. Subsequently, we devise a feature orthogonality module for enhanced separation of the two types of atrophies. Our extensive experiments demonstrate that AD-Net outperforms existing methods, highlighting the efficiency of the devised dual-task prediction and feature orthogonality modules in disentangling normal and pathological image features and further improving the diagnosis for AD and MCI. The source code is publicly avaliable at https://github.com/CVAPPS24/ADNet.git.

Predicting individual personality styles using macrostructural information: A multivariate pattern study

Advances in neuroimaging techniques have uncovered the relationships between neuroanatomy and personality traits like neuroticism and extraversion. Nevertheless, few studies have examined the reliable biological markers that identify individuals with different personality styles. Therefore, we evaluated the diagnostic capability of structural magnetic resonance imaging with a sample of 192 healthy adults by using a machine learning algorithm to divide them into four groups according to brain structural characteristics. Our results showed that personality styles, but not personality traits, could be discriminated with the extraversion candidate brain system (e. g., the frontal cortex, hippocampus, amygdala) and the neuroticism candidate brain system (e. g., the insula, the right anterior cingulate cortex). These findings indicate that personality styles that consider person-centered traits may more effectively predict individuals' mental health, providing a potential brain biomarker for early individual psychological risk assessments.

18F]FDG PET integrated with structural MRI for accurate brain age prediction.

Brain aging is a complex and heterogeneous process characterized by both structural and functional decline. This study aimed to establish a novel deep learning (DL) method for predicting brain age by utilizing structural and metabolic imaging data. The dataset comprised participants from both the Universal Medical Imaging Diagnostic Center (UMIDC) and the Alzheimer's Disease Neuroimaging Initiative (ADNI). The former recruited 395 normal control (NC) subjects, while the latter included 438 NC subjects, 51 mild cognitive impairment (MCI) subjects, and 56 Alzheimer's disease (AD) subjects. We developed a novel dual-pathway, 3D simple fully convolutional network (Dual-SFCNeXt) to estimate brain age using [18F]fluorodeoxyglucose positron emission tomography ([18F]FDG PET) and structural magnetic resonance imaging (sMRI) images of NC subjects as input. Several prevailing DL models were trained and tested using either MRI or PET data for comparison. Model accuracies were evaluated using mean absolute error (MAE) and Pearson's correlation coefficient (r). Brain age gap (BAG), deviations of brain age from chronologic age, was correlated with cognitive assessments in MCI and AD subjects. Both PET- and MRI-based models achieved high prediction accuracy. The leading model was the SFCNeXt (the single-pathway version) for PET (MAE = 2.92, r = 0.96) and MRI (MAE = 3.23, r = 0.95) on all samples. By integrating both PET and MRI images, the Dual-SFCNeXt demonstrated significantly improved accuracy (MAE = 2.37, r = 0.97) compared to all single-modality models. Significantly higher BAG was observed in both the AD (P < 0.0001) and MCI (P < 0.0001) groups compared to the NC group. BAG correlated significantly with Mini-Mental State Examination (MMSE) scores (r=-0.390 for AD, r=-0.436 for MCI) and the Clinical Dementia Rating Scale Sum of Boxes (CDR-SB) scores (r = 0.333 for AD, r = 0.372 for MCI). The integration of [18F]FDG PET with structural MRI enhances the accuracy of brain age prediction, potentially introducing a new avenue for related multimodal brain age prediction studies.

Perceptual and semantic deficits in face recognition in semantic dementia

State of the artSemantic dementia (SD) patients including semantic variant primary progressive aphasia (svPPA) and semantic behavioral variant frontotemporal dementia (sbvFTD) patients show semantic difficulties identifying faces and known people related to right anterior temporal lobe (ATL) atrophy. However, it remains unclear whether they also have perceptual deficits in face recognition. MethodologyWe selected 74 SD patients (54 with svPPA and predominant left ATL atrophy and 20 with sbvFTD and predominant right ATL atrophy) and 36 cognitively healthy controls (HC) from UCSF Memory and Aging Center. They underwent a perceptual face processing test (Benton facial recognition test-short version; BFRT-S), and semantic face processing tests (UCSF Famous people battery – Recognition, Naming, Semantic associations – pictures and words subtests), as well as structural magnetic resonance imaging (MRI). Neural correlates with the task’s performance were conducted with a Voxel-based morphometry approach using CAT12. ResultssvPPA and sbvFTD patients were impaired on all semantic face processing tests, with sbvFTD patients performing significantly lower on the famous faces’ recognition task in comparison to svPPA, and svPPA performing significantly lower on the naming task in comparison to sbvFTD. These tasks predominantly correlated with gray matter (GM) volumes in the right and left ATL, respectively. Compared to HC, both svPPA and sbvFTD patients showed preserved performance on the perceptual face processing test (BFRT-S), and performance on the BFRT-S negatively correlated with GM volume in the right posterior superior temporal sulcus (pSTS). ConclusionOur results suggest that early in the disease, with the atrophy mostly restricted to the anterior temporal regions, SD patients do not present with perceptual deficits. However, more severe SD cases with atrophy in right posterior temporal regions might show lower performance on face perception tests, in addition to the semantic face processing deficits. Early sparing of face perceptual deficits in SD patients, regardless of hemispheric lateralization, furthers our understanding of clinical phenomenology and therapeutical approaches of this complex disease.

Altered Subcortical Brain Volume and Cortical Thickness Related to Insulin Resistance in Type 2 Diabetes Mellitus.

The objective of this study is to examine the alterations in subcortical brain volume and cortical thickness among individuals diagnosed with Type 2 diabetes mellitus (T2DM) through the application of morphometry techniques and, additionally, to investigate the potential association between these modifications and insulin resistance (IR). The present cross-sectional study comprised a total of 121 participants (n=48 with healthy controls [HCs] and n=73 with T2DM) who were recruited and underwent a battery of cognitive testing and structural magnetic resonance imaging (MRI). FreeSurfer was used to process the MRI data. Analysis of covariance compared discrepancies in cortical thickness and subcortical brain volume between T2DM and HCs, adjusting for the potential confounding effects of gender, age, education, and body mass index (BMI). Exploratory partial correlations investigated links between IR and brain structure in T2DM participants. Compared with HCs, individuals with T2DM demonstrated a cortical thickness decrease in the right caudal middle frontal gyrus, right pars opercularis, left precentral gyrus, and bilateral superior frontal gyrus. Furthermore, this study for T2DM found that the severity of IR was inversely related to the volume of the left putamen and left hippocampus, as well as the thickness of the left pars orbitalis, left pericalcarine, right entorhinal area, and right rostral anterior cingulate gyrus. The evidence for structural brain changes in T2DM was observed, and alterations in cortical thickness were concentrated in the frontal lobes. Correlations between IR and frontal cortical thinning may serve as a potential neuroimaging marker of T2DM and lead to various diabetes-related brain complications.

Blood-derived microRNAs are related to cognitive domains in the general population.

Blood-derived microRNAs (miRNAs) are potential candidates for detecting and preventing subclinical cognitive dysfunction. However, replication of previous findings and identification of novel miRNAs associated with cognitive domains, including their relation to brain structure and the pathways they regulate, are still lacking. We examined blood-derived miRNAs and miRNA co-expression clusters in relation to cognitive domains, structural magnetic resonance imaging measures, target gene expression, and genetic variants in 2869 participants of a population-based cohort. Five previously identified and 14 novel miRNAs were associated with cognitive domains. Eleven of these were also associated with cortical thickness and two with hippocampal volume. Multi-omics analysis showed that certain identified miRNAs were genetically influenced and regulated genes in pathways like neurogenesis and synapse assembly. We identified miRNAs associated with cognitive domains, brain regions, and neuronal processes affected by aging and neurodegeneration, making them promising candidate blood-based biomarkers or therapeutic targets of subclinical cognitive dysfunction. We investigated the association of blood-derived microRNAs with cognitive domains. Five previously identified and 14 novel microRNAs were associated with cognition. Eleven cognition-related microRNAs were also associated with cortical thickness. Identified microRNAs were linked to genes associated with neuronal functions. Results provide putative biomarkers or therapeutic targets of cognitive aging.

The mediation effect of the inferior-parietal cortex and globus pallidus on the relationship between family conflict and major depressive disorder

ObjectiveFamily conflict is an important risk factor for major depressive disorder (MDD) and is associated with structural alterations in the brain. However, it is unclear whether structural alterations associated with family conflict would contribute to depression. This study aims to investigate the neuroimaging characteristics that connect family conflict with depression. MethodsThis study included 54 healthy controls and 53 antidepressant-free patients with MDD. Both groups completed the Beck Depression Inventory-II, the simplified Chinese version of the Family Environment Scale, and the Childhood Trauma Questionnaire. Structural Magnetic Resonance Imaging data was collected to measure cortical thickness and regional gray matter volumes. ResultsFamily conflict has a significant effect on depression diagnosis. Higher levels of family conflict were positively associated with symptoms of sadness, guilty feelings, and punishment feelings in patients, as well as with cortical thickness in the right inferior-parietal cluster and the volumes of the left globus pallidus in all participants. In the patient group, cortical thickness in the right inferior-parietal cluster and volume of the left globus pallidus were negatively related to symptoms of sadness and guilty feelings, respectively. The structural alteration in the right inferior-parietal cluster mediated the relationship of family conflict and sadness, whereas changes in the globus pallidus mediated the associations between family conflict and both depression and guilty feelings in patients. ConclusionFindings revealed the relationships between family conflict and depression, including both depression diagnosis and specific symptoms. Cortical thickness in the right inferior-parietal cortex and the volume of the left globus pallidus played mediating roles in these relationships, indicating the important contributions of these brain regions to the effect of family conflict on depression.

Radiomics with structural magnetic resonance imaging, surface morphometry features, neurology scales, and clinical metrics to evaluate the neurodevelopment of preschool children with corrected tetralogy of Fallot.

Despite the improved survival rates of children with tetralogy of Fallot (TOF), various degrees of neurodevelopmental disorders persist. Currently, there is a lack of quantitative and objective imaging markers to assess the neurodevelopment of individuals with TOF. This study aimed to noninvasively examine potential quantitative imaging markers of TOF neurodevelopment by combining radiomics signatures and morphological features and to further clarify the relationship between imaging markers and clinical neurodevelopment metrics. This study included 33 preschool children who had undergone surgical correction for TOF and 29 healthy controls (36 in the training cohort and 26 in the testing cohort), all of whom underwent three-dimensional T1-weighted high-resolution (T1-3D) head magnetic resonance imaging (MRI). Radiomics features were extracted by Pyradiomics to construct radiomics models, while surface morphometry (surface and volumetric) features were analyzed to build morphometry models. Merged models integrating radiomics and morphometry features were subsequently developed. The optimal discriminative radiomics signatures were identified via least absolute shrinkage and selection operator (LASSO). Machine learning classification models include support vector machine (SVM) with radial basis function (RBF) and multivariable logistic regression (MLR) models, both of which were used to evaluate the potential imaging biomarkers. Performances of models were evaluated based on their calibration and classification metrics. The area under the receiver operating characteristic curves (AUCs) of the models were evaluated using the Delong test. Neurodevelopmental assessments for children with corrected TOF were conducted with the Wechsler Preschool and Primary Scale of Intelligence-Fourth Edition (WPPSI-IV). Furthermore, the correlation of the significant discriminative indicators with clinical metrics and neurodevelopmental scales was evaluated. Twelve discriminative radiomics signatures, optimized for classification, were identified. The performance of the merged model (AUCs of 0.922 and 0.917 for the training set and test set with SVM, respectively) was superior to that of the single radiomics model (AUCs of 0.915 and 0.917 for the training set and test set with SVM, respectively) and that of the single morphometric models (AUCs of 0.803 and 0.756 for the training set and test set with SVM, respectively). The radiomics model demonstrated higher significance than did the morphometric models in training set with SVM (AUC: 0.915 vs. 0.803; P<0.001). Additionally, the significant indicators showed a correlation with clinical indicators and neurodevelopmental scales. MRI-based radiomics features combined with morphometry features can provide complementary information to identify neurodevelopmental abnormalities in children with corrected TOF, which will provide potential evidence for clinical diagnosis and treatment.

Joint self-supervised and supervised contrastive learning for multimodal MRI data: Towards predicting abnormal neurodevelopment

The integration of different imaging modalities, such as structural, diffusion tensor, and functional magnetic resonance imaging, with deep learning models has yielded promising outcomes in discerning phenotypic characteristics and enhancing disease diagnosis. The development of such a technique hinges on the efficient fusion of heterogeneous multimodal features, which initially reside within distinct representation spaces. Naively fusing the multimodal features does not adequately capture the complementary information and could even produce redundancy. In this work, we present a novel joint self-supervised and supervised contrastive learning method to learn the robust latent feature representation from multimodal MRI data, allowing the projection of heterogeneous features into a shared common space, and thereby amalgamating both complementary and analogous information across various modalities and among similar subjects. We performed a comparative analysis between our proposed method and alternative deep multimodal learning approaches. Through extensive experiments on two independent datasets, the results demonstrated that our method is significantly superior to several other deep multimodal learning methods in predicting abnormal neurodevelopment. Our method has the capability to facilitate computer-aided diagnosis within clinical practice, harnessing the power of multimodal data. The source code of the proposed model is publicly accessible on GitHub: https://github.com/leonzyzy/Contrastive-Network.

Reproducible brain PET data analysis: easier said than done.

While a great deal of recent effort has focused on addressing a perceived reproducibility crisis within brain structural magnetic resonance imaging (MRI) and functional MRI research communities, this article argues that brain positron emission tomography (PET) research stands on even more fragile ground, lagging behind efforts to address MRI reproducibility. We begin by examining the current landscape of factors that contribute to reproducible neuroimaging data analysis, including scientific standards, analytic plan pre-registration, data and code sharing, containerized workflows, and standardized processing pipelines. We then focus on disparities in the current status of these factors between brain MRI and brain PET. To demonstrate the positive impact that further developing such reproducibility factors would have on brain PET research, we present a case study that illustrates the many challenges faced by one laboratory that attempted to reproduce a community-standard brain PET processing pipeline. We identified key areas in which the brain PET community could enhance reproducibility, including stricter reporting policies among PET dedicated journals, data repositories, containerized analysis tools, and standardized processing pipelines. Other solutions such as mandatory pre-registration, data sharing, code availability as a condition of grant funding, and online forums and standardized reporting templates, are also discussed. Bolstering these reproducibility factors within the brain PET research community has the potential to unlock the full potential of brain PET research, propelling it toward a higher-impact future.

The volumes of amygdala subregions and peripheral programmed cell death protein-1 levels are associated with cognitive decline in individuals with knee osteoarthritis.

Persistent pain is a prominent symptom of knee osteoarthritis (KOA) and has been associated with cognitive decline in individuals with KOA. The amygdala, a complex structure consisting of nine subnuclei, and programmed cell death protein-1 (PD-1) levels play crucial roles in pain regulation and cognitive processing. This study aims to investigate the relationships among amygdala subregion volumes, cognitive function, and PD-1 levels to elucidate the underlying mechanism of cognitive decline in KOA. In this cross-sectional study, we recruited 36 patients with KOA and 25 age/gender-matched healthy controls for neuropsychological tests, structural magnetic resonance imaging scanning, and measurement of serum PD-1 levels. We used the atlas provided by FreeSurfer software to automatically segment the amygdala subnuclei. Subsequently, we compared the volumes of amygdala subregions between groups and explored their correlation with clinical scores and PD-1 levels. Compared to healthy controls, individuals with KOA exhibited significantly lower scores on global cognition tasks, such as long-delay free recall, short-delay free recall, and immediate recall tasks. Moreover, they displayed decreased volumes in lateral nucleus basal nucleus paralaminar nucleus while showing increased volumes in accessory basal nucleus, central nucleus, medial nucleus, and cortical nucleus. Within the KOA group specifically, paralaminar volume was negatively correlated with immediate recall scores; pain scores were negatively correlated with global cognition; basal volume was negatively correlated with PD-1 levels. Our findings highlight those alterations in amygdala subregion volumes along with changes in serum PD-1 levels may contribute to observe cognitive decline among individuals suffering from KOA.

Validation of a Paralimbic-Related Subcortical Brain Dysmaturation MRI Score in Infants with Congenital Heart Disease.

Background: Brain magnetic resonance imaging (MRI) of infants with congenital heart disease (CHD) shows brain immaturity assessed via a cortical-based semi-quantitative score. Our primary aim was to develop an infant paralimbic-related subcortical-based semi-quantitative dysmaturation score, termed brain dysplasia score (BDS), to detect abnormalities in CHD infants compared to healthy controls and secondarily to predict clinical outcomes. We also validated our BDS in a preclinical mouse model of hypoplastic left heart syndrome. Methods: A paralimbic-related subcortical BDS, derived from structural MRIs of infants with CHD, was compared to healthy controls and correlated with clinical risk factors, regional cerebral volumes, feeding, and 18-month neurodevelopmental outcomes. The BDS was validated in a known CHD mouse model named Ohia with two disease-causing genes, Sap130 and Pchda9. To relate clinical findings, RNA-Seq was completed on Ohia animals. Findings: BDS showed high incidence of paralimbic-related subcortical abnormalities (including olfactory, cerebellar, and hippocampal abnormalities) in CHD infants (n = 215) compared to healthy controls (n = 92). BDS correlated with reduced cortical maturation, developmental delay, poor language and feeding outcomes, and increased length of stay. Ohia animals (n = 63) showed similar BDS findings, and RNA-Seq analysis showed altered neurodevelopmental and feeding pathways. Sap130 mutants correlated with a more severe BDS, whereas Pcdha9 correlated with a milder phenotype. Conclusions: Our BDS is sensitive to dysmaturational differences between CHD and healthy controls and predictive of poor outcomes. A similar spectrum of paralimbic and subcortical abnormalities exists between human and Ohia mutants, suggesting a common genetic mechanistic etiology.

Potential Biomarkers of Dysmenorrhea Relief: A MEG Study of Hinoki Aromatherapy and Working Memory.

Background/Objectives: This study explored the potential of Hinoki aromatherapy to induce biomarkers of dysmenorrhea relief through working memory. Structural magnetic resonance imaging and magnetoencephalography (MEG) were used to examine their effects on neurophysiological responses to a visual working memory (VWM) test. Behavioral performance was measured to understand its effects on the overall working memory. Methods: Twenty-four healthy participants completed the VWM task during nonmenstruation and menstruation. Behavioral (accuracy and reaction time) and neurophysiological (event-related fields, source estimation, and permutation t-test on source data) measures were assessed without and with Hinoki aromatherapy. Results: A significant difference in the ratio of accuracy to reaction time was found without and with aromatherapy in participants with dysmenorrhea, suggesting that aromatherapy may improve working memory performance in this population. MEG analysis revealed high temporal resolution of evoked latency and intensity during the VWM task. Source localization of the activation aimed to identify brain areas involved in dysmenorrhea. Aromatherapy reduced signals in these areas, which may also contribute to reducing dysmenorrhea-related visual signals. Conclusions: Based on these findings, Hinoki aromatherapy may be a promising treatment option for alleviating dysmenorrhea and improving related symptoms by reducing activity in brain pain processing regions. These regions include the left entorhinal cortex, inferior temporal gyrus, primary visual cortex, retrosplenial cortex, and presubiculum. Furthermore, decreased activity in these areas with aromatherapy suggests that they could be used as biomarkers of dysmenorrhea relief.

The German research consortium for the study of bipolar disorder (BipoLife): a quality assurance protocol for MR neuroimaging data

BackgroundThe German multicenter research consortium BipoLife aims to investigate the mechanisms underlying bipolar disorders. It focuses in particular on people at high risk of developing the disorder and young patients in the early stages of the disease. Functional and structural magnetic resonance imaging (MRI) data was collected in all participating centers. The collection of neuroimaging data in a longitudinal, multicenter study requires the implementation of a comprehensive quality assurance (QA) protocol. Here, we outline this protocol and illustrate its application within the BipoLife consortium.MethodsThe QA protocol consisted of (1) a training of participating research staff, (2) regular phantom measurements to evaluate the MR scanner performance and its temporal stability across the course of the study, and (3) the assessment of the quality of human MRI data by evaluating a variety of image metrics (e.g., signal-to-noise ratio, ghosting level). In this article, we will provide an overview on these QA procedures and show exemplarily the influence of its application on the results of standard neuroimaging analysis pipelines.DiscussionThe QA protocol helped to characterize the various MR scanners, to record their performance over the course of the study and to detect possible malfunctions at an early stage. It also assessed the quality of the human MRI data systematically to characterize its influence on various analyses. Furthermore, by setting up and publishing this protocol, we define standards that must be considered when analyzing data from the BipoLife consortium. It further promotes a systematic evaluation of data quality and a definition of subject inclusion criteria. In the long term, it will help to increase the chance of achieving clinically relevant results.

MRI-Based Deep Learning for Differentiating Between Bipolar and Major Depressive Disorders

Mood disorders, particularly bipolar disorder (BD) and major depressive disorder (MDD), manifest changes in brain structure that can be detected using structural magnetic resonance imaging (MRI). Although structural MRI is a promising diagnostic tool, prevailing diagnostic criteria for BD and MDD are predominantly subjective, sometimes leading to misdiagnosis. This challenge is compounded by a limited understanding of the underlying causes of these disorders. In response, we present SE-ResNet, a Residual Network (ResNet)-based framework designed to discriminate between BD, MDD, and healthy controls (HC) using structural MRI data. Our approach extends the traditional Squeeze-and-Excitation (SE) layer by incorporating a dedicated branch for spatial attention map generation, equipped with soft-pooling, a 7 × 7 convolution, and a sigmoid function, intended to detect complex spatial patterns. The fusion of channel and spatial attention maps through element-wise addition aims to enhance the model's ability to discriminate features. Unlike conventional methods that use max-pooling for downsampling, our methodology employs soft-pooling, which aims to preserve a richer representation of input features and reduce data loss. When evaluated on a proprietary dataset comprising 303 subjects, the SE-ResNet achieved an accuracy of 85.8%, a recall of 85.7%, a precision of 85.9%, and an F1 score of 85.8%. These performance metrics suggest that the SE-ResNet framework has potential as a tool for detecting psychiatric disorders using structural MRI data.

Reduced pineal gland volume in patients with obsessive-compulsive disorder.

Patients with obsessive-compulsive disorder (OCD) can have hyperactivity of the hypothalamic-pituitary-adrenal axis and may have increased secretion of adrenocorticotropic hormone and cortisol and reduced secretion of melatonin. Examination of pineal gland volumes in patients with OCD compared to healthy controls, thus, is an important consideration and the focus of this study. A total of 20 patients with OCD and 20 healthy controls were enrolled. Demographic and clinical characteristics of participants were assessed, and structural magnetic resonance imaging was performed. Patients with OCD had a statistically significant smaller pineal gland volumes compared to healthy controls. In this initial small study, patients with OCD exhibited smaller pineal gland volumes compared to healthy control subjects. While this finding suggests a potential link between the pineal gland and OCD pathophysiology, further research with larger sample sizes and measurement of hormonal changes are necessary.

3-1-3 Weight averaging technique-based performance evaluation of deep neural networks for Alzheimer’s disease detection using structural MRI

Alzheimer’s disease (AD) is a progressive neurological disorder. It is identified by the gradual shrinkage of the brain and the loss of brain cells. This leads to cognitive decline and impaired social functioning, making it a major contributor to dementia. While there are no treatments to reverse AD’s progression, spotting the disease’s onset can have a significant impact in the medical field. Deep learning (DL) has revolutionized medical image classification by automating feature engineering, removing the requirement for human experts in feature extraction. DL-based solutions are highly accurate but demand a lot of training data, which poses a common challenge. Transfer learning (TL) has gained attention for its knack for handling limited data and expediting model training. This study uses TL to classify AD using T1-weighted 3D Magnetic Resonance Imaging (MRI) from the Alzheimer’s Disease Neuroimaging (ADNI) database. Four modified pre-trained deep neural networks (DNN), VGG16, MobileNet, DenseNet121, and NASNetMobile, are trained and evaluated on the ADNI dataset. The 3-1-3 weight averaging technique and fine-tuning improve the performance of the classification models. The evaluated accuracies for AD classification are VGG16: 98.75%; MobileNet: 97.5%; DenseNet: 97.5%; and NASNetMobile: 96.25%. The receiver operating characteristic (ROC), precision–recall (PR), and Kolmogorov-Smirnov (KS) statistic plots validate the effectiveness of the modified pre-trained model. Modified VGG16 excels with area under the curve (AUC) values of 0.99 for ROC and 0.998 for PR curves. The proposed approach shows effective AD classification by achieving high accuracy using the 3-1-3 weight averaging technique and fine-tuning.