IntroductionExplainable Artificial Intelligence (XAI) methods enhance the diagnostic efficiency of clinical decision support systems by making the predictions of a convolutional neural networks (CNN) on brain imaging more transparent and trustworthy. However, their clinical adoption is limited due to limited validation of the explanation quality. Our study introduces a framework that evaluates XAI methods by integrating neuroanatomical morphological features with CNN-generated relevance maps for disease classification. MethodsWe trained a CNN using brain MRI scans from six cohorts: ADNI, AIBL, DELCODE, DESCRIBE, EDSD, and NIFD (N=3253), including participants that were cognitively normal, with amnestic mild cognitive impairment, dementia due to Alzheimers disease and frontotemporal dementia. Clustering analysis benchmarked different explanation space configurations by using morphological features as proxy-ground truth. We implemented three post-hoc explanations methods: i) by simplifying model decisions, ii) explanation-by-example, and iii) textual explanations. A qualitative evaluation by clinicians (N=6) was performed to assess their clinical validity. ResultsClustering performance improved in morphology enriched explanation spaces, improving both homogeneity and completeness of the clusters. Post hoc explanations by model simplification largely delineated converters and stable participants, while explanation-by-example presented possible cognition trajectories. Textual explanations gave rule-based summarization of pathological findings. Clinicians qualitative evaluation highlighted challenges and opportunities of XAI for different clinical applications. ConclusionOur study refines XAI explanation spaces and applies various approaches for generating explanations. Within the context of AI-based decision support system in dementia research we found the explanations methods to be promising towards enhancing diagnostic efficiency, backed up by the clinical assessments.

Alzheimer disease and related dementias (ADRD) are complex disorders with overlapping symptoms and pathologies. Comprehensive records of symptoms in electronic health records (EHRs) are critical for not only reaching an accurate diagnosis but also supporting ongoing research studies and clinical trials. However, these symptoms are frequently obscured within unstructured clinical notes in EHRs, making manual extraction both time-consuming and labor-intensive. We aimed to automate symptom extraction from the clinical notes of patients with ADRD using fine-tuned large language models (LLMs), compare its performance to regular expression-based symptom recognition, and validate the results using brain magnetic resonance imaging (MRI) data. We fine-tuned LLMs to extract ADRD symptoms across the following 7 domains: memory, executive function, motor, language, visuospatial, neuropsychiatric, and sleep. We assessed the algorithm's performance by calculating the area under the receiver operating characteristic curve (AUROC) for each domain. The extracted symptoms were then validated in two analyses: (1) predicting ADRD diagnosis using the counts of extracted symptoms and (2) examining the association between ADRD symptoms and MRI-derived brain volumes. Symptom extraction across the 7 domains achieved high accuracy with AUROCs ranging from 0.97 to 0.99. Using the counts of extracted symptoms to predict ADRD diagnosis yielded an AUROC of 0.83 (95% CI 0.77-0.89). Symptom associations with brain volumes revealed that a smaller hippocampal volume was linked to memory impairments (odds ratio 0.62, 95% CI 0.46-0.84; P=.006), and reduced pallidum size was associated with motor impairments (odds ratio 0.73, 95% CI 0.58-0.90; P=.04). These results highlight the accuracy and reliability of our high-throughput ADRD phenotyping algorithm. By enabling automated symptom extraction, our approach has the potential to assist with differential diagnosis, as well as facilitate clinical trials and research studies of dementia.

One in ten children experiences sleep-disordered breathing (SDB). Untreated SDB is associated with poor cognition, but the underlying mechanisms are less understood. We assessed the relationship between magnetic resonance imaging (MRI)-derived upper airway volume and children's cognition and regional cortical gray matter volumes. We used five-year data from the Adolescent Brain Cognitive Development study (n=11,875 children, 9-10 years at baseline). Upper airway volumes were derived using a deep learning model applied to 5,552,640 brain MRI slices. The primary outcome was the Total Cognition Composite score from the National Institutes of Health Toolbox (NIH-TB). Secondary outcomes included other NIH-TB measures and cortical gray matter volumes. The habitual snoring group had significantly smaller airway volumes than non-snorers (mean difference=1.2 cm³; 95% CI, 1.0-1.4 cm³; P<0.001). Deep learning-derived airway volume predicted the Total Cognition Composite score (estimated mean difference=3.68 points; 95% CI, 2.41-4.96; P<0.001) per one-unit increase in the natural log of airway volume (~2.7-fold raw volume increase). This airway volume increase was also associated with an average 0.02 cm³ increase in right temporal pole volume (95% CI, 0.01-0.02 cm³; P<0.001). Similar airway volume predicted most NIH-TB domain scores and multiple frontal and temporal gray matter volumes. These brain volumes mediated the relationship between airway volume and cognition. We demonstrate a novel application of deep learning-based airway segmentation in a large pediatric cohort. Upper airway volume is a potential biomarker for cognitive outcomes in pediatric SDB, offers insights into neurobiological mechanisms, and informs future studies on risk stratification. This article is open access and distributed under the terms of the Creative Commons Attribution Non-Commercial No Derivatives License 4.0 (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Sodium neuroimaging provides unique insights into the cellular and metabolic properties of brain tumors. However, at 3T, sodium neuroimaging MRI's low signal-to-noise ratio (SNR) and resolution discourages routine clinical use. We evaluated the recently developed Anatomically constrained GAN using physics-based synthetic MRI artifacts" (ATHENA) for high-resolution sodium neuroimaging of brain tumors at 3T. We hypothesized the model would improve the image quality while preserving the inherent sodium information. 4,573 proton MRI scans from 1,390 suspected brain tumor patients were used for training. Sodium and proton MRI datasets from Twenty glioma patients were collected for validation. Twenty-four image-guided biopsies from seven patients were available for sodium-proton exchanger (NHE1) expression evaluation on immunohistochemistry. High-resolution synthetic sodium images were generated using the ATHENA model, then compared to native sodium MRI and NHE1 protein expression from image-guided biopsy samples. The ATHENA produced synthetic-sodium MR with significantly improved SNR (native SNR 18.20 ± 7.04; synthetic SNR 23.83 ± 9.33, P = 0.0079). The synthetic-sodium values were consistent with the native measurements (P = 0.2058), with a strong linear correlation within contrast-enhancing areas of the tumor (R² = 0.7565, P = 0.0005), T2-hyperintense (R² = 0.7325, P < 0.0001), and necrotic areas (R² = 0.7678, P < 0.0001). The synthetic-sodium MR and the relative NHE1 expression from image-guided biopsies were better correlated for the synthetic (ρ = 0.3269, P < 0.0001) than the native (ρ = 0.1732, P = 0.0276) with higher sodium signal in samples expressing elevated NHE1 (P < 0.0001). ATHENA generates high-resolution synthetic-sodium MRI at 3T, enabling clinically attainable multinuclear imaging for brain tumors that retain the inherent information from the native sodium. The resulting synthetic sodium significantly correlates with tissue expression, potentially supporting its utility as a non-invasive marker of underlying sodium homeostasis in brain tumors.

To develop and validate a deep-learning-based automatic method for vessel walls and atherosclerotic plaques segmentation for quantitative evaluation in MR vessel wall images. A total of 193 patients (107 patients for training and validation, 39 patients for internal test, 47 patients for external test) with atherosclerotic plaque from five centers underwent T1-weighted MRI scans and were included in the dataset. The first step of the proposed method was constructing a purely learning-based convolutional neural network (CNN) named Vessel-SegNet to segment the lumen and the vessel wall. The second step is using the vessel wall priors (including manual prior and Tversky-loss-based automatic prior) to improve the plaque segmentation, which utilizes the morphological similarity between the vessel wall and the plaque. The Dice similarity coefficient (DSC), intraclass correlation coefficient (ICC), etc., were used to evaluate the similarity, agreement, and correlations. Most of the DSCs for lumen and vessel wall segmentation were above 90%. The introduction of vessel wall priors can increase the DSC for plaque segmentation by over 10%, reaching 88.45%. Compared to dice-loss-based vessel wall priors, the Tversky-loss-based priors can further improve DSC by nearly 3%, reaching 82.84%. Most of the ICC values between the Vessel-SegNet and manual methods in the 6 quantitative measurements are greater than 85% (p-value < 0.001). The proposed CNN-based segmentation model can quickly and accurately segment vessel walls and plaques for quantitative evaluation. Due to the lack of testing with other equipment, populations, and anatomical studies, the reliability of the research results still requires further exploration. Question How can the accuracy and efficiency of vessel component segmentation for quantification, including the lumen, vessel wall, and plaque, be improved? Findings Improved CNN models, manual/automatic vessel wall priors, and Tversky loss can improve the performance of semi-automatic/automatic vessel components segmentation for quantification. Clinical relevance Manual segmentation of vessel components is a time-consuming yet important process. Rapid and accurate segmentation of the lumen, vessel walls, and plaques for quantification assessment helps patients obtain more accurate, efficient, and timely stroke risk assessments and clinical recommendations.

Concurrent neurodegenerative and vascular pathologies pose a diagnostic challenge in the clinical setting, with histopathology remaining the definitive modality for dementia-type diagnosis. To address this clinical challenge, we introduce a neuropathology-based, data-driven, multi-label deep-learning framework to identify and quantify in vivo biomarkers for Alzheimer's disease (AD), vascular dementia (VD) and Lewy body dementia (LBD) using antemortem T1-weighted MRI scans of 423 demented and 361 control participants from National Alzheimer's Coordinating Center and Alzheimer's Disease Neuroimaging Initiative datasets. Based on the best-performing deep-learning model, explainable heat maps were extracted to visualize disease patterns, and the novel Deep Signature of Pathology Atrophy REcognition (DeepSPARE) indices were developed, where a higher DeepSPARE score indicates more brain alterations associated with that specific pathology. A substantial discrepancy in clinical and neuropathological diagnosis was observed in the demented patients: 71% had more than one pathology, but 67% were diagnosed clinically as AD only. Based on these neuropathological diagnoses and leveraging cross-validation principles, the deep-learning model achieved the best performance, with a balanced accuracy of 0.844, 0.839 and 0.623 for AD, VD and LBD, respectively, and was used to generate the explainable deep-learning heat maps and DeepSPARE indices. The explainable deep-learning heat maps revealed distinct neuroimaging brain alteration patterns for each pathology: (i) the AD heat map highlighted bilateral hippocampal regions; (ii) the VD heat map emphasized white matter regions; and (iii) the LBD heat map exposed occipital alterations. The DeepSPARE indices were validated by examining their associations with cognitive testing and neuropathological and neuroimaging measures using linear mixed-effects models. The DeepSPARE-AD index was associated with Mini-Mental State Examination, the Trail Making Test B, memory, hippocampal volume, Braak stages, Consortium to Establish a Registry for Alzheimer's Disease (CERAD) scores and Thal phases [false-discovery rate (FDR)-adjusted P < 0.05]. The DeepSPARE-VD index was associated with white matter hyperintensity volume and cerebral amyloid angiopathy (FDR-adjusted P < 0.001), and the DeepSPARE-LBD index was associated with Lewy body stages (FDR-adjusted P < 0.05). The findings were replicated in an out-of-sample Alzheimer's Disease Neuroimaging Initiative dataset by testing associations with cognitive, imaging, plasma and CSF measures. CSF and plasma tau phosphorylated at threonine-181 (pTau181) were significantly associated with DeepSPARE-AD in the AD and mild cognitive impairment amyloid-β positive (AD/MCIΑβ+) group (FDR-adjusted P < 0.001), and CSF α-synuclein was associated solely with DeepSPARE-LBD (FDR-adjusted P = 0.036). Overall, these findings demonstrate the advantages of our innovative deep-learning framework in detecting antemortem neuroimaging signatures linked to different pathologies. The newly deep-learning-derived DeepSPARE indices are precise, pathology-sensitive and single-valued non-invasive neuroimaging metrics, bridging the traditional widely available in vivo T1 imaging with histopathology.

The ability to predict the spreading of pathology in patients with frontotemporal dementia (FTD) is crucial for early diagnosis and targeted interventions. In this study, we examined the relationship between network vulnerability and longitudinal progression of atrophy in FTD patients, using the network diffusion model (NDM) of the spread of pathology. Thirty behavioural variant FTD (bvFTD), 13 semantic variant primary progressive aphasia (svPPA), 14 non-fluent variant primary progressive aphasia (nfvPPA) and 12 semantic behavioural variant FTD (sbvFTD) patients underwent longitudinal T1-weighted MRI. Fifty young controls (20-31 years of age) underwent multi-shell diffusion MRI scan. An NDM was developed to model progression of FTD pathology as a spreading process from a seed through the healthy structural connectome, using connectivity measures from fractional anisotropy and intracellular volume fraction in young controls. Four disease epicentres were initially identified from the peaks of atrophy of each FTD variant: left insula (bvFTD), left temporal pole (svPPA), right temporal pole (sbvFTD) and left supplementary motor area (nfvPPA). Pearson's correlations were calculated between NDM-predicted atrophy in young controls and the observed longitudinal atrophy in FTD patients over a follow-up period of 24 months. The NDM was then run for all 220 brain seeds to verify whether the four epicentres were among those that yielded the highest correlation. Using the NDM, predictive maps in young controls showed progression of pathology from the peaks of atrophy in svPPA, nfvPPA and sbvFTD over 24 months. svPPA exhibited early involvement of the left temporal and occipital lobes, progressing to extensive left hemisphere impairment. nfvPPA and sbvFTD spread in a similar manner bilaterally to frontal, sensorimotor and temporal regions, with sbvFTD additionally affecting the right hemisphere. Moreover, the NDM-predicted atrophy of each region was positively correlated with longitudinal real atrophy, with a greater effect in svPPA and sbvFTD. In bvFTD, the model starting from the left insula (the peak of atrophy) demonstrated a highly left-lateralized pattern, with pathology spreading to frontal, sensorimotor, temporal and basal ganglia regions, with minimal extension to the contralateral hemisphere by 24 months. However, unlike the atrophy peaks observed in the other three phenotypes, the left insula did not show the strongest correlation between the estimated and real atrophy. Instead, the bilateral superior frontal gyrus emerged as optimal seeds for modelling atrophy spread, showing the highest correlation ranking in both hemispheres. Overall, NDM applied on the intracellular volume fraction connectome yielded higher correlations relative to NDM applied on fractional anisotropy maps. The NDM implementation using the cross-sectional structural connectome is a valuable tool to predict patterns of atrophy and spreading of pathology in FTD clinical variants.

Despite decades of advancements in diagnostic MRI, 30%-50% of temporal lobe epilepsy (TLE) patients remain categorized as 'non-lesional' (i.e. MRI negative) based on visual assessment by human experts. MRI-negative patients face diagnostic uncertainty and significant delays in treatment planning. Quantitative MRI studies have demonstrated that MRI-negative patients often exhibit a TLE-specific pattern of temporal and limbic atrophy that might be too subtle for the human eye to detect. This signature pattern could be translated successfully into clinical use via advances in artificial intelligence in computer-aided MRI interpretation, thereby improving the detection of brain 'lesional' patterns associated with TLE. Here, we tested this hypothesis by using a three-dimensional convolutional neural network applied to a dataset of 1178 scans from 12 different centres, which was able to differentiate TLE from healthy controls with high accuracy (85.9% ± 2.8%), significantly outperforming support vector machines based on hippocampal (74.4% ± 2.6%) and whole-brain (78.3% ± 3.3%) volumes. Our analysis focused subsequently on a subset of patients who achieved sustained seizure freedom post-surgery as a gold standard for confirming TLE. Importantly, MRI-negative patients from this cohort were accurately identified as TLE 82.7% ± 0.9% of the time, an encouraging finding given that clinically these were all patients considered to be MRI negative (i.e. not radiographically different from controls). The saliency maps from the convolutional neural network revealed that limbic structures, particularly medial temporal, cingulate and orbitofrontal areas, were most influential in classification, confirming the importance of the well-established TLE signature atrophy pattern for diagnosis. Indeed, the saliency maps were similar in MRI-positive and MRI-negative TLE groups, suggesting that even when humans cannot distinguish more subtle levels of atrophy, these MRI-negative patients are on the same continuum common across all TLE patients. As such, artificial intelligence can identify TLE lesional patterns, and artificial intelligence-aided diagnosis has the potential to enhance the neuroimaging diagnosis of TLE greatly and to redefine the concept of 'lesional' TLE.

Multi-modal brain magnetic resonance imaging (MRI) plays a crucial role in clinical diagnostics by providing complementary information across different imaging modalities. However, a common challenge in clinical practice is missing MRI modalities. In this paper, we apply SwinUNETR to the synthesize of missing modalities in brain MRI. SwinUNETR is a novel neural network architecture designed for medical image analysis, integrating the strengths of Swin Transformer and convolutional neural networks (CNNs). The Swin Transformer, a variant of the Vision Transformer (ViT), incorporates hierarchical feature extraction and window-based self-attention mechanisms, enabling it to capture both local and global contextual information effectively. By combining the Swin Transformer with CNNs, SwinUNETR merges global context awareness with detailed spatial resolution. This hybrid approach addresses the challenges posed by the varying modality characteristics and complex brain structures, facilitating the generation of accurate and realistic synthetic images. We evaluate the performance of SwinUNETR on brain MRI datasets and demonstrate its superior capability in generating clinically valuable images. Our results show significant improvements in image quality, anatomical consistency, and diagnostic value.

The thalamus plays a crucial role in sensory processing, emotional regulation, and cognitive functions, and its dysregulation may be implicated in psychosis. The aim of the present study was to examine the differences in thalamic subregional volumes between pediatric bipolar disorder patients with (P-PBD) and without psychotic symptoms (NP-PBD). Participants including 28 P-PBD, 26 NP-PBD, and 18 healthy controls (HCs) underwent structural magnetic resonance imaging (sMRI) scanning using a 3.0T MRI scanner. All T1-weighted imaging data were processed by FreeSurfer 7.4.0 software. The volumetric differences of thalamic subregions among three groups were compared by using analyses of covariance (ANCOVA) and post-hoc analyses. Additionally, we applied a standard support vector classification (SVC) model for pairwise comparison among the three groups to identify brain regions with significant volumetric differences. The ANCOVA revealed that significant volumetric differences were observed in the left pulvinar anterior (L_PuA) and left reuniens medial ventral (L_MV-re) thalamus among three groups. Post-hoc analysis revealed that patients with P-PBD exhibited decreased volumes in the L_PuA and L_MV-re when compared to the NP-PBD group and HCs, respectively. Furthermore, the SVC model revealed that the L_MV-re volume exhibited the best capacity to discriminate P-PBD from NP-PBD and HCs. The present findings demonstrated that reduced thalamic subregional volumes in the L_PuA and L_MV-re might be associated with psychotic symptoms in PBD.