Frontiers in Neuroinformatics

EegFun.jl is an open-source package for electroencephalography (EEG) analysis implemented in the Julia programming language. EegFun.jl provides a flexible framework for EEG research, covering data import from standard file formats, filtering and re-referencing, Independent Component Analysis (ICA) for artifact detection/correction, epoch extraction, and ERP averaging and visualisation. The Julia language provides the readability of a high-level scripting environment together with execution speeds comparable to compiled code. EegFun.jl combines interactive data visualization with high-performance execution, making large-scale analyses both efficient and easy. Here, we provide a brief overview and introductory tutorial of the core stages of the EEG analysis workflow to illustrate the packages capabilities. The package is freely available under the MIT license.

Purpose: Attenuation correction (AC) of PET images is essential for accurate quantification. Brain PET studies comprising simultaneous EEG (PETEEG) may suffer from metal artifacts in CT images (CTEEG), or improper correction when electrodes are not present in the CT (CT0). As these influences are not well-characterized, we aim to compare metal artifact reduction (MAR) techniques for CTEEG images, and evaluate differences between attenuated-corrected PETEEG using CT0 and CTEEG with MAR, synthetically placed electrodes (CTEEG-synth) and extended Hounsfield unit (HU) range. Methods: 19 healthy participants underwent two total-body PET/CT scans with [18F]FDG, with and without 32 EEG scalp electrodes, respectively. We evaluated five MARs to reduce streaks caused by the EEG electrodes in the CTEEG. Finally, CT0, CTEEG with (CTEEG-iMAR-Ext) and without extended HU range (CTEEG-iMAR) and CTEEG-synth were used to perform attenuation correction of PETEEG. We compared our results to PET0/CT0 scan using relative differences. Results: CTEEG and CTEEG-iMAR showed the smallest differences to CT0. PETEEG/CTEEG-iMAR-Ext exhibited the lowest differences to PET0/CT0 (average bias across all regions of -0.46%), followed by similar performance of PETEEG/CTEEG-iMAR (-0.73%) and PETEEG/CTEEG (-0.76%). Conversely, PETEEG/CT0 demonstrated the largest average differences (-1.81%), with values reaching -2.71% in the parietal lobe. These differences were consistent across subjects, yielding significant effects in most of the brain (pFWE < 0.05). CTEEG-synth performed not as good as CTEEG (-1.21%). Conclusions: CTEEG with extended HU range is most suitable for attenuation correction of PETEEG images, with MAR correction offering little additional improvement.

Traditional fMRI studies rely on predefined task paradigms, where fixed stimulus designs limit the flexibility with which brain-stimulus relationships can be explored. Here, we introduce Reinforcement Learning via Brain Feedback (RLBF), a framework and open-source software package for adaptive stimulus optimization using real-time fMRI. RLBF reverses the conventional direction of inference by using neural responses to guide the exploration of stimulus spaces through reinforcement learning, enabling optimization of predefined brain targets such as regional activity or multivariate neural signatures. The accompanying Python-based software provides a modular framework integrating real-time fMRI data processing, reinforcement learning agents, adaptive stimulus generation, simulation-based testing, and experiment monitoring. Its flexible architecture allows researchers to customize preprocessing pipelines, reward functions, stimulus spaces, and RL strategies for diverse closed-loop neuroimaging applications. We validate the framework in a proof-of-concept study (N=10), demonstrating real-time optimization of a simple visual stimulus space by adapting checkerboard contrast and frequency to maximize primary visual cortex (V1) responses within a single 10-minute fMRI session. RLBF provides an extensible foundation for brain-guided stimulus optimization and enables new approaches for investigating neural specificity, individualized brain-stimulus relationships, and adaptive experimental design.

Dynamic positron emission tomography (PET) with long axial field-of-view (LAFOV) scanners enables multi-organ imaging and kinetic quantification beyond static (late-phase) imaging; however, the long times typically required for dynamic acquisitions remain clinically impractical. This study evaluates a deep learning (DL) framework to enable abbreviated dynamic PET acquisitions, comparing single-time-window (STW, early dynamic data only) and dual-time-window (DTW, early dynamic data plus a late 5-min static frame) protocols with early dynamic scan durations of 5-30 min and dose levels ranging from 360 MBq to 18 MBq. Seventeen 60-min dynamic [18F]FDG datasets were first motion-corrected using a staggered FALCON pipeline and then used to train and test a spatiotemporal DL model for autoregressive frame prediction. Performance was assessed across the full quantitative workflow, from DL-predicted frames and time-activity curves to organ-based kinetic modeling and voxel-wise parametric imaging in multiple tissues and two patient cohorts. DTW protocols consistently outperformed STW, better preserving late-phase kinetics. For a 15-min early dynamic scan, adding a late 5-min scan reduced mean absolute Ki difference from 23% (STW) to 17% (DTW) in the liver and from 26% to 15% in the thalamus. DTW + DL further reduced errors to [&le;]10% in the liver, thalamus, and breast lesion, and 16% in muscle. Our recommended protocol, 15-min early dynamic scan plus a 5-min late scan with DL, remained robust to up to a 5-fold dose reduction (~74 MBq). Overall, these findings support DL-enabled abbreviated, low-dose dynamic LAFOV PET as a clinically feasible approach for accurate kinetic quantification

Intracranial electroencephalography (iEEG) provides an unprecedented opportunity to directly record neural activity and causally perturb the human brain through electrical stimulation. Yet, the increasingly collaborative nature and complexity of modern iEEG studies pose substantial challenges for experimental control, data quality, and standardization. Unlike most experimental modalities, human iEEG data are acquired within dynamic clinical environments, where patient condition, recording quality, hardware configuration, and experimental protocols may vary across recording sessions and collaborating sites. The resulting heterogeneity creates opportunities for technical and procedural failures that often remain undetected until downstream analyses, when corrective action is no longer possible. Here, we present a framework for standardized session-level quality assurance in human iEEG research and provide an open-source implementation compatible with Brain Imaging Data Structure (BIDS)-organized datasets. The framework defines four complementary domains of quality assessment crucial for human iEEG studies: protocol fidelity, behavioral integrity, stimulation validation, and signal quality. These domains integrate electrophysiological recordings, behavioral event logs, and stimulation metadata to verify data completeness, confirm participant engagement, validate stimulation delivery, and identify potentially compromised recording channels. Automated quality metrics and standardized diagnostic visualizations are generated following each testing session, enabling rapid identification of technical and procedural failures while corrective action is still possible. By providing a standardized approach to session-level quality assurance, the framework improves data integrity, enhances reproducibility, facilitates analyst training, and supports harmonized data collection across laboratories and clinical sites.

Despite the growing use of functional near-infrared spectroscopy (fNIRS) hyperscanning to record brain activity simultaneously from interacting individuals in naturalistic settings, most analyses quantify functional connectivity separately for each channel pair. The resulting collection of pairwise estimates is difficult to integrate into a network-level characterization of intra- and inter-brain organization. Here, we present an open, configuration-driven Python toolkit that transforms preprocessed fNIRS hyperscanning time series into functional connectivity graphs. The toolkit constructs a bipartite inter-brain network for each dyad and separate intra-brain networks for each participant, computes node- and graph-level measures, and exports adjacency matrices, edge lists, analysis-ready summary tables, reproducibility metadata, and standardized visualizations. Dataset-specific parameters, including directory structure, participant naming, channel selection, epoch extraction, and edge-retention criteria, are defined in a human-readable YAML configuration file, enabling the same workflow to accommodate differently organized datasets without changes to the source code. We illustrate the pipeline using a representative recording from a mother-infant fNIRS hyperscanning dataset and present the resulting network outputs. The toolkit provides a reproducible framework for moving from pairwise functional connectivity estimates to network-level analyses of dyadic and individual brain organization.

Brain-computer interface (BCI) research relies on multistage computational pipelines, but progress has been slowed by fragmented data formats, heterogeneous decoder implementations and hardware-specific deployment toolchains. Here, we introduce BCIJelly, a unified ecosystem that standardizes 18 BCI datasets into AI-ready inputs and integrates 15 benchmark decoders, 80 reusable modules, automated architecture search (AAS) and hardware-aware neuromorphic deployment. Our AAS constructs task-specific decoders without manual design and extends into a large language model (LLM)-driven closed-loop mode supporting single-task, multitask and cross-species decoder design. A single-command pipeline compiles trained decoders for neuromorphic hardware, reducing power consumption by 30 to 50 times while preserving decoding performance. An interactive visualization software enables code-free exploration of neural recordings and decoding outputs. BCIJelly is validated across five BCI paradigms (motor, visual, speech, emotion and auditory) in humans, macaques and mice, providing an extensible ecosystem connecting data standardization, decoder development, systematic evaluation and hardware-aware deployment for BCI research.

Purpose Open-source vision-language models (VLMs) can be locally deployed without external internet access, potentially enhancing data security. This study compared the diagnostic performance of general-purpose and medical-purpose open-source VLMs and evaluated their ability to characterize brain metastases on contrast-enhanced (CE) MRI. Materials and Methods Sixty lesion-positive axial CE T1-weighted images and sixty matched lesion-negative images from 60 patients were analyzed using three general-purpose VLMs-InternVL3-8B, Qwen2.5-VL-7B-Instruct, and MiniCPM-V-4.5-and three medical-purpose VLMs-MedGemma-4B-it, LLaVA-Med v1.5, and HuatuoGPT-Vision-7B. Lesion detection performance was assessed using sensitivity, specificity, and balanced accuracy. On lesion-positive images, accuracy was evaluated for lesion count, laterality, anatomic location, enhancement pattern, necrosis, vasogenic edema, and mass effect. Model differences were assessed using Cochran's Q tests followed by pairwise McNemar tests with Benjamini-Hochberg correction. Results The median age of the study patients was 67 years (IQR, 61.0-70.5 years), and 35 patients were male (58.3%). MiniCPM-V-4.5 showed the most balanced diagnostic performance, with a sensitivity of 78.3% (95% CI, 66.4-86.9%) and a specificity of 85.0% (95% CI, 73.9-91.9%), and significantly higher balanced accuracy than all other models. Significant overall differences were observed for lesion count, laterality, location, enhancement pattern, necrosis, and mass effect, but not for vasogenic edema (FDR-adjusted P = 0.056). HuatuoGPT-Vision-7B and MedGemma-4B-it showed relatively consistent accuracy across multiple image assessment tasks, although their performance remained modest. Conclusion Our study demonstrated substantial heterogeneity in the performance of open-source VLMs in brain metastasis evaluation, and medical-purpose VLMs did not outperform general-purpose VLMs.

Purpose: Respiratory motion remains a major source of quantitative bias in PET and becomes increasingly relevant for high-sensitivity long axial field-of-view (LAFOV) PET/CT. Although numerous respiratory motion correction (MoCo) methods have been proposed, their quantitative accuracy cannot be established clinically because a patient-specific motion-free reference is fundamentally unavailable in vivo. This study combined clinical PET imaging with a digital twin, a realistic representation of both the PET/CT system and the patient, to objectively validate respiratory MoCo against a corresponding motion-free reference. Methods: Twenty patients (10 [18F]FDG with predominantly pulmonary lesions and 10 [18F]SiFAlin-TATE with predominantly hepatic lesions; total 135 lesions) were analyzed. The digital twin combined a validated LAFOV PET/CT simulation model with an anatomically realistic phantom containing 14 lung and liver lesions, two patient-derived respiratory patterns, and respiratory motion amplitudes of 2 and 3 cm, generating patient-like datasets with corresponding motion-free references. Data-driven and image-based MoCo were evaluated using lesion morphology, SUVmean, SUVmax, and metabolic tumor volume (MTV). Results: In patients, data-driven MoCo produced larger SUVmean increases than image-based MoCo for liver (48.1{+/-}18.9% vs. 17.0 {+/-} 12.0%; p<0.01), lower-lung (32.5{+/-}21.2% vs. 16.3{+/-}15.6%, p=0.06), and upper-lung lesions (28.4{+/-}32.0% vs. 10.4 {+/-} 17.2%; p<0.01), with similar findings for SUVmax and larger MTV reductions. Simulation revealed marked motion-induced SUVmean underestimation before correction, particularly in liver (-31.2{+/-}6.8%) and lower lung (-15.5{+/-}13.9%). Relative to the motion-free reference, data-driven MoCo most accurately recovered hepatic uptake (4.3{+/-}11.7% vs. -10.0 {+/-} 9.2%; p=0.01) but overestimated pulmonary uptake (lower lung: 19.8{+/-}16.3% vs. -1.6 {+/-} 10.2%; p=0.02). SUVmax showed the same regional behavior, whereas image-based MoCo yielded MTV estimates closer to the reference. Quantitative recovery was largely independent of respiratory pattern, while larger motion amplitudes mainly affected image-based MoCo. Conclusion: Combining clinical PET with a realistic digital twin and corresponding motion-free ground truth enabled objective validation of respiratory MoCo beyond conventional clinical evaluation. Larger correction-induced quantitative changes should not be equated with greater quantitative accuracy. Instead, MoCo performance was region- and metric-dependent, highlighting the value of ground-truth-based validation for developing and benchmarking respiratory motion correction and quantitative PET on LAFOV PET/CT systems.

Background: Precise localisation of intracranial electrode contacts is essential for the interpretation of stereoelectroencephalography recordings and planning epilepsy surgery. In current clinical practice, this is typically a manual process, which is time-consuming and prone to variability. Existing automated solutions are often fragmented across multiple tools requiring technical expertise, limiting their adoption in routine clinical workflows. This study presents an open-source extension for 3D Slicer that provides an integrated, user-friendly standalone solution for the direct automatic detection of electrode contacts within a widely used medical imaging platform. Results: The proposed method combines anchor bolt-based initialisation, probabilistic segmentation of electrode structures, and non-linear modelling to precisely track true electrode trajectories. The approach was evaluated on a dataset comprising 78 cases from 73 patients, including 1,078 electrodes with 14,480 contacts. The method achieved high localisation accuracy, with a median (interquartile range) deviation of 0.10 (0.06, 0.15) mm. Only 7/1078 (0.65%) electrodes required manual correction; these specific cases were handled using tools provided within the proposed extension. Conclusions: The presented extension enables fast, accurate, and reproducible electrode contact localisation within a single integrated environment. By combining automation with intuitive user interaction, it significantly reduces processing time while maintaining clinical reliability. The tool's free availability as an extension in 3D Slicer lowers the barrier to adoption and supports the standardisation of workflows across clinical and research centres.

Head-fixed behavior in rodents is a foundational technique in systems neuroscience which enables use of sophisticated imaging techniques in combination with animal behavior. However, accessibility of head-fixed behavior techniques is limited. Animal training consumes a large amount of experimenter labor and commercial setups, when available, are largely inflexible and financially burdensome. Here, we present a low-cost, modular, and open-source hardware and software implementation for head-fixed rodent decision-making tasks. Our design lowers experimenter labor and enables large teams of researchers to participate in animal training with minimal experimenter error using a simple touchscreen GUI and automated training progression. We demonstrate the efficacy of the platform by training a cohort of animals in a two-choice probabilistic rapid-reversal task in which mice continuously update action choices based on recent reward history. The presented design lowers the barrier to entry for laboratories seeking to conduct head-fixed rodent behavior and provides modular solutions for developing custom rigs based on experimental demands.

Automated lesion segmentation is increasingly central to PSMA PET/CT interpretation, supporting staging, treatment planning, and response assessment at a scale that outpaces available nuclear-medicine expertise. However, automated PSMA-PET/CT whole-body lesion segmentation models are trained on images alone, with no knowledge of where in the body prostate metastases actually tend to occur. Radiologists use clinical domain knowledge of metastatic spread, but its absence in machine learning models produces false positives in anatomically implausible locations and missed lesions in high-risk sites such as the liver. In this work, we explore whether population-level spatial knowledge of metastatic spread can be used to augment deep learning segmentation predictions, and how such a prior should be fused with a network's output, without additional training. We build a data-driven metastasis atlas from 375 expert-annotated whole-body PSMA PET/CT scans and investigate its fusion with a trained segmentation network under a Bayesian framework, in which prediction probabilities from an nnU-Net-based lesion segmentation model serve as the likelihood and the data-driven atlas as the prior. Because metastases occupy only a small fraction of whole-body voxels, the atlas's peak probability is too low, and standard power-scaled or naive Bayesian pooling references lack the tools to deal with this shortcoming. This causes these standard fusion strategies to fail and, in the naive Bayesian case, to sharply degrade performance. We instead derive a calibrated, background-referenced log-odds fusion, one of many possible approaches to combine a population atlas with a deep learning model's predictions, distinct from classical multi-atlas label fusion in that it fuses a single population prior with a trained network's softmax rather than combining several registered atlases. Furthermore, this approach is neutral outside atlas support by construction, reduces exactly to the baseline network when unweighted, and requires no retraining. This atlas fusion significantly improved mean Dice over the baseline nnU-Net on a disjoint internal test set ($+0.011$, Holm-adjusted $p=0.021$) and on an independent external cohort ($+0.0129$, Holm-adjusted $p=3.8\times10^{-16}$), with lesion sensitivity improving from 0.849 to 0.861 internally and Dice improving over baseline in every stratified anatomic region, including the rare, high-risk sites motivating this work, while naive Bayesian pooling degrades performance sharply and power-scaled pooling underperforms it throughout. Our findings suggest that population-level spatial priors can meaningfully augment deep learning predictions in whole-body oncologic segmentation, provided the fusion rule is calibrated to where the prior actually carries signal.

AO_SCPLOWBSTRACTC_SCPLOWElectroencephalography (EEG) is a promising tool for automated detection of mild cognitive impairment (MCI) and dementia, but comparisons across studies are limited by inconsistent datasets and evaluation protocols. This study benchmarks ten deep learning models across four resting-state EEG datasets and eight binary classification tasks using a unified preprocessing pipeline and five-fold subject-wise cross-validation. Each experiment was repeated ten times. SCCNet obtained the highest mean subject-level accuracy, sensitivity, and F1 score, while ShallowConvNet achieved the highest mean segment-level accuracy, specificity, and precision. Subject-level aggregation improved mean accuracy for all evaluated models, and performance varied substantially across datasets and diagnostic tasks. Higher computational cost did not consistently correspond to better classification performance, with several compact architectures remaining competitive with substantially larger models. The results provide a reproducible reference for comparing EEG-based dementia classification models under consistent subject-independent evaluation conditions.

Drosophila connectomic datasets provide increasingly comprehensive maps of neuronal morphology and synaptic connectivity, offering an unprecedented opportunity to explore the structural organization of its neural circuits. This calls for designing automated tools to interact with connectomic datasets at scale for efficiently exploring structural features embedded in the vast amount of data. Yet the central challenge remains the understanding of the functional logic of neural circuits. In order to understand how elements of the functional logic may emerge from this structural organization, it is critical to (i) characterize the objects in the natural environment in which brain circuits operate, and (ii) formulate how brain circuits represent and process the defined objects in the natural environment. To develop and demonstrate a methodology for these requirements, we focus on the Drosophila looming-evoked escape pathway. We modeled the trajectory of looming objects that are on a collision course (direct-hits) or pass-by the fly (near-misses): their projected images on the retina can be characterized by the solid angle (angular size) and elevation. We then analyzed the pathway's morphology across the OpticLobe, Hemibrain, and FlyWire connectome datasets. By abstracting their sub-neuronal structure and retinotopic organization, we constructed an executable circuit model that maps each structural element to a processing block. We demonstrate that this model separates direct hits from near misses well before the angular size could tell them apart. To accelerate the connectomic analysis step, we developed a Python toolset with an agentic, code-free workspace interface called NeuroGraphBench (NGB). NGB provides four composable morphology-analysis primitives and an AI agent that composes them to interactively respond to natural-language queries aided by visualization on an interactive 3D canvas. Thus, NGB automates tedious and repetitive tasks to enable faster and scalable connectomic exploration, keeping human reasoning, instead of writing code, at the center of an open-ended research inquiry.

Introduction: Anesthesia is known to modulate glymphatic clearance and cerebrospinal fluid (CSF) transport in rodents, but how these effects translate to a larger, gyrencephalic brain is unknown. With its anatomical similarity to the human brain, the pig offers a valuable translational model for examining anesthesia-dependent CSF-to-brain transport. Methods: We used dynamic in vivo SPECT/CT imaging for six hours following cisterna magna injection of [99mTc]-DTPA to quantify CSF-to-brain tracer transport in pigs under two anesthesia regimens: ketamine/dexmedetomidine (K/D, n=5) which previously has been shown in rodents to enhance glymphatic influx relative to GABAergic anesthesia, and propofol (PRO, n=5). Brain and CSF spaces were delineated using a data-driven non-negative matrix factorization approach, and tracer kinetics were quantified using a one-tissue compartment model. Results: Brain influx could be stably estimated from 2 hours post-injection. Hierarchical sub-division of the brain parenchyma identified two kinetically distinct components with different anatomical distributions: a surface component, located ventrally and within the interhemispheric fissure, showed faster kinetics than the anatomically deeper and lateral-dorsal component. Consistent with rodent findings, K/D-anesthetized pigs showed 62% (p=0.002) greater brain tracer accumulation than PRO-anesthetized pigs. However, while the brain influx rates did not differ substantially (p=0.047), a 52% higher cumulative CSF tracer concentration (p=0.047) could account for most of the difference by providing greater tracer availability for brain entry. Conclusions: In the larger gyrencephalic pig brain, we found higher brain tracer accumulation under K/D anesthesia compared to PRO anesthesia. A significant portion of this difference is readily explained by higher CSF retention, likely driven by a slower CSF turnover. This underscores the necessity of dynamic CSF tracer concentration measurements when assessing CSF-brain influx, a factor we suggest that future glymphatic studies should take into account.

Conventional sleep electroencephalography (EEG) measures often rely on predefined bands, thresholds, and averages that incompletely capture transient oscillatory dynamics across an entire night. Here, we introduce the Dynamic Oscillation (DYNAM-O) Toolbox, an open-source, cross-platform (MATLAB, Python, and Rust) software package for data-driven characterization of individualized neural dynamics in sleep EEG. DYNAM-O identifies transient oscillations as time-frequency peaks on multitaper spectrograms using a novel multi-resolution procedure, computes intrinsic and sleep-state-dependent extrinsic features for each event, and represents the overnight distributions of tens of thousands of TF-peaks as feature histograms spanning oscillation frequency, slow oscillation power, and slow oscillation phase. This distributional representation preserves continuous brain-state variation that could be obscured by averaging within conventional sleep stages. The toolbox further provides Gaussian and spline basis-based dimensionality reduction, visualization, and whole-histogram statistical testing tools to support both exploratory and hypothesis-driven analyses. To demonstrate its use for group-level inference, we analyzed overnight C3-channel EEG from 133 adults (71 females, 72 males; ages 20-35 years) in the Cleveland Family Study. Whole-histogram and parameterized-mode analyses reproduced the established higher center frequency of fast-spindle activity in females and additionally revealed greater low-alpha transient oscillatory activity in females, a pattern outside the conventional sleep spindle range. By completing the analysis cycle from TF-peak extraction to statistical inference, DYNAM-O provides an accessible and interpretable framework for studying individualized sleep physiology and identifying subtle, reproducible electrophysiological patterns.

Automated detection of brain tumors from Magnetic Resonance Imaging (MRI) can accelerate diagnosis and reduce inter-reader variability, yet many existing studies report only top-line accuracy on small datasets, omit efficiency analysis, and provide no interpretability, limiting their clinical credibility. We present a reproducible, comparative, and explainable transfer- learning framework for binary brain-tumor classification. Our framework (i) standardizes a configurable preprocessing pipeline combining CLAHE contrast enhancement and unsharp-mask sharpening, (ii) evaluates a custom CNN baseline and pretrained backbones under an identical training budget, (iii) reports a full metric suite (accuracy, precision, recall, F1, ROC-AUC, PR-AUC, parameter count, and inference latency), and (iv) applies Grad- CAM for spatial interpretability. On a public 253-image MRI dataset (38-image held-out test set), MobileNetV2 achieves the best overall performance (94.74% accuracy, 0.994 ROC-AUC, 0.996 PR-AUC) with only 2.59M parameters and 5.9 ms per- image inference, making it the most deployment-friendly model. Larger backbones (Xception, EfficientNetB0) and the custom CNN converge to degenerate all-positive predictions under the same limited budget, illustrating the small-data overfitting risk that accuracy-only reporting conceals. Grad-CAM confirms that the best model attends to the tumor region. All source code, con- figuration files, and trained evaluation scripts are publicly avail- able at https://github.com/blck-iris/explainable-brain-tumor-mr

Over the last five years, over 13,000 tissue datasets with 200+ million cells from 20 consortia have been spatially registered into the Human Reference Atlas (HRA) common coordinate framework (CCF). The shared 3D spatial and semantic reference system enables exploration of datasets in the context of all other data across organs, assay types, and spatial scales. However, manual registration of individual samples remains resource intensive, posing feasibility challenges exacerbated by the proliferation of samples, assays, and atlasing efforts. This paper presents two approaches to scale up HRA construction: (1) projecting data across biomedical reference atlas systems and (2) using millitomes to bulk register tissue blocks into a reference organ. Both methods use the AutoMated Alignment and Projection (AMAP) pipeline to align 3D mesh models using point cloud registration. We demonstrate the evolving HRA-aligned atlas ecosystem for 6 models from the SPARC Program (heart), Gut Cell Atlas (large intestine), 500-subject consensus kidneys, and the Julich Brain Atlas. Additionally, we used AMAP to project 7 millitome models across 5 organs onto the HRA ecosystem, integrating 300+ tissue extraction sites. AMAP enables scalable tissue registration of data across atlas ecosystems enabling the construction of detailed reference maps of the human body.

Background and Objective Epileptic seizure prediction is a critical challenge requiring the discrimination of subtle preictal physiological changes from interictal brain activity. While deep learning has shown promise in this domain, existing models often face limitations due to small EEG datasets, high computational costs for training from scratch, and a lack of patient-independent generalizability. In this paper, we present a novel framework for EEG-based seizure prediction that leverages pre-trained Vision Transformers (ViTs) through custom architectural modifications and optimized re-training strategies. Methods Our primary contributions include: [bullet]CVIT-ESP: A family of vision transformer architectures that replaces standard patch embedding layers with custom N-dimensional CNN stages to refine EEG representations. [bullet] ESPFormer: A lightweight, custom-designed transformer specifically engineered to mitigate overfitting on limited-scale EEG datasets. We identified optimal fine-tuning combinations for transformer blocks by devising a heuristic search-space reduction strategy, significantly reducing the training complexity. We validated our methods using the patient-independent MLSPred-Bench, involving 12 diverse benchmarks with varying seizure prediction horizons. Results Results demonstrate a clear progression in performance: while prior ResNet and vanilla Transformer models achieved an AUC-ROC of 69.0%, our CVIT-ESP architectures achieved the highest performance with a maximum average AUC of 76.4%. Conclusions These findings suggest that adapting pre-trained ViTs with domain-specific CNN front-ends and strategic fine-tuning offers a robust, generalizable, and resource-efficient path forward for clinical seizure prediction systems. Our code is available at: https://github.com/pcdslab/CVitEsp and https://github.com/pcdslab/ESPFormer

Missing or degraded sequences can limit prostate multiparametric MRI. We developed MSCNet, a sequence-conditioned cross-modal generative framework for reconstructing unavailable contrasts and restoring degraded acquisitions. Across ten completion tasks, task-specific MSCNet achieved mean structural similarity of 0.818 versus 0.798 for the strongest task-matched comparators; matched-capacity analyses showed larger differences in lesion fidelity and boundary preservation. In a blinded 1,000-case reader study, overall image quality met the prespecified non-inferiority criterion for DWI, ADC and T2W completion, but not T1W. In a separate 200-case diagnostic assessment, AUCs for clinically significant cancer were 0.860 with acquired images, 0.841 with MSCNet and 0.797 with baseline-generated images. A locked 186-case three-hospital cohort supported multicentre transportability. These retrospective results support quality-controlled cross-modal reconstruction as an adjunct to acquired prostate MRI.