Applied Sciences

Research

Jump to: Review, Other

36 pages, 1305 KB

Open AccessArticle

Multi-ROI Multimodal 3D Vision Transformer for Alzheimer’s Disease Classification with Attention-Based Interpretability

by Juan A. Castro-Silva, María N. Moreno-García and Diego H. Peluffo-Ordóñez

Appl. Sci. 2026, 16(11), 5705; https://doi.org/10.3390/app16115705 - 5 Jun 2026

Viewed by 398

Abstract

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder for which early and accurate diagnosis remains a critical challenge. In this work, we propose a Multi-ROI Multimodal 3D Vision Transformer for AD classification that integrates structural MRI data with clinical and volumetric biomarkers within [...] Read more.

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder for which early and accurate diagnosis remains a critical challenge. In this work, we propose a Multi-ROI Multimodal 3D Vision Transformer for AD classification that integrates structural MRI data with clinical and volumetric biomarkers within a unified attention-based framework. The proposed approach leverages anatomically guided multi-region-of-interest (ROI) decomposition to focus on disease-relevant brain structures, including the hippocampus, entorhinal cortex, fornix, and major cortical lobes. Each ROI is encoded using 3D tubelet embeddings, while clinical and volumetric features are transformed into feature-wise tokens, enabling seamless multimodal fusion through self-attention mechanisms. A hemisphere-aware selection strategy is introduced to identify the most discriminative ROI representations, enhancing both performance and interpretability. The model is evaluated on a merged multi-cohort dataset combining ADNI, AIBL, and OASIS using a 7-fold cross-validation protocol. Experimental results demonstrate that the proposed method achieves high classification performance, reaching an accuracy of 97.62% and an AUC of 0.9940, outperforming single-modality and whole-brain baselines. Furthermore, attention-based analysis provides interpretable insights into the relative importance of clinical and neuroanatomical features, revealing consistency with established AD biomarkers. These findings highlight the effectiveness of multimodal integration and ROI-based representation for robust and explainable AD classification. Full article

(This article belongs to the Special Issue MR-Based Neuroimaging, 2nd Edition)

► Show Figures

Figure 1

15 pages, 3122 KB

Open AccessArticle

AttentionMS-Net: An Attention-Enhanced Multi-Scale Framework for Alzheimer’s Disease Classification with Subject-Level Validation

by Osman Yildiz and Abdulhamit Subasi

Appl. Sci. 2026, 16(9), 4338; https://doi.org/10.3390/app16094338 - 29 Apr 2026

Cited by 1 | Viewed by 610

Abstract

Many MRI-based Alzheimer’s disease (AD) classification studies report near-perfect accuracy; however, these results are often inflated by data leakage caused by slice-level splitting, where correlated slices of the same subject appear in both the training and test sets. In this study, we introduce [...] Read more.

Many MRI-based Alzheimer’s disease (AD) classification studies report near-perfect accuracy; however, these results are often inflated by data leakage caused by slice-level splitting, where correlated slices of the same subject appear in both the training and test sets. In this study, we introduce AttentionMS-Net, an attention-enhanced multi-scale deep learning architecture that combines channel-spatial attention using the Convolutional Block Attention Module (CBAM) with multi-scale feature aggregation from intermediate EfficientNet-B3 layers for binary AD classification (Non-Demented vs. Demented). Using strict subject-level 10-fold cross-validation on the OASIS dataset (347 subjects, 86,437 slices), our experiments clearly show the impact of data leakage: image-level 10-fold CV achieves about 99.9% accuracy, whereas subject-level 10-fold CV with the same model results in 80.8% accuracy—a reduction of roughly 19 percentage points. Aggregating predictions at the subject level further improves accuracy to 82.4% (AUC: 0.889), suggesting that prediction errors are mainly slice-specific rather than subject-specific. Systematic ablation reveals a complementary interaction between attention and multi-scale components, neither of which performs as well alone. Post-hoc Grad-CAM++ visualization and SHAP analysis suggest that AttentionMS-Net’s attention patterns are focused on ventricular regions—visual biomarkers indicative of overall brain atrophy rather than early hippocampal degeneration. These findings highlight the unreliability of current benchmarks and establish methodologically rigorous baselines for future AD classification research. Full article

(This article belongs to the Special Issue MR-Based Neuroimaging, 2nd Edition)

► Show Figures

Figure 1

20 pages, 13821 KB

Open AccessArticle

Improved MRSI in a Murine Glioma Model Using semiLASER: Refining the Metabolomics Data Obtained from Murine Models

by Zoona Javed, Gary V. Martinez, Marta Mulero-Acevedo, Ana Paula Candiota, Carles Arus, Miquel E. Cabañas and Silvia Lope-Piedrafita

Appl. Sci. 2026, 16(8), 3788; https://doi.org/10.3390/app16083788 - 13 Apr 2026

Viewed by 530

Abstract

Background: Magnetic resonance spectroscopic imaging (MRSI) offers valuable metabolic information for assessing brain tumor progression and therapeutic response, but its performance in rodent models is often hindered by the low signal-to-noise ratio (SNR) and spatially heterogeneous spectral quality, particularly in peripheral voxels. These [...] Read more.

Background: Magnetic resonance spectroscopic imaging (MRSI) offers valuable metabolic information for assessing brain tumor progression and therapeutic response, but its performance in rodent models is often hindered by the low signal-to-noise ratio (SNR) and spatially heterogeneous spectral quality, particularly in peripheral voxels. These issues reduce the number of usable spectra available for quantitative and classifier-based analyses. To address this, we implemented a multi-voxel MRSI-semiLASER sequence—widely recommended in clinical practice—on a 7T Bruker Biospec system running ParaVision 5.1 to improve spectral homogeneity in mouse brain tumor studies. Results: Compared with the vendor CSI-PRESS sequence, MRSI-semiLASER produced more uniform spectra across the grid and achieved up to a 1.2-fold SNR increase in murine glioma, enabling a 20% reduction in slice thickness without compromising spectral quality. Importantly, the sequence produced a substantial gain in the proportion of spectra amenable to analysis, particularly in outer grid voxels that frequently fail with CSI-PRESS. The implemented MRSI-semiLASER sequence and instructions are openly available to the community. Conclusions: MRSI-semiLASER improves spectral homogeneity, increases the proportion of usable spectra, and supports higher spatial detail. These technical improvements may enhance data yield per subject and may facilitate future applications such as more robust pattern recognition workflows and greater data efficiency in longitudinal studies, although such aspects were not evaluated here. Full article

(This article belongs to the Special Issue MR-Based Neuroimaging, 2nd Edition)

► Show Figures

Figure 1

17 pages, 3640 KB

Open AccessArticle

A 3D Global-Patch Transformer for Brain Age Prediction Using T1-Weighted MRI with Gray and White Matter Maps

by Seung-Jun Lee, Myungeun Lee, Yoo Ri Kim and Hyung-Jeong Yang

Appl. Sci. 2026, 16(6), 3004; https://doi.org/10.3390/app16063004 - 20 Mar 2026

Cited by 1 | Viewed by 592

Abstract

With the increasing prevalence of neurodegenerative diseases driven by population aging, imaging-based biomarkers are needed to quantify brain aging at an early stage. Brain age, which estimates structural brain aging relative to chronological age, has emerged as a useful indicator. Prior work has [...] Read more.

With the increasing prevalence of neurodegenerative diseases driven by population aging, imaging-based biomarkers are needed to quantify brain aging at an early stage. Brain age, which estimates structural brain aging relative to chronological age, has emerged as a useful indicator. Prior work has mainly used T1-weighted MRI with deep learning models such as convolutional neural networks (CNNs) or transformers; however, many approaches insufficiently capture three-dimensional structural continuity and localized anatomical patterns, and tissue-specific aging in gray matter (GM) and white matter (WM) is often treated as auxiliary. To address these limitations, we propose a 3D Global–Patch Transformer framework for brain age prediction that directly processes volumetric data while jointly learning global brain structure and local anatomical features. Our model runs global and patch pathways in parallel and explicitly incorporates GM and WM structural maps alongside T1-weighted MRI to encode tissue-specific aging signals. Experiments on multiple public datasets, including IXI and OASIS, show that the proposed method reduces mean absolute error (MAE) by approximately 10–15% compared with CNN-based and single-input transformer baselines, with notably improved performance in older populations, highlighting the value of tissue-level structural information for brain age estimation. Full article

(This article belongs to the Special Issue MR-Based Neuroimaging, 2nd Edition)

► Show Figures

Figure 1

Review

Jump to: Research, Other

21 pages, 1747 KB

Open AccessReview

The Role of Advanced MR Imaging in Gliomas

by Anastasia K. Zikou, Eleni Romeo, George A. Alexiou, Marios Lampros, Spyridon Voulgaris, Loukas Astrakas and Maria I. Argyropoulou

Appl. Sci. 2026, 16(2), 1027; https://doi.org/10.3390/app16021027 - 20 Jan 2026

Viewed by 1684

Abstract

Gliomas are a significant health problem with a lot of imaging challenges. The role of imaging is no longer limited to only providing anatomic details, but with the advancement of Magnetic Resonance Imaging (MRI) techniques, it now permits the assessment of the freedom [...] Read more.

Gliomas are a significant health problem with a lot of imaging challenges. The role of imaging is no longer limited to only providing anatomic details, but with the advancement of Magnetic Resonance Imaging (MRI) techniques, it now permits the assessment of the freedom of water molecule movement, the microvascular structure, the hemodynamic characteristics, and the chemical makeup of certain metabolites of lesions. These advanced imaging techniques include diffusion-weighted imaging, diffusion tensor imaging, dynamic contrast-enhanced MRI, Magnetic Resonance (MR) perfusion, MR angiography, and magnetic resonance spectroscopy. their role in the diagnosis, classification, and post-treatment follow-up of gliomas, as well as their application in radiogenomics and glioma analysis with the aid of artificial intelligence, is presented and discussed. Full article

(This article belongs to the Special Issue MR-Based Neuroimaging, 2nd Edition)

► Show Figures

Figure 1

Other

Jump to: Research, Review

14 pages, 2117 KB

Open AccessPerspective

Noninvasive Neuromodulation and Neuroimaging to Enhance Glymphatic Function for Neurodegenerative and Autoimmune Disorders in Next-Generation Personalized Treatments and Precision Neuropsychiatry: A Perspective Proposal

by Castañeyra-Perdomo Agustín, José L. González-Mora, Sophocles Goulis, Risto J. Ilmoniemi, Pantelis Lioumis, Nikos Makris and Stefano Pallanti

Appl. Sci. 2026, 16(13), 6593; https://doi.org/10.3390/app16136593 - 2 Jul 2026

Viewed by 342

Abstract

Transcranial magnetic stimulation-electroencephalography (TMS–EEG) biomarkers have recently become available as a means to obtain new understanding of the causal chains of neuronal signaling in the brain. This is a key piece in the puzzle of how the brain is organized and how it [...] Read more.

Transcranial magnetic stimulation-electroencephalography (TMS–EEG) biomarkers have recently become available as a means to obtain new understanding of the causal chains of neuronal signaling in the brain. This is a key piece in the puzzle of how the brain is organized and how it works. Using dMRI tractography, we can map the circuit beneath a chosen cortical target; TMS can then stimulate it, and EEG records responses that reflect—and may even be caused by—activity in that structural circuit. The chain of events after stimulus delivery can be observed and quantified using current neuroimaging and TMS–EEG technology, a matter of tremendous relevance on how to approach novel therapeutic approaches in clinical conditions. Herein, we elaborate upon a perspective of how groundbreaking multi-locus TMS (mTMS) technology associated with EEG and multimodal neuroimaging can be applied to modulate the flow dynamics of the glymphatic system (GS). The enhancement of the GS waste clearance functionality has been shown to improve significantly symptom severity in neurodegenerative disorders such as Alzheimer’s (AD) and Parkinson’s disease (PD) or long COVID. In this perspective paper, we consider that next-generation therapeutics using versatile technologies such as noninvasive neuromodulation and neuroimaging will provide important benefits in public health and in how society can address the management of these difficult-to-deal-with ailments more effectively. Full article

(This article belongs to the Special Issue MR-Based Neuroimaging, 2nd Edition)

► Show Figures

Figure 1

23 pages, 2783 KB

Open AccessSystematic Review

Radiological Findings in Gliosarcoma: A Diagnostic Challenge

by Domenico La Torre, Attilio Della Torre, Prospero Longo, Ilaria Rania, Emilio De Bartolo, Giada Garufi, Valeria Garufi and Salvatore Massimiliano Cardali

Appl. Sci. 2026, 16(13), 6374; https://doi.org/10.3390/app16136374 - 25 Jun 2026

Viewed by 892

Abstract

Aim: This study aims to identify the primary differential MRI features between gliosarcoma (GS) and glioblastoma (GBM) based on a comprehensive review of existing literature. Given their close relationship, identifying preoperative neuroradiological markers is essential for distinguishing these two entities. Methods and [...] Read more.

Aim: This study aims to identify the primary differential MRI features between gliosarcoma (GS) and glioblastoma (GBM) based on a comprehensive review of existing literature. Given their close relationship, identifying preoperative neuroradiological markers is essential for distinguishing these two entities. Methods and Analysis: We reviewed studies involving patients with histologically confirmed GS and primary GBM. Our analysis, aligned with the 2021 WHO Classification of Tumors of the Central Nervous System, evaluated specific radiological characteristics in order to determine their diagnostic value in differential diagnosis. Results: While GS is categorized as a variant of GBM, distinct radiological patterns emerged from our literature review. Gliosarcomas are typically larger due to their sarcomatous component and exhibit more intense contrast enhancement and frequent cortical involvement. Notably, hemorrhagic patterns are more prevalent in GS, whereas GS displays less necrosis and a lower incidence of midline-crossing edema compared to GBM. Conversely, glioblastomas more frequently present with cystic components and pial involvements as well as a higher risk of ependymal invasion. Conclusions: Although specific radiological features—such as hemorrhage frequency and tumor size—can strongly suggest GS over GBM diagnosis, no single feature was considered pathognomonic. Among radiological tools, MRI remains the most important for guiding a diagnostic suspicion, but histopathological confirmation remains the gold standard for definitive diagnosis. Full article

(This article belongs to the Special Issue MR-Based Neuroimaging, 2nd Edition)

► Show Figures

Figure 1

15 pages, 769 KB

Open AccessPerspective

Concurrent/Interleaved TMS–fMRI as an MR-Guided Framework for Target Engagement

by Chiara Di Fazio and Sara Palermo

Appl. Sci. 2026, 16(9), 4135; https://doi.org/10.3390/app16094135 - 23 Apr 2026

Viewed by 548

Abstract

Concurrent/interleaved transcranial magnetic stimulation combined with functional MRI (TMS–fMRI) enables causal perturbation of targeted cortical regions while measuring whole-brain MR-based responses during stimulation. This perspective argues that the main translational value of concurrent/interleaved TMS–fMRI lies in operationalizing target engagement and network-level propagation as [...] Read more.

Concurrent/interleaved transcranial magnetic stimulation combined with functional MRI (TMS–fMRI) enables causal perturbation of targeted cortical regions while measuring whole-brain MR-based responses during stimulation. This perspective argues that the main translational value of concurrent/interleaved TMS–fMRI lies in operationalizing target engagement and network-level propagation as measurable endpoints, bridging stimulation “dose” to clinically meaningful effects. Rather than proposing a validated gold-standard protocol, we frame concurrent/interleaved TMS–fMRI as a measurement-driven translational approach in which MRI-informed targeting and MR-based readouts can be integrated to quantify target engagement under clearly specified methodological and quality-control conditions. This perspective specifically aims to make explicit an intermediate verification step that remains only partially formalized in current clinical neuromodulation workflows. We propose that MRI-based neuronavigation should move beyond template coordinates toward individualized anatomical and network-informed targeting, with the aim of improving precision, reproducibility, and safety. Building on the field’s evolution from technical feasibility to emerging clinical applications, we outline a staged framework from feasibility to biomarker potential, summarize representative protocol archetypes, and provide pragmatic recommendations for reporting and study design to improve comparability. This framework is intended to guide future concurrent/interleaved TMS–fMRI studies toward biomarker-ready designs and more clinically informative network neuromodulation. We further distinguish offline MRI-informed targeting from potential future real-time or closed-loop implementations, and we emphasize that current biomarker claims should remain proportional to the still heterogeneous evidence base. Full article

(This article belongs to the Special Issue MR-Based Neuroimaging, 2nd Edition)

► Show Figures

Figure 1

Journal Menu

Journal Browser

MR-Based Neuroimaging, 2nd Edition

Share This Special Issue

Editors

Special Issue Information

Keywords

Benefits of Publishing in a Special Issue

Related Special Issue

Published Papers (8 papers)

Research

Review

Other

Further Information

Guidelines

MDPI Initiatives

Follow MDPI