Keywords: Multiple Sclerosis, Neuro

Motivation: Deep learning (DL) reconstructions show promise in accelerating MRI yet have not been extensively validated clinically, particularly for 3D sequences.

Goal(s): To evaluate the diagnostic quality of DL-based 3D FLAIR compared to Wave-CAIPI-accelerated FLAIR in a clinical setting.

Approach: This prospective study included 26 patients undergoing evaluation for demyelinating disease with Wave-CAIPI-FLAIR and a resolution-matched 6-fold-under-sampled Cartesian FLAIR acquisition with DL reconstruction.

Results: DL-FLAIR reduced scan time (1:53 vs. 2:50) and showed better image quality with higher SNR/CNR, greater lesion conspicuity, and reduced noise compared to Wave-CAIPI-FLAIR, with high agreement in lesional and regional brain volumes between both methods.

Impact: Deep learning reconstruction of 3D-FLAIR provides 30% less acquisition time and improved subjective image quality compared to a state-of-the-art accelerated technique. The excellent agreement in quantitative lesion and regional brain volumes suggests robustness for use in clinical and research studies.

Keywords: Tumors (Post-Treatment), Non-Proton, Sodium

Motivation: Sodium MRI is a promising technique for understanding the brain tumor microenvironment. However, sodium MRI at 3T suffers from extremely low SNR, resulting in compromised resolution and long acquisition times.

Goal(s): Our goal is to create a high-resolution sodium MRI at 3T using generative AI to improve biological characterization, treatment monitoring, and surgical planning for brain tumor patients.

Approach: We developed a physics-informed synthetic dataset to train an anatomically-constrained GAN for high-resolution neuroimaging of brain tumors.

Results: When applied to brain tumor patients' images, the synthetic-sodium MRI improved resolution, SNR, and correlated with expression of sodium-proton exchanger (NHE1) on image-guided biopsy.

Impact: High-resolution sodium neuroimaging at 3T using physics-informed anatomically-constrained GAN has the potential to make multinuclear MRI feasible in the clinical environment, leading to conceivable improvements in diagnosis, monitoring, treatment, and our understanding of the biology of brain tumors.

Keywords: Tumors (Pre-Treatment), Brain, Glioma, Habitat imaging

Motivation: Accurate preoperative identification of isocitrate dehydrogenase (IDH) mutation is  crucial for improving patients’ management in clinical practice. Intratumor heterogeneity in gliomas limits the accurate determination of IDH mutation to some extent.

Goal(s):  T1-CE-derived BBB permeability and DWI-derived cell density habitat imaging may enable more precise prediction of IDH mutation by parcellating similar voxels using a clustering method.

Approach: We developed and validated imaging habitats based on T1-CE and DWI to predict IDH mutation by localized mapping of tumor heterogeneity.

Results: The damaged vascular and hypocellular imaging habitat performed best and robust to predict the IDH mutation, and was considered as the sensitive habitat. 

Impact: Fully recognizing and exploiting this heterogeneity can contribute to improving the prediction accuracy of IDH mutation status, providing more precise treatment and management strategies, and ultimately improving survival and quality of life. 

Keywords: White Matter, Brain Connectivity, Transcriptome

Motivation: The research aims to uncover intricate gray-white matter structural connectivity (GWSC) patterns and associated gene expression profiles in ADHD.

Goal(s): Develop machine-learning classifiers based on GWSC to distinguish ADHD from controls, bridging its gap with gene expression to unveil neurobiological mechanisms.

Approach: Utilize T1-weighted and diffusion-weighted MRI data to construct GWSC networks. Employed four machine-learning classifiers for classification. Analyzed transcriptomes from the Allen Human Brain Atlas to link with gene expression.

Results: Classifiers achieved over 75% accuracy, with Gaussian-kernel SVM leading at 82.6%. Ventromedial prefrontal cortex emerged as a key contributor. Transcriptome analysis identified enrichment in "neuron projection development."

Impact: These findings empower clinicians with accurate ADHD classifiers and pinpoint the ventromedial prefrontal cortex as a hub. The revelation of gene expression nuances in neuron projection development advances targeted interventions, fostering a shift towards more personalized and effective ADHD treatments.

Keywords: Alzheimer's Disease, Relaxometry, xAI, Explainable, Deep Learning

Motivation: While recent studies show high accuracy in the classification of Alzheimer’s disease using deep neural networks, the underlying learned concepts have not been investigated.

Goal(s): To systematically identify the concepts learned by the deep neural network for model validation.

Approach: Using R2* maps we separated Alzheimer's patients (n=117) from healthy controls (n=219) by using a deep neural network and systematically investigated the learned concepts using Concept Relevance Propagation (CRP).

Results: In line with established histological findings, highly relevant concepts were primarily found in and adjacent to the basal ganglia.

Impact: The identification of concepts learned by deep neural networks for disease classification enables validation of the models and improves reliability.

Keywords: Parkinson's Disease, Machine Learning/Artificial Intelligence, Few-shot learning

Motivation: Recent research indicates that various atypical Parkinsonian syndromes (APSs) exhibit distinct and subtle patterns of iron accumulation in the globus pallidus and putamen, typically detected through susceptibility-weighted imaging (SWI).

Goal(s): We propose a novel automated framework for distinguishing between APSs, specifically MSA-P and PSP, in SWI allowing the model to learn from a small amount of labeled data.

Approach: We combined T1-weighted and SWI to create a Hybrid Contrast Image, facilitating precise registration. Furthermore, we used Hyperbolic Few-shot contrastive learning for similarity-based.

Results: The model achieved a balanced accuracy of approximately 94.29%, demonstrating its superior robustness compared to other models and distance metrics.

Impact: Our proposed approach demonstrated the potential to classify specific APS with high performance using a small amount of labeled data. Furthermore, it can be extended to apply not only to binary-classification of specific APS but also to the entire APS.

Keywords: Neuroinflammation, Brain, Encephalitis · Gliomas · Magnetic resonance imaging

Motivation: Encephalitis and glioma can appear very similar in atypical cases. However, their treatment protocols differ significantly. As such, distinguishing between these two diseases is crucial. 

Goal(s): Our objective is to assess and compare the performance of various machine learning (ML) techniques in discriminating between encephalitis and glioma in atypical cases.

Approach: We compare the performance of the classical machine learning (CML) model and the deep learning (DL) model, and assess the effectiveness of utilizing radiomics features extracted from both CML and DL in distinguishing encephalitis from glioma in atypical cases.

Results: ML models can distinguish between encephalitis and glioma in atypical cases.

Impact: Surgery is commonly considered as the initial treatment for glioma, while non-operative therapy is the primary approach for managing encephalitis. Precise identification of glioma and encephalitis facilitates physicians in avoiding misdiagnosis and delays in treatment. 

Keywords: Tumors (Post-Treatment), Tumor, Radiomics; fractal dimensions; 3D-printing technology

Motivation: This study sought to forecast the prognosis of glioblastoma patients by conducting a retrospective analysis of their fractal dimensions (FD) from postoperative multimodal MRI and radiomics features within surgical regions. Additionally, it aimed to assess the potential for improving clinical therapeutic outcomes using preoperative personalized three dimensional (3D)-printing technology.

Goal(s): Exploring whether personalised 3D-printing technology can improve surgical precision and thus prolong survival in glioblastoma patients

Approach: Using questionnaires, radiomics, and FD to evaluate whether preoperative 3D-printing technology improves postoperative outcomes and survival

Results: The FD of surgical regions was associated with overall survival, and preoperative 3D-printing improves patient prognosis and prolongs survival

Impact: Multimodal magnetic resonance imaging radiomics and fractal dimension can predict patient survival by analyzing postoperative images, while personalized 3D printing technology can improve surgical accuracy, reduce the fractal dimension of the surgical regional, and prolong the overall survival of patients