Brain age gap as a potential biomarker for schizophrenia: A multi-site structural MRI study

Schizophrenia Detection and Classification: A Systematic Review of the Last Decade

BACKGROUND/OBJECTIVES

Artificial Intelligence (AI) in healthcare employs advanced algorithms to analyze complex and large-scale datasets, mimicking aspects of human cognition. By automating decision-making processes based on predefined thresholds, AI enhances the accuracy and reliability of healthcare data analysis, reducing the need for human intervention. Schizophrenia (SZ), a chronic mental health disorder affecting millions globally, is characterized by symptoms such as auditory hallucinations, paranoia, and disruptions in thought, behavior, and perception. The SZ symptoms can significantly impair daily functioning, underscoring the need for advanced diagnostic tools.

METHODS

This systematic review has been conducted following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 guidelines and examines peer-reviewed studies from the last decade (2015-2024) on AI applications in SZ detection as well as classification. The review protocol has been registered in the International Prospective Register of Systematic Reviews (PROSPERO) under registration number: CRD42024612364. Research has been sourced from multiple databases and screened using predefined inclusion criteria. The review evaluates the use of both Machine Learning (ML) and Deep Learning (DL) methods across multiple modalities, including Electroencephalography (EEG), Structural Magnetic Resonance Imaging (sMRI), and Functional Magnetic Resonance Imaging (fMRI). The key aspects reviewed include datasets, preprocessing techniques, and AI models.

RESULTS

The review identifies significant advancements in AI methods for SZ diagnosis, particularly in the efficacy of ML and DL models for feature extraction, classification, and multi-modal data integration. It highlights state-of-the-art AI techniques and synthesizes insights into their potential to improve diagnostic outcomes. Additionally, the analysis underscores common challenges, including dataset limitations, variability in preprocessing approaches, and the need for more interpretable models.

CONCLUSIONS

This study provides a comprehensive evaluation of AI-based methods in SZ prognosis, emphasizing the strengths and limitations of current approaches. By identifying unresolved gaps, it offers valuable directions for future research in the application of AI for SZ detection and diagnosis.

New clues for the role of cerebellum in schizophrenia and the associated cognitive impairment

Schizophrenia (SZ) is a complex neuropsychiatric disorder associated with severe cognitive dysfunction. Although research has mainly focused on forebrain abnormalities, emerging results support the involvement of the cerebellum in SZ physiopathology, particularly in Cognitive Impairment Associated with SZ (CIAS). Besides its role in motor learning and control, the cerebellum is implicated in cognition and emotion. Recent research suggests that structural and functional changes in the cerebellum are linked to deficits in various cognitive domains including attention, working memory, and decision-making. Moreover, cerebellar dysfunction is related to altered cerebellar circuit activities and connectivity with brain regions associated with cognitive processing. This review delves into the role of the cerebellum in CIAS. We initially consider the major forebrain alterations in CIAS, addressing impairments in neurotransmitter systems, synaptic plasticity, and connectivity. We then focus on recent findings showing that several mechanisms are also altered in the cerebellum and that cerebellar communication with the forebrain is impaired. This evidence implicates the cerebellum as a key component of circuits underpinning CIAS physiopathology. Further studies addressing cerebellar involvement in SZ and CIAS are warranted and might open new perspectives toward understanding the physiopathology and effective treatment of these disorders.

Brain Age Prediction Using 2D Projections Based on Higher-Order Statistical Moments and Eigenslices from 3D Magnetic Resonance Imaging Volumes

Brain age prediction from 3D MRI volumes using deep learning has recently become a popular research topic, as brain age has been shown to be an important biomarker. Training deep networks can be very computationally demanding for large datasets like the U.K. Biobank (currently 29,035 subjects). In our previous work, it was demonstrated that using a few 2D projections (mean and standard deviation along three axes) instead of each full 3D volume leads to much faster training at the cost of a reduction in prediction accuracy. Here, we investigated if another set of 2D projections, based on higher-order statistical central moments and eigenslices, leads to a higher accuracy. Our results show that higher-order moments do not lead to a higher accuracy, but that eigenslices provide a small improvement. We also show that an ensemble of such models provides further improvement.