Neurofind: using deep learning to make individualised inferences in brain-based disorders.

Area of Science:

• Neuroscience
• Psychiatry
• Medical Imaging

Background:

• Precision psychiatry increasingly uses normative models to understand brain heterogeneity.
• Developing these models typically requires significant technical expertise and resources, limiting accessibility for many researchers.

Purpose of the Study:

• To introduce Neurofind, a freely available web-based tool designed to democratize the use of advanced normative models for brain morphology analysis.
• To provide researchers with an accessible platform for generating Outlier Index Scores and Brain Age from structural MRI data.

Main Methods:

• Neurofind was developed using harmonized structural MRI data from 3362 healthy controls (ages 20-80).
• Brain region volumes were extracted and analyzed using an adversarial autoencoder for the Outlier Index Score and support vector regression for Brain Age.
• The platform requires minimal user input, processing structural MRI images through a four-step web interface (www.neurofind.ai).

Main Results:

• Neurofind outputs Outlier Index Scores (deviation from normative morphology) and Brain Age (predicted age from morphometry).
• Application to Alzheimer's disease (AD) and schizophrenia (SCZ) datasets revealed distinct patterns: AD patients showed extreme outliers (55.2%) with temporal-limbic deviations, while SCZ patients (30.1%) had hippocampal/pallidal deviations and greater heterogeneity.
• Both AD and SCZ patient groups exhibited accelerated brain aging compared to normative data.

Conclusions:

• Neurofind offers an accessible solution for researchers to apply sophisticated normative modeling techniques to structural MRI data.
• The tool facilitates the identification of deviations in brain morphology and estimates brain age, supporting research in psychiatric and neurodegenerative disorders.
• Findings in AD and SCZ highlight the utility of Neurofind in characterizing disease-specific patterns of brain morphology and aging.