Swin fMRI Transformer Predicts Early Neurodevelopmental Outcomes from Neonatal fMRI

2024-11-25

AI Summary

This study addresses the challenge of early neurodevelopmental outcome prediction in newborns. We propose Swin 4D fMRI Transformer (SwiFT), the first adaptation of the Swin Transformer to spatiotemporal modeling of 4D resting-state fMRI for predicting Bayley-III cognitive, motor, and language scores. To mitigate severe data scarcity in neonates, we innovatively integrate group-level independent component analysis (ICA) for dimensionality reduction with transfer pretraining initialized from adult fMRI models. Additionally, we introduce IG-SQ—a gradient-based interpretability method—to precisely localize developmentally relevant functional brain representations. Evaluated on the dHCP dataset, SwiFT significantly outperforms all baseline models (p < 0.001), achieving high-accuracy, multi-domain developmental score prediction. Interpretability analysis reveals robust associations between early cognitive development and canonical functional systems—particularly the default mode network—thereby offering both a novel predictive tool and mechanistic insights into neonatal brain development.

Technology Category

• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Machine Learning: Neuro-Symbolic Learning
• Natural Language Processing: Interpretability, Analysis, and Evaluation of NLP Models

Application Category

• Web Mining and Content Analysis: Large pretrained models with web data
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Semantics and Knowledge: Methods, algorithms and applications for the development of semantic models, knowledge graphs and other forms of structured data models with machine-interpretable semantics

Abstract

Brain development in the first few months of human life is a critical phase characterized by rapid structural growth and functional organization. Accurately predicting developmental outcomes during this time is crucial for identifying delays and enabling timely interventions. This study introduces the SwiFT (Swin 4D fMRI Transformer) model, designed to predict Bayley-III composite scores using neonatal fMRI data from the Developing Human Connectome Project (dHCP). To enhance predictive accuracy, we apply dimensionality reduction via group independent component analysis (ICA) and pretrain SwiFT on large adult fMRI datasets to address the challenges of limited neonatal data. Our analysis shows that SwiFT significantly outperforms baseline models in predicting cognitive, motor, and language outcomes, leveraging both single-label and multi-label prediction strategies. The model's attention-based architecture processes spatiotemporal data end-to-end, delivering superior predictive performance. Additionally, we use Integrated Gradients with Smoothgrad sQuare (IG-SQ) to interpret predictions, identifying neural spatial representations linked to early cognitive and behavioral development. These findings underscore the potential of Transformer models to advance neurodevelopmental research and clinical practice.

Problem

• Neonatal Brain Development
• Early Prediction
• Developmental Delay

Innovation

• Swin fMRI Transformer
• Early Brain Development Prediction
• Interpretable Machine Learning in Pediatrics