Is Texture Predictive for Age and Sex in Brain MRI?

TL;DR

This paper investigates whether smaller receptive fields in neural networks are sufficient for predicting age and sex from brain MRI scans, challenging the assumption that large receptive fields are always necessary in medical image analysis.

Contribution

The study evaluates the effectiveness of neural networks with varying receptive fields for age and sex prediction in brain MRI, providing insights into model design for medical imaging tasks.

Findings

Smaller receptive fields can achieve comparable accuracy to larger ones.

Large receptive fields may not be essential for certain brain MRI prediction tasks.

Efficient models with limited receptive fields can reduce computational complexity.

Abstract

Deep learning builds the foundation for many medical image analysis tasks where neuralnetworks are often designed to have a large receptive field to incorporate long spatialdependencies. Recent work has shown that large receptive fields are not always necessaryfor computer vision tasks on natural images. We explore whether this translates to certainmedical imaging tasks such as age and sex prediction from a T1-weighted brain MRI scans.