Harnessing spatial MRI normalization: patch individual filter layers for CNNs

Neuroimaging studies based on magnetic resonance imaging (MRI) typically employ rigorous forms of preprocessing. Images are spatially normalized to a standard template using linear and non-linear transformations. Thus, one can assume that a patch at location (x, y, height, width) contains the same brain region across the entire data set. Most analyses applied on brain MRI using convolutional neural networks (CNNs) ignore this distinction from natural images. Here, we suggest a new layer type called patch individual filter (PIF) layer, which trains higher-level filters locally as we assume that more abstract features are locally specific after spatial normalization. We evaluate PIF layers on three different tasks, namely sex classification as well as either Alzheimer's disease (AD) or multiple sclerosis (MS) detection. We demonstrate that CNNs using PIF layers outperform their counterparts in several, especially low sample size settings.

[1]  Chuan Li,et al.  Precomputed Real-Time Texture Synthesis with Markovian Generative Adversarial Networks , 2016, ECCV.

[2]  David H. Miller,et al.  Diagnosis of multiple sclerosis: 2017 revisions of the McDonald criteria , 2017, The Lancet Neurology.

[3]  Alexei A. Efros,et al.  Image-to-Image Translation with Conditional Adversarial Networks , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[4]  Nico Karssemeijer,et al.  Deep multi-scale location-aware 3D convolutional neural networks for automated detection of lacunes of presumed vascular origin , 2016, NeuroImage: Clinical.

[5]  Andrew J. Saykin,et al.  Deep Learning in Alzheimer's Disease: Diagnostic Classification and Prognostic Prediction Using Neuroimaging Data , 2019, Front. Aging Neurosci..

[6]  Brian B. Avants,et al.  Symmetric diffeomorphic image registration with cross-correlation: Evaluating automated labeling of elderly and neurodegenerative brain , 2008, Medical Image Anal..

[7]  G. Ascoli,et al.  Computational Neuroanatomy , 2002, Humana Press.

[8]  A. Mechelli,et al.  Using deep learning to investigate the neuroimaging correlates of psychiatric and neurological disorders: Methods and applications , 2017, Neuroscience & Biobehavioral Reviews.

[9]  Konstantinos Kamnitsas,et al.  DeepMedic for Brain Tumor Segmentation , 2016, BrainLes@MICCAI.

[10]  Ritesh Kumar,et al.  A Comprehensive Study of Alzheimer's Disease Classification Using Convolutional Neural Networks , 2019, ArXiv.

[11]  Roger C. Tam,et al.  Deep learning of joint myelin and T1w MRI features in normal-appearing brain tissue to distinguish between multiple sclerosis patients and healthy controls , 2017, NeuroImage: Clinical.

[12]  Jeffrey A. Cohen,et al.  Diagnostic criteria for multiple sclerosis: 2010 Revisions to the McDonald criteria , 2011, Annals of neurology.

[13]  D. Louis Collins,et al.  Unbiased average age-appropriate atlases for pediatric studies , 2011, NeuroImage.