Benchmarking CNN on 3D Anatomical Brain MRI: Architectures, Data Augmentation and Deep Ensemble Learning
Authors:
Preprint
Abstract
Deep Learning (DL) and specifically CNN models have become a de facto method for a wide range of vision tasks, outperforming traditional machine learning (ML) methods. Consequently, they drew a lot of attention in the neuroimaging field in particular for phenotype prediction or computer-aided diagnosis. However, most of the current studies often deal with small single-site cohorts, along with a specific pre-processing pipeline and custom CNN architectures, which make them difficult to compare to. We propose an extensive benchmark of recent state-of-the-art (SOTA) 3D CNN, evaluating also the benefits of data augmentation and deep ensemble learning, on both Voxel-Based Morphometry (VBM) pre-processing and quasi-raw images. Experiments were conducted on a large multi-site 3D brain anatomical MRI data-set comprising N=10k scans on 3 challenging tasks: age prediction, sex classification, and schizophrenia diagnosis. We found that all models provide significantly better predictions with VBM images than quasi-raw data. This finding evolved as the training set approaches 10k samples where quasi-raw data almost reach the performance of VBM. Moreover, we showed that linear models perform comparably with SOTA CNN on VBM data. We also demonstrated that DenseNet and tiny-DenseNet, a lighter version that we proposed, provide a good compromise in terms of performance in all data regime. Therefore, we suggest to employ them as the architectures by default. Critically, we also showed that current CNN are still very biased towards the acquisition site, even when trained with N=10k multi-site images. In this context, VBM pre-processing provides an efficient way to limit this site effect. Surprisingly, we did not find any clear benefit from data augmentation techniques. Finally, we proved that deep ensemble learning is well suited to re-calibrate big CNN models without sacrificing performance.Keywords
All Related Versions
This publication has 44 references indexed in Scilit:
- TorchIO: A Python library for efficient loading, preprocessing, augmentation and patch-based sampling of medical images in deep learningComputer Methods and Programs in Biomedicine, 2021
- Accurate brain age prediction with lightweight deep neural networksMedical Image Analysis, 2021
- Detect and correct bias in multi-site neuroimaging datasetsMedical Image Analysis, 2021
- Convolutional neural networks for classification of Alzheimer's disease: Overview and reproducible evaluationMedical Image Analysis, 2020
- Deep residual learning for neuroimaging: An application to predict progression to Alzheimer’s diseaseJournal of Neuroscience Methods, 2020
- Self-supervised learning for medical image analysis using image context restorationMedical Image Analysis, 2019
- Automated image quality evaluation of structural brain MRI using an ensemble of deep learning networksJournal of Magnetic Resonance Imaging, 2019
- Using deep autoencoders to identify abnormal brain structural patterns in neuropsychiatric disorders: A large‐scale multi‐sample studyHuman Brain Mapping, 2018
- Detection of Schizophrenia in brain MR images based on segmented ventricle region and deep belief networksNeural Computing and Applications, 2018
- Brain tumor segmentation with Deep Neural NetworksMedical Image Analysis, 2017