Komura, T. | Search bioRxiv

bioRxiv2018

Transfer Learning for Task Adaptation of Brain Lesion Assessment and Prediction of Brain Abnormalities Progression/Regression using Irregularity Age Map in Brain MRI

The Irregularity Age Map (IAM) for the unsupervised assessment of brain white matter hyperintensities (WMH) opens several opportunities in machine learning-based brain MRI analysis, including transfer task adaptation learning in the MRI brain lesions segmentation and prediction of lesion progression and regression. The lack of need for manual labels is useful for transfer learning. Whereas, the nature of IAM itself can be exploited for predicting lesion progression/regression. In this study, we propose the use of task adaptation transfer learning for WMH segmentation using CNN through weakly-training UNet and UResNet using the output from IAM and the use of IAM for predicting patterns of WMH progression and regression.

neuroscience↗

bioRxiv2018

Limited One-time Sampling Irregularity Age Map (LOTS-IAM): Automatic Unsupervised Detection of Brain White Matter Abnormalities in Structural Magnetic Resonance Images

We present a complete study of limited one-time sampling irregularity map (LOTS-IM), a fully automatic unsupervised approach to extract brain tissue irregularities in magnetic resonance images (MRI), including its application and evaluation for quantitative assessment of white matter hyperintensities (WMH) of presumed vascular origin and assessing multiple sclerosis (MS) lesion progression. LOTS-IM is unique compared to similar other methods because it yields irregularity map (IM) which represents WMH as irregularity values, not probability values, and retains the original MRIs texture information. We tested and compared the usage of IM for WMH segmentation on T2-FLAIR MRI with various methods, including the well established unsupervised WMH segmentation Lesion Growth Algorithm from the public toolbox Lesion Segmentation Toolbox (LST-LGA), conventional supervised machine learning schemes and state-of-the-art supervised deep neural networks. In our experiments, LOTS-IM outperformed unsupervised method LST-LGA, both in performance and processing speed, thanks to the limited one-time sampling scheme and its implementation on GPU. Our method also outperformed supervised conventional machine learning algorithms (i.e., support vector machine (SVM) and random forest (RF)) and deep neural networks algorithms (i.e., deep Boltzmann machine (DBM) and convolutional encoder network (CEN)), while yielding comparable results to the convolutional neural network schemes that rank top of the algorithms developed up to date for this purpose (i.e., UResNet and UNet). The high sensitivity of IM on depicting signal change deems suitable for assessing MS progression, although care must be taken with signal changes not reflective of a true pathology.

neuroscience↗