Limited One-time Sampling Irregularity Map (LOTS-IM): Automatic Unsupervised Quantitative Assessment of White Matter Hyperintensities in Structural Brain Magnetic Resonance Images

Abstract

We present the application of limited one-time sampling irregularity map (LOTS-IM): a fully automatic unsupervised approach to extract brain tissue irregularities in magnetic resonance images (MRI), for quantitatively assessing white matter hyperintensities (WMH) of presumed vascular origin, and multiple sclerosis (MS) lesions and their progression. LOTS-IM generates an irregularity map (IM) that represents all voxels as irregularity values with respect to the ones considered "normal". Unlike probability values, IM represents both regular and irregular regions in the brain based on the original MRI's texture information. We evaluated and compared the use of IM for WMH and MS lesions segmentation on T2-FLAIR MRI with the stateof-the-art unsupervised lesions' segmentation method, Lesion Growth Algorithm from the public toolbox Lesion Segmentation Toolbox (LST-LGA), with several well established conventional supervised machine learning schemes and with state-of-the-art supervised deep learning methods for WMH segmentation. In our experiments, LOTS-IM outperformed unsupervised method LST-LGA on WMH segmentation, 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 learning 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). LOTS-IM also performed well on MS lesions segmentation, performing similar to LST-LGA. On the other hand, 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.Keywords: white matter hyperintensities (WMH), multiple sclerosis (MS) lesion, unsupervised lesion segmentation, irregularity map, penumbra of brain's lesion, characterisation of WMH and MS lesions.

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

“…2B shows, peak mean performances are 0.4704 (0.1587) for IAM, 0.2888 (0.0990) for IAM-UResNet and 0.4409 (0.1410) for IAM-UNet. The latter performs 15.21% better than the former, which is quite opposite to when TL is not used (see [5]). Our guess is that residual blocks in UResNet perform poorly if it has to learn from real values of IAM.…”

“…Whereas, task adaptation TL, where tasks in training and testing processes are different, has not widely explored in medical image analysis. However, the newly proposed unsupervised method of Limited One Time Sampling IAM (LOTS-IAM) [5] has been reported to serve the purpose of white matter hyperintensities (WMH) segmentation performing at the level of DNN architectures trained for this specific purpose while executing a different task i.e., extracting irregular brain tissue texture in the form of irregularity age map (IAM).…”

“…For our experiments we selected UNet [7] and UResNet [2] architectures used in various natural/medical image segmentation studies. They were also evaluated in the original study of LOTS-IAM [5]. In this study we made two modifications to allow UNet and UResNet to learn IAM: 1) no non-linear activation function (e.g., sigmoid, softmax or ReLU) is used in the last layer of both architectures and 2) mean squared error loss function is used instead of Dice similarity coefficient or binary cross entropy in both architectures.…”

“…Fig. 2A shows the DSC of the WMH masks obtained for the 60 MRI data by both algorithms without TL(Aa) and with TL(Ab) using the results of LOTS-IAM thresholded at 0.18 (see [5]). Both DNN schemes could yield better results if task-adaptation TL using IAM is used.…”

“…Our guess is that residual blocks in UResNet perform poorly if it has to learn from real values of IAM. On the other hand, UNet learned from IAM with minimal performance drop (i.e., 2.95% drops from the LOTS-IAM) compared with when learned from manual WMH labels (i.e., 6.21% differences as per [5]). Although the performance of IAM-UResNet and IAM-UNet apparently follow the LOTS-IAM's performance at different thresholds in terms of DSC, a closer look at the learning process shows these relationships are not linear.…”