Ahmed Al-Taie


PhD student

Office: 119, Research I
Address: Campus Ring 1, 28759 Bremen, Germany
Telephone: (+49-421) 200-3182
Email: a.altaie [@jacobs-university.de]

Uncertainty Estimation in Multi-modal Medical Images. Uncertainty is widespread when interpreting medical imaging data sets due to different sources of artifacts such as signal measurement errors or noise and partial volume effects in the acquisition process. The effectiveness of visualization methods for supporting decision making and diagnostic exploration is limited by the lack of suitable uncertainty estimation and visualization tools. This project is concerned with deriving estimates for uncertainty in multimodal medical imaging data and using these estimations in both improving image segmentation results and the analysis of these results. The project aims to develop uncertainty-aware medical image segmentation system that incorporate methods to estimates and visualize the uncertainties associated with segmenting multi-modal medical images. The study identifies the uncertainty sources related to medical image visualization pipeline and propose visualization methods to convey the different classes of uncertainty sources to the user in an intuitive way.We want to develop interactive tools that can help the user in exploring the spatial distribution and uncertainty level distribution of the different uncertainty sources in the segmented image and visually analyze their effect in certain medical applications such as treatment planning. In a first step, we want to analyze the available data and the respective manual segmentations to draw statistics of the characteristics of the individual segments (training data). As tumor can be quite diverse, we mainly want to use healthy patient data for this step. Then, we can use those statistics to compute probabilities for segmentations of other data sets using probabilistic segmentation algorithm with prior knowledge. We also want to apply these to data with tumor and investigate how the tumor will be segmented in this regard. Based on the probabilistic segmentation, we will compute uncertainties in the segmentation results. Further, we will develop visualization methods that show uncertainty-enhanced segmentation results. Uncertainty estimates are of particular interest for the tumor segmentation. One focus of our approach is to combine multi-modal data. The uncertainty estimations will be performed per modality and then the resulting uncertainty-aware segmentations shall be combined. The combination shall lead to a reduction of the uncertainty, as more information shall allow for a better segmentation. We will investigate methods to best combine the uncertainty-aware estimations. Finally, the uncertainty-aware visualizations shall be evaluated, where they shall be compared to visualizations without uncertainty display.

Uncertainty Estimation and Visualization for Multi-modal Image Segmentation.
Multi-modal imaging allows for the integration of complementary information from multiple medical imaging modalities for an improved analysis. The multiple information channels may lead to a reduction of the uncertainty in the analysis and decision-making process. Recently, efforts have been made to estimate the uncertainty in unimodal image segmentation decisions and visually convey this information to the medical experts that examine the image segmentation results. We propose an approach to extend uncertainty estimation and visualization methods to multi-modal image segmentations. We combine probabilistic uni-modal image segmentation results using the concept of ensemble of classifiers. The uncertainty is computed using a measure that is based on the Kullback- Leibler divergence. We apply our approach for an improved segmentation of Multiple Sclerosis (MS) lesions from multiple MR brain imaging modalities. Moreover, we demonstrate how our approach can be used to estimate and visualize the growth of a brain tumor area for imaging data taken at multiple points in time. Both the MS lesion and the area of tumor growth are detected as areas of high uncertainty due to different characteristics in different imaging modalities and changes over time, respectively.

Point-wise Diversity Measure and Visualization for Ensemble of with Application to Image Segmentation

The idea of using ensembles of classifiers is to increase the performance when compared to applying a single classifier. Crucial to the performance improvement is the diversity of the ensemble. A classifier ensemble is considered to be diverse, if the classifiers make no coinciding errors. Several studies discuss the diversity issue and its relation to the ensemble accuracy. Most of them proposed measures that are based on an ”Oracle” classification. In this paper, we propose a new probability-based diversity measure for ensembles of unsupervised classifiers, i.e., when no Oracle machine exists. Our measure uses a point-wise definition of diversity, which allows for a distinction of diverse and non-diverse areas. Moreover, we introduce the concept of further categorizing the diverse areas into healthy and unhealthy diversity areas. A diversity area is healthy for the ensemble performance, if there is enough redundancy to compensate for the errors. Then, the performance of the ensemble can be based on two parameters, the non-diversity area, i.e., the size of all regions where the classifiers of the ensemble agree, and the healthy diversity area, i.e., the size of the regions where the diversity is healthy. Furthermore, our point-wise diversity measure allows for an intuitive visualization of the ensemble diversity for visual ensemble performance comparison in the context of image segmentation.

Uncertainty-aware Ensemble of Classifiers for Segmenting Brain MRI Data

Estimating and visualizing uncertainty in medical image segmentation has become an active research area due to the necessity of making medical experts aware of possibly wrong segmentation decisions. Still, to our knowledge all these methods are based on a single choice of the underlying segmentation approach. Segmentation using an ensemble of classifiers (or committee machine) use multiple classifiers to increase the performance when compared to applying a single classifier. In this paper, we propose methods to estimate uncertainties in segmentations produced by ensembles of classifiers. We investigate and compare the different combining strategies of the segmentation results of the ensemble members from an uncertainty point of view. We discuss why some combining  strategies tend to perform better than others. Also, we visualize the estimated uncertainties using a color mapping in image space and propose a post-segmentations correction step to reclassify the noisy pixels in the final result based on the statistical uncertainty.

Majority Rule Uncertainty

Improved Bias-corrected Fuzzy C-means Segmentation of Brain MRI Data

Errors in the scanning procedures lead to uncertainties when trying to segment the scanned images. Fuzzy c-means is a clustering method that can be applied to segment images with uncertainty estimates. Bias-corrected fuzzy c-means (BCFCM) clustering compensates for two sources of uncertainty by modeling noise and bias fields during the segmentation process. In this paper, we present an approach to improve BCFCM clustering and apply it to magnetic resonance imaging (MRI) data of the human brain. Our approach is based on two variants of BCFCM clustering, the classical one and the one with distance-based weights. We improve both variants by slightly modifying their main algorithms for better bias field estimation. To evaluate the improved algorithms, we apply the algorithms to synthetic data, simulated MRI brain data, and real MRI brain data with ground truth in form of manual segmentation. All experiment results show that our improved methods outperform the original methods in both the segmentation accuracy and efficiency (the number of iterations).



  • Ahmed Al-Taie, Horst K. Hahn, Lars Linsen:
    Computers & Graphics, vol. 39, iss. 0, pp. 48 — 59
    @article{AlTaie2014, title = "Uncertainty estimation and visualization in probabilistic segmentation ", journal = "Computers & Graphics ", volume = "39", number = "0", pages = "48 -- 59", year = "2014", note = "", issn = "0097-8493", doi = "http://dx.doi.org/10.1016/j.cag.2013.10.012",
      author = "Ahmed Al-Taie and Horst K. Hahn and Lars Linsen", note = {linsenvcglaltaie},
      url = {http://www.sciencedirect.com/science/article/pii/S0097849313001532},
      annote = {http://vcgl.jacobs-university.de/wp-content/uploads/2013/12/AlTaie2013.png},