Gerardo Hermosillo Valadez

Supervised learning from multiple experts: whom to trust when everyone lies a bit

We describe a probabilistic approach for supervised learning when we have multiple experts/annotators providing (possibly noisy) labels but no absolute gold standard. The proposed algorithm evaluates the different experts and also gives an estimate of the actual hidden labels. Experimental results indicate that the proposed method is superior to the commonly used majority voting baseline.

A comparison study of atlas-based image segmentation: the advantage of multi-atlas based on shape clustering

Purpose: By incorporating high-level shape priors, atlas-based segmentation has achieved tremendous success in the area of medical image analysis. However, the effect of various kinds of atlases, e.g., average shape model, example-based multi-atlas, has not been fully explored. In this study, we aim to generate different atlases and compare their performance in segmentation. Methods: We compare segmentation performance using parametric deformable model with four different atlases, including 1) a single atlas, i.e., average shape model (SAS); 2) example-based multi-atlas (EMA); 3) cluster-based average shape models (CAS); 4) cluster-based statistical shape models (average shape + principal shape variation modes)(CSS). CAS and CSS are novel atlases constructed by shape clustering. For comparison purpose, we also use PDM without atlas (NOA) as a benchmark method. Experiments: The experiment is carried on liver segmentation from whole-body CT images. Atlases are constructed by 39 manually delineated liver surfaces. 11 CT scans with ground truth are used as testing data set. Segmentation accuracy using different atlases are compared. Conclusion: Compared with segmentation without atlas, all of the four atlas-based image segmentation methods achieve better results. Multi-atlas based segmentation behaves better than single-atlas based segmentation. CAS exhibit superior performance to all other methods.

A method for automatic matching of multi-timepoint findings for enhanced clinical workflow

Non-interventional diagnostics (CT or MR) enables early identification of diseases like cancer. Often, lesion growth assessment done during follow-up is used to distinguish between benign and malignant ones. Thus correspondences need to be found for lesions localized at each time point. Manually matching the radiological findings can be time consuming as well as tedious due to possible differences in orientation and position between scans. Also, the complicated nature of the disease makes the physicians to rely on multiple modalities (PETCT, PET-MR) where it is even more challenging. Here, we propose an automatic feature-based matching that is robust to change in organ volume, subpar or no registration that can be done with very less computations. Traditional matching methods rely mostly on accurate image registration and applying the resulting deformation map on the findings coordinates. This has disadvantages when accurate registration is time-consuming or may not be possible due to vast organ volume differences between scans. Our novel matching proposes supervised learning by taking advantage of the underlying CAD features that are already present and considering the matching as a classification problem. In addition, the matching can be done extremely fast and at reasonable accuracy even when the image registration fails for some reason. Experimental results∗ on real-world multi-time point thoracic CT data showed an accuracy of above 90% with negligible false positives on a variety of registration scenarios.