This paper presents a comparison study between 10 automatic and six interactive methods for liver segmentation from contrast-enhanced CT images. It is based on results from the "MICCAI 2007 Grand Challenge" workshop, where 16 teams evaluated their algorithms on a common database. A collection of 20 clinical images with reference segmentations was provided to train and tune algorithms in advance. Participants were also allowed to use additional proprietary training data for that purpose. All teams then had to apply their methods to 10 test datasets and submit the obtained results. Employed algorithms include statistical shape models, atlas registration, level-sets, graph-cuts and rule-based systems. All results were compared to reference segmentations five error measures that highlight different aspects of segmentation accuracy. All measures were combined according to a specific scoring system relating the obtained values to human expert variability. In general, interactive methods reached higher average scores than automatic approaches and featured a better consistency of segmentation quality. However, the best automatic methods (mainly based on statistical shape models with some additional free deformation) could compete well on the majority of test images. The study provides an insight in performance of different segmentation approaches under real-world conditions and highlights achievements and limitations of current image analysis techniques.

Abstract-Defining the support (or frequency) of a subgraph is trivial when a database of graphs is given: it is simply the number of graphs in the database that contain the subgraph. However, if the input is one large graph, an appropriate support definition is much more difficult to find. In this paper we study the core problem, namely overlapping embeddings of the subgraph, in detail and suggest a definition that relies on the non-existence of equivalent ancestor embeddings in order to guarantee that the resulting support is anti-monotone. We prove this property and describe a method to compute the support defined in this way.

This paper uses an ensemble of classifiers and active learning strategies to predict radiologists’ assessment of the nodules of the Lung Image Database Consortium (LIDC). In particular, the paper presents machine learning classifiers that model agreement among ratings in seven semantic characteristics: spiculation, lobulation, texture, sphericity, margin, subtlety, and malignancy. The ensemble of classifiers (which can be considered as a computer panel of experts) uses 64 image features of the nodules across four categories (shape, intensity, texture, and size) to predict semantic characteristics. The active learning begins the training phase with nodules on which radiologists’ semantic ratings agree, and incrementally learns how to classify nodules on which the radiologists do not agree. Using our proposed approach, the classification accuracy of the ensemble of classifiers is higher than the accuracy of a single classifier. In the long run, our proposed approach can be used to increase consistency among radiological interpretations by providing physicians a “second read”

No abstract

Research studies have shown that advances in computed tomography (CT) technology allow better detection of pulmonary nodules by generating higher-resolution images. However, the new technology also generates many more individual transversal reconstructions, which as a result may affect the efficiency and accuracy of the radiologists interpreting these images.The goal of our research study is to build a content-based image retrieval (CBIR) system for pulmonary CT nodules. Currently, texture is used to quantify the image content, but any other image feature could be incorporated into the proposed system. Unfortunately, there is no texture model or similarity measure known to work best for encoding nodule texture properties or retrieving most similar nodules. Therefore, we investigated and evaluated several texture models and similarity measures with respect to nodule size, number of retrieved nodules, and radiologist agreement on the nodules' texture characteristic.The results were generated on 90 thoracic CT scans collected by the Lung Image Database Consortium (LIDC). Every case was annotated by up to four radiologists marking the contour of nodules and assigning nine characteristics (including texture) to each identified nodule. We found that Gabor texture descriptors produce the best retrieval results regardless of the nodule size, number of retrieved items or similarity metric. Furthermore, when analyzing the radiologists' agreement on the texture characteristic, we found that when just two radiologists agreed, the average precision increased from 88% to 96% for both Gabor and Markov texture features. Moreover, once three or four radiologists agreed the precision increased to nearly 100%.