Biomedical subjects
Publications and source records attributed to Daniel Keysers.
We integrate the tangent method into a statistical framework for classification analytically and practically. The resulting consistent framework for adaptation allows us to efficiently estimate the tangent vectors representing the variability. The framework improves classification results on two real-world pattern recognition tasks from the domains handwritten character recognition and automatic speech recognition.
The impact of content-based access to medical images is frequently reported but existing systems are designed for only a particular modality or context of diagnosis. Contrarily, our concept of image retrieval in medical applications (IRMA) aims at a general structure for semantic content analysis that is suitable for numerous applications in case-based reasoning or evidence-based medicine. Within IRMA, stepwise processing results in six layers of information modeling (raw data layer, registered data layer, feature layer, scheme layer, object layer, knowledge layer) incorporating medical expert knowledge. At the scheme layer, medical images are represented by a hierarchical structure of ellipses (blobs) describing image regions. Hence, image retrieval transforms to graph matching. The multilayer processing is implemented using a distributed system designed with only three core elements. The central database holds program sources, process-ing schemes, images, features, and blob trees; the scheduler balances distributed computing by addressing daemons running on all connected workstations; and the web server provides graphical user interfaces for data entry and retrieval..
Automatic identification of frontal (posteroanterior/anteroposterior) vs. lateral chest radiographs is an important preprocessing step in computer-assisted diagnosis, content-based image retrieval, as well as picture archiving and communication systems. Here, a new approach is presented. After the radiographs are reduced substantially in size, several distance measures are applied for nearest-neighbor classification. Leaving-one-out experiments were performed based on 1,867 radiographs from clinical routine. For comparison to existing approaches, subsets of 430 and 5 training images are also considered. The overall best correctness of 99.7% is obtained for feature images of 32 x 32 pixels, the tangent distance, and a 5-nearest-neighbor classification scheme. Applying the normalized cross correlation function, correctness yields still 99.6% and 99.3% for feature images of 32 x 32 and 8 x 8 pixel, respectively. Remaining errors are caused by image altering pathologies, metal artifacts, or other interferences with routine conditions. The proposed algorithm outperforms existing but sophisticated approaches and is easily implemented at the same time.
Categorization of medical images means selecting the appropriate class for a given image out of a set of pre-defined categories. This is an important step for data mining and content-based image retrieval (CBIR). So far, published approaches are capable to distinguish up to 10 categories. In this paper, we evaluate automatic categorization into more than 80 categories describing the imaging modality and direction as well as the body part and biological system examined. Based on 6231 reference images from hospital routine, 85.5% correctness is obtained combining global texture features with scaled images. With a frequency of 97.7%, the correct class is within the best ten matches, which is sufficient for medical CBIR applications.