Dr D Duda – STSM Report

Dorota Duda, Poland


Report concerning my Short Term Scientific Mission (STSM), undertaken in September, in France, within the framework of the COST Action BM1304

My Short Term Scientific Mission took place from 1st to 21st September 2014, in France. The purpose of this mission was the visit to the Nuclear Magnetic Resonance Laboratory at the Institute of Myology, which is a part of the Pitié-Salpétrière University Hospital in Paris. This laboratory provided us with the data analyzed within the framework of the COST Action BM1304. The visit to the NMR Laboratory, IoM, enabled me to meet the other members of our working group (WG4), who – like me – are working on the muscle imaging texture analysis based on magnetic resonance images (MRI). The objective of such an analysis is to characterize the dystrophic muscle in golden retriever muscular dystrophy dogs (GRMD) at different stages of the development of the disease. The considered texture characterization methods are tested for their usefulness in the classification process (GRMD dogs versus the healthy control dogs) and in the assessment of the disease progression.

Our meetings enabled us to discuss the texture classification experiments already done, using the GRMD MRI. Although the preliminary results are quite satisfactory, many problems related to the choice of the best possible image preprocessing- and analysis- settings still remain unresolved. One of them is to determine the smallest accepted size of analyzed image regions, so called “Regions of Interest” (ROIs). Such a problem becomes important when dealing with considered image data, where many considered ROIs are extremely small. In this case, several broadly used and powerful texture analysis methods cannot be applied.

Another goal of our exchange was to establish the most important directions for further analyses based on the available database. Finding the smallest possible sets of texture characteristics, used when classifying GRMD and healthy muscle tissue at several moments of GRMD development, is obviously important. However, much more essential is to characterize the texture evolution under the disease progression for each tissue type and for each muscle group (TC, EDL, GASLAT, GASMED). It would enable doctors to assess in a less invasive way the response to potential treatment progress.

During my stay in the NMR Laboratory, IoM, we initiated the work aimed at selecting the most robust features for the muscular tissue differentiation at each stage of disease development and for each muscle group. In the near future we will try to find the best textural features for characterization of the MR texture evolution under the disease progression in GRMD dogs. The methods for such a characterization are currently being developed in my laboratory: Faculty of Computer Science, Bialystok University of Technology, Poland. We also plan to jointly publish the results of the first step of or experiments, aiming at differentiation between the GRMD dogs versus the healthy ones, at each stage of muscle development, and for each type of muscle.

My travel to France was also a good occasion to visit another institution: Signal and Image Processing Laboratory (LTSI) at the University of Rennes 1. This laboratory is part of the French National Institute for Health and Medical Research (INSERM) and brings together many specialists on texture analysis and MRI imaging. Several of them are also cooperating with our COST Action members. In Rennes, I could work directly with the coordinator of my workgroup (WG4). We started preparing a joint publication combining the most interesting initial results presented during the first Working Groups and Management Committee Meeting (in Milan, on March 2014). Such a publication will contain a description of applied texture analysis methods, their usability and restrictions.