Computational Image Analysis and RadiologyRA quantification
We work on the automatic quantification of erosion development and joint space narrowing during the course of rheumatoid arthritis (RA). The work is aimed at a more precise monitoring of RA during therapy and during multi center clinical trials. This was the main application topic in my PhD work in cooperation with the Medical University of Vienna and Vienna General Hospital and is still ongoing in the AAMIR project. The work resulted in the RAquantify package, that allows for the automatic RA assessment based on radiographs in a clinical context.
Automatic RA assessment
The basic two building blocks of radiography based RA progression monitoring are ''erosion detection'' and ''joint space width measurement'':
1. Joint space width measurement: The first important marker of RA progression is ''joint space narrowing''. Based on models of appearance and shape, we locate the bones in hand radiographs, and subsequently measure joint spaces. A first experience regarding joint space width measurement has been published recently in Radiology [Peloschek et al. 07], and a comperative study was reported in the Journal of Rheumatology [Sharp et al. 07].
2. Erosion detection: We analyze the bone contour texture by classifiers, that have been trained on healthy and diseased cases. This results in an indication of regions on the bone contour, with high probability of erosive destruction. The method for erosion detection and joint space width measurement has been accepcted for publication in IEEE TMI [Langs et al.].
3. Erosion visualization: Besides the automatic detection of erosions, the visualizations of these findings to the radiologist is important. Currently we use the generative appearance model, to visualize the deviation from a helathy plausible bone texture. It provides an intuitive interface, and highlights different degrees of change. In an Academic Radiology paper [Langs et al. 07] we explore the capabilities of generative appearance models to ''detect and visualize erosions''.
Literature
- Automatic Quantification of Joint Space Narrowing and Erosions in Rheumatoid Arthritis G. Langs, P. Peloschek, H. Bischof, F. Kainberger. in IEEE Transactions on Medical Imaging 28(1):151-164, January 2009 [PMID:19116197]
- Model-Based Erosion Spotting and Visualization in Rheumatoid Arthritis G. Langs, P. Peloschek, H. Bischof, F. Kainberger. in Academic Radiology, 2007 Oct;14(10):1179-88, [PMID:17889335]
- An automatic model-based system for joint space measurements on hand radiographs: initial experience [abstract+pdf] Philipp Peloschek, Georg Langs, Michael Weber, Johannes Sailer, Michael Reisegger, Herwig Imhof, Horst Bischof, Franz Kainberger. in Radiology 2007 Dec;245(3):855-62
- Computer based methods for measurement of joint space width: Update of an Ongoing OMERACT Project [pdf], John T. Sharp, Jane Angwin, Maarten Boers, Jeff Duryea, Gabriele von Ingersleben, James R. Hall, Joost A. Kauffman, Robert Landewe, Georg Langs, Cédric Lukas, ~Jean-Francis Maillefert, H. J. Bernelot Moens, Philipp Peloschek, Vibeke Strand and Desiree van der Heijde, in //Journal of Rheumatology//, 34, 2007, pp. 874-883, [PMID:17407243]