Evaluating Partial Shape Queries for Pathology-based Retrieval of Vertebra

There is a growing research interest in reliable contentbased image retrieval (CBIR) methods for biomedical images. Feature representation algorithms used in indexing medical images have to address conflicting goals of reducing feature dimensionality while retaining important and often subtle biomedical features. In a collection of digitized X-rays of the spine, for example, a typical user is interested in only a small region of the vertebral body pertinent to the pathology: for this experiment, the Anterior Osteophyte (AO). To address this, we have proposed the use of partial shape matching (PSM) methods for spinal X-ray image retrieval. This paper describes an evaluation of a Procrustes distancebased PSM method for pathology-based retrieval. It was conducted on a subset of data selected from a collection of 17,000 digitized X-rays of the spine from the Second National Health and Nutritional Examination Survey (NHANES II) maintained at the U.S. National Library of Medicine. Two classifications for AO were used to evaluate the PSM algorithm, viz., Macnab classification, which is a published method, and a grading system developed by us. The evaluation based on ground truth established by the classifications shows that the PSM algorithm is a promising approach for content-based image retrieval of biomedical images. *

[1]  Euripides G. M. Petrakis,et al.  Matching and Retrieval of Distorted and Occluded Shapes Using Dynamic Programming , 2002, IEEE Trans. Pattern Anal. Mach. Intell..

[2]  D Resnick,et al.  Traction osteophytes of the lumbar spine: radiographic-pathologic correlation. , 1988, Radiology.

[3]  Hemant D. Tagare.  Deformable 2-D template matching using orthogonal curves , 1997, IEEE Transactions on Medical Imaging.

[4]  Kazuo Toraichi,et al.  A partial shape matching using wedge wave feature extraction , 1997, 1997 IEEE Pacific Rim Conference on Communications, Computers and Signal Processing, PACRIM. 10 Years Networking the Pacific Rim, 1987-1997.

[5]  L. Rodney Long,et al.  Partial shape matching for CBIR of spine x-ray images , 2003, IS&T/SPIE Electronic Imaging.

[6]  Sameer Antani,et al.  Anatomical Shape Representation in Spine X-ray Images , 2003 .

[7]  B. J. Doherty,et al.  Morphologic Study of Lumbar Vertebral Osteophytes , 1998, Southern medical journal.

[8]  Sameer Antani,et al.  Biomedical information from a national collection of spine x-rays: film to content-based retrieval , 2003, SPIE Medical Imaging.