Abstract 13

REDUCING THE FRACTURE MISS RATE: AIDING DIAGNOSIS THROUGH IMAGE PROCESSING

Martin Donnelley Greg Knowles

Embedded Systems Laboratory, School of Informatics and Engineering, Flinders University and the Department of Orthopaedics, Division of Surgery, Repatriation General Hospital.

Accurate detection and evaluation of musculoskeletal trauma is often problematic, regardless of the treating physician's level of experience. While bone fractures are a relatively common occurence, their presence can often be missed during x-ray diagnosis. For a trauma patient, a delay in diagnosis can lead to ineffective patient management, increased dissatisfaction, and expensive litigation. Fast and accurate detection of long bone fractures is therefore an important orthopaedic and radiologic problem. We have developed a computer aided fracture detection system - initially for the detection of long bone fractures - that in the future will help lower the miss rate.

We have created a number of image processing software algorithms useful for automating the fracture detection process. The first of these is a tool for fullyautomatic long bone segmentation that is capable of identifying boundaries - as defined by the AO - between the metaphyseal and diaphyseal segments in an x-ray image. Anatomic structure varies markedly between the segmented regions so different fracture detection criteria are required. Our tools are then able to accurately detect abnormalities, including fractures, in the diaphyseal segment of the longbone. Our experiments on a set of sample images show that these tools can consistently identify the boundaries between bone segments, and then accurately highlight midshaft long bone fractures within the marked diaphysis. We believe that in the future this type of scheme will help reduce the number of fractures that are missed during the reading process.

Return to Listing of 2005 Abstracts
View Listing of 2004 Abstracts