Image-matching as a medical diagnostic support tool (DST) for brain diseases in children

Abstract

Imaging-matching is an important research area in imaging informatics. We have developed and evaluated a novel diagnostic support tool (DST) based on medical image matching using MR brain images. The approach consists of two steps, database generation and image matching. The database contains pre-diagnosed MR brain images. As the images are added to the database, they are registered to the 3D Talairach coordinate system. In addition, regions of interests (ROI) are generated, and image-processing techniques are used to extract relevant image parameters related to the brain and diseases from the ROIs and from the entire MR image. The second step is to retrieve relevant information from the database by performing image matching. In this step, the physician first submits a query image. The DST computes the similarity between the query image and each of the images in the database, and then presents the most similar images to the user. Since the database contains pre-diagnosed images, the retrieved cases tend to contain relevant diagnostic information. To evaluate the usefulness of the DST in a clinical setting, pediatric brain diseases were used. The database contains 2500 pediatric patients between ages 0 and 18 with brain Magnetic Resonance (MR) images of known brain lesions. A testbed was established at the Childrens Hospital Los Angeles (CHLA) for acquiring MR images from the PACS server of patients with known lesions. These images were matched against those in the DST pediatric brain MR database. An expert pediatric neuroradiologist evaluated the matched results. We found that in most cases, the image-matching method was able to quickly retrieve images with relevant diagnostic content. The evaluation method and results are given.