Evaluation of Image Compression for Computer-Aided Diagnosis of Breast Tumors in Sonography

摘要

Medical imaging examinations form the basis for physicians diagnosing diseases, as evidenced by the increasing use of digital medical images for picture archiving and communications systems (PACS). However, with enlarged medical image databases and rapid advances in storage and transmission technologies, PACS requires image compression to accelerate the image transmission rate and conserve disk space in order to reduce implementation costs. Therefore, this study evaluates the compression ratios of JPEG and JPEG2000 standards for computer-aided diagnosis (CAD) of breast tumors in medical ultrasound images. The ultrasound images with various compression ratios are compressed by using the two efficacious image compression standards. The reconstructed images are then diagnosed by CAD. Next, diagnostic accuracy is measured based on receiver operating characteristic (ROC) analysis. Analysis results ensure a comparison of the compression ratios with JPEG and JPEG2000 for ultrasound images. Results of this study provide the possible bit rates using JPEG and JPEG2000 for breast ultrasound images.