Mosaic ultrasound medical image compression using TTA10 algorithm

Abstract

This paper describes an ultrasound image compression algorithm, related to mosaic image compression, where each compressed object is represented as a set of indexes to the database of mosaic elements. The proposed approach is considered as an alternative method for ultrasound biomedical image compression. A memory efficient implementation based upon the tree tessellation algorithm v.10 (TTA10) indexing/retrieval solution for managing mosaic elements. The principal advantage of the current approach is that for the very specific kind of ultrasound images (cardiology), it classifies a particular type of image (compression stage) first, and secondly uses a specially constructed schema to decompose an object into parts. Each part of the image is identified as the most similar mosaic image in the database using TTA10, and finally only the index is stored in compressed structure of an image. Since the size of ultrasound image storage is similar to the video one, while the specific biomedical content is highly correlated, the usage of mosaic image compression is an effective solution for storage.