Progressive transmission of neurological images: iconic modelling

We present a transmission scheme that reduces the latency in transmitting high resolution medical image sets. Traditional compression methods use pixel statistics to reduce transmission bit-rate and ignore visual cues that are important in identifying visually informative regions. Context-based image transmission (CBIT) utilises gross structure, represented by an approximate iconic image, to provide initial information about image content. This iconic image is represented using a novel superelliptic shape-tree. Each part of the iconic image is progressively updated, using rules, to provide an informative image build-up, in a timely manner. CBIT is realised using knowledge about image composition to segment, label and prioritise regions of an image for transmission. Experiments using CT X-ray and MR images show that an informative structure delineation is obtained for the iconic image and that the progressive image build-up is beneficial