A multiprocessor architecture for medical image compression in a PACS environment.

We present a new multiprocessor architecture for medical image compression using digital signal processors. The system has been designed to be used in a NuBus-based workstation and to apply different image compression algorithms to medical images. The system has been tested with a full-frame cosine transform and bit allocation table on several imaging modalities ranging from ultrasound to magnetic resonance imaging. Results show that image quality is acceptable when the allocation tables are calculated for each specific image modality. Implementation of algorithms such as wavelet transforms and vector quantization which respond to the characteristics of each type of image is in progress.