Reliable medical imaging transmission for PACS over ATM networks

In an ATM (Asynchronous Transfer Mode)-based PAC system, cell losses during the transmission might cause degradation on the quality of medical images. This in turn will affect the accuracy of the diagnosis. A three-step scheme to minimize the effect of cell losses on the quality of medical images is proposed in this paper. The first step is related to the medical imaging coding before it enters the network, in which, ROIs (Regions of Interest) of the medical imaging which are crucial to diagnosis are kept non-compressed and packetized with pixel-level inter-leaving. Non-compression can make the data of ROIs to be more robust to cell losses than any compression algorithms, while pixel-level inter-leaving is strong for bursty cell losses recovery combined with FEC (Forward Error Correction) at the receiver. The background part of the medical imaging will be compressed using fractal algorithm, which can get very high compression ratio to balance the large amount data of the ROIs. In the second step, the ROIs will be allocated the highest priority during transmission. While in the third step, FEC will be used to minimize the existed cell losses at the receiver. The balance and optimal of these three stages are discussed from the system-level point of view.