Adaptive entropy coded DCT coding of magnetic resonance image sequences

Compression of magnetic resonance (MR) image sequences have proved to be difficult as compared to other medical imaging modalities. Some of the factors partially responsible for this are significant amount of noise and image artifacts which are present in a 3-D MR dataset. Another factor is use of unsuitable intensity matching techniques in removal of inter-slice dependencies. In this paper we introduce an adaptive DCT coding scheme that employs a triangle-based matching (TBM) algorithm to explore the inter-slice dependencies. The proposed scheme is used to encode a heart sequence. An improvement of 0.3 dB is achieved over adaptive intra-slice schemes at low bit-rates