Progressive transmission of images over fading channels using rate-compatible LDPC codes.

In this paper, we propose a combined source/channel coding scheme for transmission of images over fading channels. The proposed scheme employs rate-compatible low-density parity-check codes along with embedded image coders such as JPEG2000 and set partitioning in hierarchical trees (SPIHT). The assignment of channel coding rates to source packets is performed by a fast trellis-based algorithm. We examine the performance of the proposed scheme over correlated and uncorrelated Rayleigh flat-fading channels with and without side information. Simulation results for the expected peak signal-to-noise ratio of reconstructed images, which are within 1 dB of the capacity upper bound over a wide range of channel signal-to-noise ratios, show considerable improvement compared to existing results under similar conditions. We also study the sensitivity of the proposed scheme in the presence of channel estimation error at the transmitter and demonstrate that under most conditions our scheme is more robust compared to existing schemes.

Unbalanced 3-D tree structure for region-based coding of volumetric medical images.

In this paper, we propose an unbalanced three-dimensional (3-D) coefficient tree structure for 3-D region-of-interest (ROI) coding of medical volumetric data. We compare the proposed coding scheme against 3-D SPIHT and a recent technique that employs optimal tree construction shown to improve the performance of the 3-D SPIHT algorithm. For a MR volumetric dataset, at 0.1 bpp, the proposed region-based coding scheme outperforms by about 0.5 dB. The improvement is more significant at higher bitrates, up to 1 dB compared to SPIHT algorithm and optimal tree construction. In addition, the proposed tree structure provides a more general multiregion multiquality coding framework rather than ROI/non-ROI coding.

Region of interest image coding for digital mammography.

In this paper, we investigate region-based wavelet compression methods and describe a region-based coder based on the Set Partitioning in Hierarchical Trees (SPIHT) algorithm, called unbalanced spatial orientation trees (UBT), applied to digital mammograms. We compare this method against the region-based extension of SPIHT (ROI-SPIHT), and the ROI coding unit of JPEG2000 (JP2K) algorithm on five digital mammograms compressed at rates ranging from 0.1 to 1.0 bits per pixel (bpp). We show that UBT is competitive in PSNR with the other two region-based coding methods, also providing a more general multiregion multiquality coding framework rather than ROI/non-ROI coding. Unlike ROI-SPIHT, UBT allows encoding of diagnostically significant regions with best possible fidelity while allocating less number of bits to remaining regions.

Spatially adaptive wavelet thresholding of rubidium-82 cardiac PET images.

82Rb positron emission tomography (PET) images can be used to diagnose coronary artery disease (CAD), but deterioration of the images due to noise can compromise their quality. This work presents our work on a wavelet-based thresholding method for denoising cardiac /sup 82/Rb PET images. Our approach is based on a three-dimensional (3D) discrete dyadic wavelet transform, with iterative thresholding of coefficients being performed in a spatially adaptive context. A subjective analysis indicates that this technique leads to better discrimination between diagnostically significant features and noise when observing the denoised images and the corresponding cardiac polar maps. Results are presented for cardiac PET images of a normal subject (low risk of CAD) using a figure of merit based on expected image properties for such a patient. The increase in image quality is substantial when the proposed method is compared to conventional denoising protocols.