A new quantization algorithm for a wavelet compression scheme of coronary angiograms

New methods for image compression which are able to store and retrieve efficiently the images produced by a digital angiographic unit are needed for clinical use. Currently discrete cosine transform (DCT) based compression schemes (i.e. JPEG, MPEG) generate unacceptable block artefacts in coronary angiograms at high compression ratios, because of the block coefficients quantization. We propose a wavelet compression algorithm that overcomes fundamentally this drawback. Using a set of 40 coronary angiograms, we evaluated the method by analyzing several image quality factors: peak signal to noise ratio, compression ratio, maximum absolute difference and mean difference of gray levels and subjective evaluation. At the same compression ratio, the proposed technique outperforms the JPEG method in image quality presentation.