Wavelet compression algorithm applied to abdominal ultrasound images

摘要: We sought to investigate acceptable compression ratios of lossy wavelet on 640 x 480 8 abdominal ultrasound (US) images. acquired 100 US images with normal and abnormal findings from the view station a 932 bed teaching hospital. The were then compressed at quality factors (QFs) 3, 10, 30, 50 followed outcomes pilot study. This was equal average 4.3:1, 8.5:1, 20:1 36.6:1, respectively. Four objective measurements carried out examine compare image degradation between original Receiver operating characteristic (ROC) analysis also introduced for subjective assessment. Five experienced qualified radiologists as reviewers blinded corresponding pathological findings, analysed paired 400 randomly ordered two 17 inch thin film transistor/ liquid crystal display (TFT/LCD) monitors. At ROC analysis, area under curve (Az) 0.874 ratio 36.6:1. size only 2.7% this ratio. parameters showed higher mean squared error (MSE) or root (RMSE) values, poorer quality. signal-to-noise (SNR) peak (PSNR) values indicated better RMSE, PSNR 36.6:1 4.84 +/- 0.14, 35.45 dB, finding suggests that, basis patient sample, did not adversely affect diagnostic performance evaluation radiologists' interpretation so risk affecting diagnosis. (author abstract)