Perceptual quantitative quality assessment of JPEG2000 compressed ct images with various slice thicknesses

Abstract

Modern medical equipments produce huge amounts of data that need to be archived for long periods and efficiently transferred over networks. Data compression plays an essential role in reducing the amount of medical imaging data. Medical images can usually be compressed by a factor of three before any degradation appears. Higher compression levels are desirable but can only be achieved with lossy compression, thus scarifying image quality. The diagnosis value of compressed medical images has been studied and recommendations about maximum acceptable compression ratios have been provided based on qualitative visual analysis. It has been suggested, without further investigation, that CT images, with thicknesses below five mm, cannot undergo lossy compression if diagnostic value needed to be preserved. In this paper, we present an objective quantitative quality assessment of compressed CT images using Visual Signal to Noise Ratio. Our results show that visual fidelity can be significantly affected by two factors, slice thickness and exposure time, for images compressed using the same compression ratio.

QoE-based optimal resource allocation in wireless healthcare networks: opportunities and challenges

“…Pambrun et al in [47] study the quality assessment of compressed CT images. They present the diagnosis of compressed medical images and recommend the maximum acceptable compression ratios based on qualitative visual analysis.…”

“…Such a need is especially clear for [45] approaches were used to with PSNR correlate with MOS and WSNR A. Przelaskowski Used MSE, PSNR, The proposed [46] Picture Quality Scale distortion metrics showed a to correlate MOS correlation of 0.9 J. Pambrun, Used VSNR to correlate VSNR showed et al [47] subjective scores for a high correlation JPEG 2000 K. J. Kim, Used various objective metrics HDR-VDP metric et al [48] with subjective metrics is best correlated Di Lin, Used for multiple Reached to a et al [49] diagnosis purposes correlation of 0.98 elderly patients: several studies have shown that about half of people 65 years or older have a comorbid condition [33]. Thus, in [49], an image quality metric (defined as metric for comorbidity, IF C) applicable for multiple diagnosis purposes is proposed, and it is able to assess how well a medical image can be used for the diagnosis on a comorbid patient's conditions as well as on a single disease of this patient.…”