- 期刊
以客觀評量數學式比較MRI影像經失真壓縮後之影像品質
Image Quality Evaluation on Compressed Abdominal MR Images by Using Objective and Mathematical Methods
摘要
本次研究目的在於評估各種客觀數學式對於MRI的壓縮影像品質差異偵測敏感度實驗，客觀的方法以傳統點對點測量法(pixel based-metric, PBM)有均方誤差(MSE)、訊雜比(SNR)、峰值訊雜比(PSNR)及窗對窗測量方法(window based-metric, WBM)有Qindex、MSSIM、MPR來作為影像品質測量工具。透過JJ2000與Apollo兩種同樣以離散小波轉換(Discrete Wavelet Transfer, DWT)作為核心的影像壓縮軟體來對MRI影像進行小波壓縮，影像的壓縮比設定10個等級，將30張的MRI腹部T1影像從10倍至100倍以間隔10倍做為一個單位進行壓縮，最後將壓縮影像使用各種客觀數學方法來進行品質測定。結果發現從MSE、RMSE、SNR及PSNR的傳統點對點測量法中得知，JJ2000之表現優於Apollo；然就本研究整體而言，MPR測量的H index具有較佳之偵測敏感性，並可建議做此類型影像時之壓縮測量品質之工具。
並列摘要
Medical images generated daily with developing diagnostic modalities. The purpose of this study was to survey different objective methods on compressed abdominal MRI images. The researchers conducted two commercial software, JJ2000 and Apollo, to detect the sensitivity of these images, respectively. The core of algorithm in this study was Discrete Wavelet Transform (DWT). 30 T1 MR images were compressed from 10:1 to 100:1 with 10 interval each. The compressed images then were evaluated by the algorithms of Mean Squared Error (MSE)、Signal-to-noise ratio (SNR)、Peak SNR (PSNR)、Q index、Mean Structural Similarly (MSSIM)、and Moran's Peak Ratio (MPR) to detect the degree of image degradation. Two objective metrics were used to compare the differences between compressed images. There were pixel based-metric (PBM) and window based-metric (WBM). The results find that JJ2000 is superior than Apollo in terms of test by MSE, RMSE,SNR and PSNR. The study suggests that H index in MPR could be applied to these kind of images.