0.15 bpp ROI

0.4 bpp ROI

1.75 bpp ROI

Left CC

44.30 dB, 0.11 bpp

45.03 dB, 0.24 bpp

46.44 dB, 0.91 bpp

Right CC

44.53 dB, 0.11 bpp

45.21 dB, 0.22 bpp

46.88 dB, 0.85 bpp

Left MLO

44.91 dB, 0.11 bpp

45.73 dB, 0.25 bpp

47.28 dB, 1.00 bpp

Right MLO

45.22 dB, 0.11 bpp

46.06 dB, 0.25 bpp

47.96 dB, 0.96 bpp

Left side (MLO and CC)

44.60 dB, 0.11 bpp

45.38 dB, 0.24 bpp

46.89 dB, 0.96 bpp

Right side (MLO and CC)

44.88 dB, 0.11 bpp

45.63 dB, 0.24 bpp

47.41 dB, 0.92 bpp


44.74 dB, 0.11 bpp

45.51 dB, 0.24 bpp

47.14 dB, 0.93 bpp

Reprinted with permission from S.M. Perlmutter, P.C. Cosman, R.M. Gray, R.A. Olshen, D. Ikeda, C.N. Adams, B.J. Betts, M. Williams, K.O. Perlmutter, J. Li, A. Aiyer, L. Fajardo, R. Birdwell, and B.L. Daniel, Image Quality in Lossy Compressed Digital Mammograms, Signal Processing, 59:189-210, 1997. © Elsevier.

distortion measures to subjective quality is the choice of image distortions used in the tests. Some of the literature on the subject has considered signal-independent distortions such as additive noise and blurring, yet it has been implied that the results were relevant for strongly signal dependent distortions such as quantization error. Experiments should imitate closely the actual distortions to be encountered.

The assessment of subjective quality attempted to relate subjective image quality to diagnostic utility. For the MR study, each radiologist was asked at the time of measuring the vessels to "assign a score of 1 (least) to 5 (most) to each image based on its usefulness for the measurement task." The term "usefulness" was defined as "your opinion of whether the edges used for measurements were blurry or distorted, and your confidence concerning the measurement you took." The question was phrased in this way because our concern is whether measurement accuracy is in fact maintained even when the radiologist perceives the image quality as degraded and may have lost some confidence in the utility of the image for the task at hand. It is not clear to us whether radiologists are inculcated during their training to assess quality visually based on the entire image, or whether they rapidly focus on the medically relevant areas of the image. Indeed, one might

0 0

Post a comment