|Workshop on Standardization in E-Health
Geneva, 23-25 May 2003
User report on ITU Recommendation Clinical Study on MPEG-2 Compression Based on the Rec. ITU-R BT. 500-9
Isao Nakajima M.D., Ph.D.
Tokai University Institute of Medical Sciences
This clinical studies were performed at Tokai University Hospital in Japan to intend to identify the optimum
transmission rate of medical-use MPEG-2 motion pictures by the Double-Stimulus Continuous Quality Scale(DSCQS)
method. In addition, chronological fluctuations such as Jitter and Wandering, Block noise, Mosquito noise, Color
smearing are caused in image quality with MPEG-2 compression. MPEG-2 use coding that eliminates inter-frame and
intra-frame redundancy along the spatial and time axes. Though there are three (I, B, P) types of frames, the
motion JPEG uses only a single type. Since MPEG provides one of those three types of frames in an instant, there
is no point in extracting a single MPEG frame randomly for evaluation purposes. This implies that it is not useful
to perform an evaluation of image quality based on test signals. MPEG pictures should be evaluated subjectively
using images that have been obtained through practical user applications.
The test materials were images obtained by video endoscopy and CCD camera images of the affected areas of patients.
The images consisted of color NTSC signals(4:2:2).The DSCQS method enabled a detailed estimate of the degradation
in image quality to be made. However, the evaluation results for the examiners were found to be inconsistent.
ITU Rec. 500 recommends 4H(H: height of CRT) and 6H as the vision lengths, but we used 2H and 6H, as we did not
feel that subtle differences in image quality could be detected at the recommended vision length. We believe that at a length of 2H, differences in picture quality can be detected to obtain fine medical diagnosis. The Rec. ITU-R BT. 500-9 shall be very useful to evaluate suitable bandwidth to transmit motion picture to support clinical medicine.
[ Back ]