Kunio Doi, Ph.D., retired in September after 40 years of service to the Department of Radiology and The University of Chicago. Professor Doi is an internationally renown medical imaging physicist whose research has provided much of the scientific basis for the assessment and improvement of image quality in diagnostic radiology and for the development of computer-aided diagnosis in that field.

His career began in 1962 at the Kyokko Research Laboratories, Dai Nippon Toryo Co., Ltd., in Tokyo, where he served as Chief of Radiography Research and achieved international recognition for his pioneering work in applying transfer function analysis and Wiener spectral analysis to the evaluation of spatial resolution and noise in medical radiographic systems. Professor Doi joined The University of Chicago’s Department of Radiology as Research Associate (with the parenthetical rank of Assistant Professor) in 1969, at the invitation of the late Kurt Rossmann. He was promoted to Associate Professor with tenure at The University of Chicago in 1972, became full Professor in 1977, and was awarded a University Chair as Ralph W. Gerard Professor in the Biological Sciences in 1994. Professor Doi has served our department as Director of The University of Chicago’s Kurt Rossmann Laboratories for Radiologic Image Research since those laboratories were created in 1977; as Director of the University’s Graduate Programs in Medical Physics (offered jointly by the Department of Radiology and the Department of Radiation and Cellular Oncology) from 1985 until 1998; and as Associate Chairman for Research from 1994 until 2003.

Professor Doi’s standing in the international radiological community is reflected by his work as a consultant to the International Commission on Radiological Units and Measurements (ICRU) from 1965 to 1979; by his service as Chairman of the ICRU’s report committee on Modulation Transfer Functions of Screen-Film Systems between 1979 and 1986; and by his election as a Commissioner of the ICRU in 1989. At the national level, Professor Doi served as a member of Subcommittee PH 2-31 of the American National Standards Institute (ANSI) from 1982 until 1995; and as Associate Editor of Medical Physics from 1986 until 1989; and as a member of the Integration Panel for the United States Army Medical Research and Material Command's Breast Cancer Research Program from 1997 until 2001.

Kunio Doi played a seminal and preeminent role in laying the foundations for modern medical x-ray imaging research. With Rossmann and Morgan he was the first to apply and encourage the use of modulation transfer function (MTF) analysis for quantification of spatial resolution in radiography, and with Rossmann he introduced Wiener spectral analysis for the quantification of noise in radiographic screen-film systems. However, a continuing theme of Professor Doi’s research has been its dedication and relevance to the improvement of diagnostic image quality and the reduction of patient exposure in angiography, skeletal radiography, chest radiography, and mammography. For example, he played an early role in both theoretical and experimental studies concerning the effects of the x-ray spectrum on patient exposure, image contrast and noise. He established the use of Monte Carlo simulation to study the effect of antiscatter grid design on scattered radiation, and he developed the first practical stereoscopic magnification technique for angiography. He was also among the first to employ observer-performance experiments to investigate the impact of image processing on lesion detectability in radiography.

Perhaps Professor Doi’s most revolutionary contribution to radiology, however, began in the mid-1980s, when his research shifted toward the development of techniques for computer-aided diagnosis (CAD) in digital radiography. At that time, methods for automated image analysis had been investigated for at least two decades in computer science, where the work was guided primarily by theoretical considerations and had achieved success in detecting largely geometrical objects, such as buildings and armored vehicles in defense applications. However, attempts by others to program computers to “find” abnormalities in digitized radiologic images had been disappointing. Professor Doi perceived two aspects of the problem of automated image analysis in radiology that others had overlooked: the computer’s goal should be to aid the radiologist, rather than to replace him or her; and computer aids in radiology can be developed successfully only by taking the insight and skill of radiologists into account. Professor Doi’s vision has proven to be correct on both counts, with CAD schemes developed in our department — and subsequently by scores of other research groups around the world — now becoming a routine part of radiological practice.