Ewert Bengtsson

Ewert Bengtsson är professor emeritus i datoriserad bildanalys med inriktning mot biomedicinska tillämpningar. Har varit verksam på området sedan 1973. Var med och byggde upp ämnet vid Uppsala universitet och skapade Centrum för bildanalys, CBA, 1988. Har sedan dess varit engagerad i CBA:s utveckling till en ledande bildanalysenhet. Är särskilt engagerad i att utveckla bildanalyssystem för tidig upptäckt, diagnostik och gradering av cancer.

My main research interest is in developing computerized image analysis for medical applications. I have been working in that field since my Master Thesis work in 1973-74. That work was dealing with Analysis of cell images using video and computer technology. My doctoral thesis in 1977 was dealing with the same subject. I have since then supervised 15 doctoral theses in that field. My most recently supervised PhD on that topic, Patrik Malm, worked on developing computerized analysis of PAP-smears for early detection of cervical cancer, the same problem as I was addressing in my PhD thesis some 50 years ago. My early research resulted in a working, but far from cost-effective, solution. Since then commercial systems have appeared on the market but they are also very expensive and only used in rich countries. My most recent work on that topic was to be involved in developing a new generation system, cost effective enough to be used in countries such as India where the main part of the project took place. It lead to a product that has been commercialized by an Indian company.

A former PhD student of mine, professor Joakim Lindblad is continuing this line of working and is now focusing on applying it to oral cancer which from an image analysis point of view has many similarities to cervical cancer.

Most of the work has been related to cancer research but our research has also found applications in solving cell biology problems. Digital imaging cytometry is a field of rapidly growing importance. One of my former PhD students, Carolina Wählby, is now professor of Quantitative Microscopy and is very successfully leading this research in close collaboration with the Science for Life Laboratory project.

But cell image analysis is not my only interest. I have also worked with other medical images e.g. from radiology (CT, MR) or nuclear medicine (PET, SPECT). We have also had a project on using haptic interaction to supplement the visual one in exploring medical images in particular for planning of maxillofacial surgery. Professor Ingela Nyström is continuing work on that project as well as other visualization projects.

Our research at CBA also covers many other things. The first full professor in image analysis at the centre, starting 1993, was Gunilla Borgefors. She is well established in the field of digital geometry where she has made several fundamental contributions. Those methods are used in many of our applied projects both by her students and mine. One of her former students, Robin Strand is currently head of the IT-department. He and Filip Malmberg are working in close collaboration with the Radiology Department at Uppsala University developing fundamental medical image analysis algorithms.

We also have research on several other image analysis application fields, computer graphics and visualization headed by another of my former PhD students, professor Anders Hast.

For a much more complete account of my research over the years I refer to the other web-pages at www.cb.uu.se e.g. the personal web-pages of my colleagues and students and in particular to our annual reports which each year listed all our ongoing projects, publications, collaboration partners and other activities up until 2021.

After my retirement I took the initiative of creating a spin-off company together with professor emeritus Christer Busch whom I have collaborated with since the early 1990-ies on image analysis applications to uropathology. The goal of the company is to create an AI and image analysis based way of objectively grading prostate cancer aggressiveness. Our special approach is to train our system on patient disease outcome, not on subjective ground truth. The results are promising, we outperform previously published prostate cancer grading results.