by Ryszard Tadeusiewicz

A machine that is smarter than a specialist doctor and can produce a reliable diagnosis-this is the kind of computer that Polish researchers already have in their grasp. This technology has attracted a lot of interest overseas, including from the United States and Japan.

Telemedicine, a flagship product of biocybernetics, will become one of the pillars of the information society of the future. The whole world is working towards this goal. Simply put, this means making the most of the latest technology to carry out the majority of medical services in patients' homes with the help of specialized apparatus connected to the internet. This especially applies to early diagnoses; long-term but non-intensive treatments; the monitoring of bodily functions in chronically ill patients, especially old people without families; rehabilitation after hospital treatment; and various other forms of medical care.

Dehumanized medicine?
Some people are afraid that the health services of the future may be dehumanized. While that may in part be true, from a technical point of view, this is huge progress.
Some say that the replacement of a doctor or caring nurse by a machine, however competent, is a form of dehumanization. However, we have no alternative. The world's population is aging at an alarming rate. At the same time, the extended multi-generation family, in which senior citizens could count on help and care from the younger family members, is gradually disappearing. As a result, a growing percentage of the population needs medical and paramedical care, while the number of people who are capable of providing such care is shrinking. One method to deal with this problem could be to improve the health service's technical capabilities and equipment, mainly through telemedicine. And in this field my team has made significant achievements.

We have developed many new ways of using information and data communication technologies to diagnose various medical conditions. Modern apparatus allows us to view internal human organs and facilitates their observation from many angles: how they are built, what is their internal structure and what kind of treatment they have undergone. However, a doctor may sometimes feel overwhelmed by the volume of information available to him. It is difficult for him to analyze and take full advantage of the information while diagnosing or monitoring a particular patient's treatment.

Computer analysis
The Biocybernetics Laboratory of which I am in charge was one of the first in Poland to bring together researchers and technicians to create a method for the visual support and control of many processes. Our goal was to build a computer that could precisely analyze and interpret what it sees. Just seeing was not enough. We needed correct diagnoses too. While computer software used for processing images and improving their quality and resolution is universally known and widely available, this is not enough, either in automated industrial systems or medicine. We wanted to replace the human eye, aided by the brain, with a machine. And we have managed to do so. This means more than just using computerized image analysis in industry. Computers all over the world are already capable of controlling an assembly line to produce cars built to client specifications. For this, an intelligent robot's eye is indispensable. Such technology is also useful in medicine, mainly in semi-automatic and remote image analysis.

Why a doctor needs a diagnostic machine
Isn't it enough for a doctor to view a computer printout, ultrasound scan or graph on his own? We are not talking about completely replacing an expert's eye and brain by a machine, but about aiding the human being. To bring the point home, here are three scenarios of diagnostic procedures:

Scenario 1
Patients, before being examined by specialists, are screened by general practitioners, who are not experts in fields such as cancer and may miss something on diagnostic images. A computer can give the doctor a signal and warn him that he should re-examine the image.
Scenario 2
In health screening programs-such as mass mammography sessions to screen for breast cancer-specialist doctors have limited time to look at thousands of images. A tired doctor may easily overlook a growing tumor. A computer could warn him that he missed something.
Scenario 3
A doctor may have years of experience diagnosing images, but with newer and better examination equipment more information is available, and some of it may go unnoticed if the doctor exclusively relies on routine diagnosis. Again, in this situation, a computer is indispensable.

Doesn't a computer make mistakes?
Of course, we all know that different people have different livers, hearts, lungs and so on. But this is the essence of our idea: we have devised a way of making a computer understand an image. We have programmed the machine in such a way so that it not only gives us an image of the organ in question but also analyzes it medically and offers an indication of what disease could have affected it. In other words, the computer is not only capable of providing detailed analysis but also of interpreting the data correctly.

Thanks to this process, it is possible to more efficiently use image information held in computer data banks. I must admit that searching for information in multi-media databases is a very difficult task. It is easy enough to find, via the internet, information on the pancreas, for example, but it is almost impossible to find images supported by relevant text. To find an online picture of an organ with the same characteristics as a diseased organ is even more difficult. A diagnosis based on such a comparison is unsatisfactory, especially in view of the minuscule differences between the same organs in different people.

An export hit
We have overcome this barrier and we now have a real export hit. Even the Japanese are interested in our computer. I have just returned from Kyoto, where I gave a presentation of our work. The Americans are also interested. During the most recent conference of the Institute of Electrical and Electronics Engineers (IEEE), the world's largest computer and electronics institution, I was asked to give a public lecture on automatic image recognition. Two special sessions were held exclusively on this topic later on during the conference.

What is interesting is that there are fewer enthusiasts in Europe. The Old Continent remains loyal to classical image analysis methods. But in June I will be lecturing in France, while earlier I was in Strasbourg and reviewed a doctoral thesis at the University of Erlangen near Nuremberg.

I must admit, however, that my studies at the Medical Academy, albeit discontinued in the third year, prepared me for the work that I have been doing on systems for the visual support and control of automated processes. These experiences have helped modernize the Cracow University of Science and Technology (AGH). Today our university stands out nationwide with its first-and for the time being only-Interdepartmental School of Biomedical Engineering in Poland. And I keep reading books on the structure and work of the human brain. To find out about the brain and its secrets is a real challenge.

The author is a computer scientist and biocyberneticist at the Cracow University of Science and Technology (AGH).

Prof. Tadeusiewicz's biomedical engineering research fits into the idea of an information society as promoted by the United Nations, which passed a resolution on March 27, 2006 to declare every May 17 as Information Society Day across the world.