SSIMA – a new step towards the future of medical imaging

There are good things happening in Romania, and one of them is called SSIMA, that is International School of Medical Imaging. For five days, at the beginning of the month June, the fourth edition of this salutary initiative brought to the students 15 professors from prestigious universities and representatives from the equipment industry medical and business sphere, who lectured and workshops for those 70 participants came to Sibiu from over ten countries, including Canada, USA, France, Spain, Great Britain, Finland and Switzerland. Among the guests of this edition were Michael Lustig (Berekeley University), one of the most vocal names in technologies based on nuclear magnetic resonance, Ovidiu Andronesi (School Medical Harvard), a specialist in precision oncology and in radiogenomics and Kim Mouridsen (Aarhus University, Denmark), specialized in early detection of a diseases with artificial intelligence. I asked all three of them questions. Here’s what’s out!

Michael Lustig – University of Berkley

Photo credit:

How would you define medical imaging and what does “compressed detection” mean, in which you specialized?

Medical imaging is revolutionary for medical diagnosis by allowing the physician to look from outside to a person, which was made possible at the beginning of the last century with X-rays. As for computed detection, let’s look at magnetic resonance imaging. Collecting data by this means to create a complete picture of what we want to see takes time. And this is a limitation, MRI is very slow, which becomes a problem if you want to capture “dynamic scenes” such as beating hearts, breathing in the lungs or children’s patients. It’s very difficult.

But today, for example, when you take a picture or a digital camera shot, the device stores the file in its memory at a much smaller size than it should have given the number of pixels surprised. And this is possible through the compression / compression process that keeps the information of a frame without loss, meaning there is no loss of data at the level of perception of what is in the picture. And then you ask why accumulate all the pixels just to throw a good part of them, as a result of the computation. As such, compressed detection, which has a lot of math in the back that was developed ten years ago, involves data accumulation by means of fewer measurements and subsequent rebuilding of the desired image even if you have not collected the data completely. MRI is very well on this approach.

And this technology allows more efficient capture of more dynamic processes, what are you talking about?

Yes, it can be used in many ways. It can scan faster, for example, a reading that normally takes 4-5 minutes can take half a minute now. Or the examinations in which a patient had to inspire and expire often can now be done even with one deep breath. It is even possible to scan a moving, breathing accelerator and is not as quiet as it may be needed in a classic MRI exam.

What is the place of artificial learning / artificial intelligence in medical imaging? And how does it affect doctors?

Through automated learning, it is allowed for the AI ​​to decipher and use important hidden information in the data. I’m given lots of data sets to learn how to diagnose certain diseases or how to characterize a certain type of images. Today, we are collecting huge amounts of information and I believe that the advantage that artificial intelligence really gives us is the ability to understand the essence of this information and to use it to speed up certain processes to discover new things. There are many small keys hidden in data, not exactly obvious, which man can neglect, but a machine can detect them if properly trained.

Operations automation is an important step in medical diagnosis, and the AI ​​can extract valuable information about a patient by looking at patterns that affect their health. As to the role of physicians, I believe artificial intelligence will support them as an excellent monitoring tool that will take them from the pressure of tedious analysis activities that are more appropriate in the end for a robot, and will allow them to focus more on areas of health care where critical and creative thinking is needed. Surely the AI ​​will greatly improve the prevention capability of medical systems and will allow early detection of early or early symptoms or symptoms that we do not yet understand.

Do you think technological advancement will allow people to become, in any way measure their own doctors?

This trend can be more and more noticeable, because people use mobile devices to measure their pulse, tension, number of steps they take, sleep cycles, etc. At the same time, I can draws some hasty conclusions, which are not necessarily supported scientifically and medically, which is a cause for concern. A medical direction in which technology proves to be really useful is to empower people in remote areas that lack diagnostic tools to access such services through telemedicine – collecting data in one place and analyzing them by a expert at another location. More and more people will benefit from this.

What are you working on now and how much progress you would like to make in the future? very distant?

Over the past eight years, I have worked with several colleagues at the Stanford Pediatric Hospital to eliminate the need for anesthetized patients to be examined with magnetic resonance imaging devices. Through this project, which is also my passion, we have tried to develop child-friendly equipment that is more suited for them and which can extract clearer images, we have tried to develop faster scanning tools, ways to overcome the problem of patient movement, and we get the safety of the diagnosis under these conditions, so that the doctor does not need to sleep the baby to scan it properly. Because if you have to put on a MRI a child aged 1-3 years, he does not understand the notion of staying motionless or breathing.

Therefore, because examination is doomed to failure, it is necessary to use the anesthetic, but this comes with risks and complications in such a young patient, there are possible side effects known at the neurological level. In addition, CT (computed tomography), for example, involves ionizing radiation which, because a child needs to be scanned several times, comes with cancer-causing side effects. It’s a delicate situation. That is why we want to make sure that the MRI technology we produce is fast enough and robust to convince the doctor that it can provide a correct reading without anesthetized child. That’s what I’m working on, and that’s what I’d like to see that happens: imaging devices dedicated to children, smaller devices that are both performing at the same time, because even reducing size allows for faster reading. Also, the sensors and antennas used in these procedures do not fit well with children and we are working on an adaptation in this regard as well.

Ovidiu Andronesi – Harvard Medical School

Photo credit:

What does the radiogenic mean to ordinary people?

As the name calls it, radiogenomics is a combination of radiology and genetic testing, applicable in oncology. By combining imaging data with genetic or molecular data, patients can be categorized more accurately than possible before using molecular markers when patients were classified more by anatomo-pathological aspect. The use of radiogenicity is superior because there are large differences between patients depending on how they respond to the treatments, the survival rate, the type of treatment performed, etc.

Many cancers previously categorized into large groups are now subclassified into much smaller groups based on molecular markers – by the presence of certain proteins or genetic mutations that are important in terms of treatment and prognosis. This is combined with the development of certain treatments, especially on the basis of chemotherapy or immunotherapy, which specifically target genetic mutations in order to produce as few collateral effects as possible. But these treatments are very expensive and sometimes, even if they try to minimize the side effects, they occur. And then you want to apply the treatments to the patients who benefit from them. Radiogenomics is part of a “holy third” of cancer treatment, along with personalized oncology and precision oncology, with which it is often intertwined.

Will prevention take place in the fight against cancer?

It’s quite complicated with prevention at this time, because when you detected the mutation or molecular marker it’s a little too late. For prevention, early detection of cancer is needed – for example, when there are only a few tumor cells in the body, because you have a much greater chance of taking action so that the patient can survive for another 20 years, as if he would never have suffered from this disease. With the available imaging methods, cancer is detected at the level of thousands of cells at a minimum, which is already very much, that is, at that level it is an already well-formed tumor. But with the technological advance, this limit is still reduced, so if we move from the ability to detect 1,000 tumor cells to the performance of detecting 100, it is already considerably improving what we can do for the patient.

The cancer, described very minimalistically, is a cycle in which genetic mutations, metabolic alterations and epigenetic changes are interconnected – at the level of DNA packaging, packaging that is exposed to the external environment or exposure. And more prevention in this area can be done by limiting exposure to what we know to be harmful. It is true, and on the basis of the genealogy tree, prevention can be done because there are genetic predispositions against which, when detected, action can be taken. Only they are sometimes unpleasant, even mutilating, such as preventive mastectomy.

How advanced is the brain reading technology and how much we understand in the moment this organ?

If we are talking about imaging methods, there are two research agency initiatives now in the US and the EU that want to push the limits of human brain knowledge as an organ towards the level of observation of the individual neuron. However, what is feasible now with imaging techniques, MRI, PET or others is to aim for a mesoscopic brain knowledge, that is, an intermediate scale, neither macroscopic nor microscopic. At mesoscopic level, we are talking about observing a few hundred neurons, and things are quite interesting here because cognitive functions and brain function are given not by how a single neuron works, but by the group activity of a neuron network.

Another technique useful for understanding how brain connections are organized is called “diffusion tensor imaging”, which can determine how two extremely distant regions of the brain are connected, based on the diffusion of water molecules along the axons (neuronal bodies). Also via the functional MRI, it is possible to monitor the specific activity of a particular region in the brain by providing instant feedback from the device, which checks the proper functioning of different brain areas. We are not yet at the point of using a scanner to read a person’s thoughts, but there are researchers and laboratories working to do that.

What challenges do you face in the sphere of activity that you have?

In my field, it is important to do studies on human subjects, and this can be very costly and complicated. Because there is a need for a very large sample of subjects whose co-operation involves both substantial financial, logistical and human resources. But, after all, it is all about the financial effort. In the US, for example, the federal health agency has an annual budget of about $ 30 billion for all of America, for which a variety of research groups apply for projects, of which only about 10% receive funding. That’s why the chance to receive money is very small and you apply at least ten different conventional projects or a very good, really revolutionary project. In this, you have to combine an exceptional technical idea with the practical part, bringing a clear benefit to society, such as, for example, revolutionizing the way we treat cancer or helping people with depression. It is a very competitive area, where it is hard to always find the optimal combination of elements to attract funding. This is the biggest challenge.

Technology of The Future – Romanian point of view