1. Where did you get the idea to face CT?
Szymon Korzekwa, CEO of BrainScan: One of the company’s founders – Marek Trojanowicz, who graduated from computer science 28 years ago at Gdańsk University of Technology, worked on the theory of artificial intelligence. Even then he dreamed of the times that have now come – when the theory developed three decades ago will meet the computing power that allows its verification – the verification of theory in practice. Five years ago, on our way, we met Adam Brzeski, then one of the best graduates of Gdańsk University of Technology, who was starting his doctorate. It was he who came to us with the idea of using machine learning methods to recognize images of imaging diagnostics of the gastrointestinal tract – exactly from the images recorded by capsule endoscopy.
That’s how it all started. We invested our own money, built a team of six AI engineers, obtained image data – that’s how our five-year journey began. The capsule endoscopy project showed how there is a path between the idea and the implementation of the production software, between the results of the tests carried out in our offices and the results coming from the solution implemented at the customer.
2. If they could explain in a few sentences what BrainScan is and how it can help patients, doctors and medical institutions alike.
Sask. : BrainScan is a commercial company, its foundation – like any other company – generates profits. Of course, building a business whose source of revenue largely depends on research and development is a doubly difficult task to achieve. Why? Because before the step of – as it is nicely called – the commercialization, it is necessary to plan the resources, the time, and the team to prepare a starting solution, a product in its first commercial version.
In the case of BrainScan, it took more than three years to come up with this product. We started with the image analysis of capsule endoscopy, and the current company and its successfully implemented product deals with the analysis of brain computed tomography imaging examinations. We recognize more than a dozen pathological changes using machine learning methods, creating a tool that significantly supports the work of a radiologist.
In reference to your question, if we provide radiologists, tele-radiology companies, hospitals – with a tool that makes their work more efficient, we help patients in a natural way. In the first two cases, if the doctor can describe more tests at the same time, and with greater certainty that no lesions will be missed, the patient wins. In the second case, if the hospital receives a tool from us that allows an effective trio of patients awaiting intervention, we are talking directly about helping to save human lives. Such immediate and precise medical segregation of patients in the stroke emergency department is crucial for the life of the patient and the continuation of his therapy.
3. It’s no secret that imaging tests in Poland, especially at the National Health Fund, often have to wait a long time. Is there a chance BrainScan will really change it?
Sask. : As I mentioned before, there are two areas where our solution helps improve patient diagnosis. While in the aforementioned implementation of our solution in hospitals we facilitate trioscopy, in teleradiology companies we increase the efficiency of the work of the radiologist who describes the imaging test. With the use of our software, he can describe more tests, which directly reduces the time between when the test is performed and when the patient receives the description. Responding briefly – yes, we shorten this time, moreover, the system is able to detect very small changes that are easily ignored by a human, especially in the sixth – seventh hour of his work. The system, detecting all pathologies, will give the radiologist a second opinion with a list of detected changes, which gives greater certainty and precision.
The BrainScan system should be considered the next generation tool provided to the physician. After all, some of us still remember how the X-ray machine was the basis of imaging diagnosis, and the image taken on the film attached to the negatoscope was viewed by a radiologist. Then came the era of CT scanners and magnetic resonance. Digitally recorded images have enabled the development of software, which allows image processing to extract as much diagnostic information as possible from the examination. It provides the doctor with many options to change display settings, image reconstruction, filtering and many more – that was the second step. Now is the time to take the next step, introducing further improvements, which are possible through knowledge and technology, and which always help the patient in the end.
4. When will the average patient in Poland be able to benefit from your solution? Is there a chance that its use in our country will become widespread in the next five years?
Sask. : Some of our patients can use our solution for a year. We work on a commercial basis with several companies, such as PZU Zdrowie. At the same time, we are continuing our discussions with other customers in Poland. These are private entities – commercial networks of medical facilities providing diagnostic imaging services. But we are also talking to many state actors.
Of course, I am not in a position to tell you if everyone in Poland will benefit from our solution within five years – although this is a very good prospect for me, I would very much like that to happen. However, I am confident that systems such as BrainScan will become commonly used tools in diagnostic imaging in the years to come. Just like more than ten years ago, the fluorescent lamp backlighting the X-ray examination has been replaced by the monitor and increasingly perfect software allowing the manipulation of imaging tests, obtained with diagnostic methods of increasingly modern imaging, such as CT, MRI, PET-CT, PET-MR, etc.
5. BrainScan, using machine learning solutions, is constantly being developed. How effective is the tool now? How has this changed over time?
Marek Trojanowicz, co-founder of BrainScan: The collection of brain computed tomography imaging exams collected by the company includes several hundred thousand unique exams. Currently, as we mentioned earlier, we recognize several pathological changes, but approximately every six months we update both the number of lesions recognized and the efficiency of their detection, and, above all, the identification of the areas of the brain in which they occur. This number of exams is a huge amount of information to process, it must be remembered that each scan is made up of successive slices, and the number of unique images that we process is the product of the number of exams and the average number of cuts. To illustrate the scale, if we assume the average number of slices is 50, we are currently analyzing about 15,000,000 images. Each of these thousands of imagery studies has a text description that also needs to be interpreted when using text analysis methods. To come back to your question – we currently recognize more than ten pathological changes, each change is diagnosed with an accuracy close to 99% – that is 99 correct diagnoses for 100 cases.
6. To what extent will you operate with BrainScan? Are you going to focus mainly on your activity in Poland or rather expand it gradually to other markets?
MT: BrainScan is a complementary solution, more than just a tool. In fact, from the perspective of the client – radiologist, teleradiology company, hospital – it is a tool. But from a technical point of view, it is a complete IT solution that integrates seamlessly with the hospital’s information flow system, allows you to send anonymous imaging tests of PACS systems from the hospital to our cloud computing, run analyzes on it and return the result. The entire transfer and scan operation takes less time than the time it takes for the patient to return from the CT room to the emergency room.
Sask. : So the solution is fully scalable, today it works in branches in Poland, but we are already having discussions in Europe. Thanks to the received certification of the medical device, the CE certification – the BrainScan system can be sold and implemented throughout Europe without any problems, which we want to implement in 2022.
7. It seems that artificial intelligence has definitively imposed itself in medicine and today the use of AI solutions is no longer surprising. If you could identify three key areas, outside of imaging studies, where AI can help physicians, what would it be?
MT: Currently, methods known in terms of marketing such as Artificial Intelligence work very well, mainly in the areas of classification. In addition, they have the particularity that the number of colors, the number of shades of gray, the multidimensionality or a multitude of sections are not a barrier for them. For example – the human eye theoretically distinguishes 500 shades of gray, while current devices return an image whose elements have from 4,000 to more than 65,000 values. Although our brain probably does not compete with digital methods and is still, as a small biological creature, capable of abstract thought, our receptors, such as sight, smell and touch, are often not as efficient.
When it comes to identifying areas where machine learning methods will be useful, there will certainly be work where the analysis of a huge amount of data is required, an amount so large that it is necessary to adopt heuristics in their analysis. Good examples here are DNA strand research, vaccine research and mRNA drug research – opening a new avenue for the treatment of oncological diseases. But also more “tangible” orientations, like the whole field of intelligent prostheses. Nowadays, it is possible to think of intelligent devices of sight, hearing or movement.
Unfortunately, we will never escape the question posed in 1955 in Stanisław Lem’s radio play – “Is there a Lord, Mr. Johns?” – but that’s a topic for another interview!
Also read our other Digital Health Magazine content