A teleradiology network and an artificial intelligence-based diagnostic report analysis system have revolutionized the treatment of stroke, one of the leading causes of death, in Hungary. The initiative, launched by the UP Department of Medical Imaging, will increase the chances of survival and recovery for stroke victims severalfold and will also provide equal opportunities for people living in South Transdanubia.
Written by Miklós Stemler
40,000 cases a year, around 13,000 deaths, tens of thousands of serious, permanent health problems: these are the grim Hungarian statistics on stroke. The fact that most patients could be saved by the latest treatments is both hopeful and frustrating, but despite this, only 10% of them receive the appropriate treatment in time. An initiative in Pécs has turned these figures around in South Transdanubia, and all it took was regional cooperation and the truly 'smart' use of modern technology.
Time: the deadliest war machine of all
"There are basically two types of stroke: the first is when there is sudden bleeding in the brain, while the second is a blockage of blood vessels that cut off the blood supply to certain areas of the brain. Unfortunately, there is little that can be done about the former, but in the case of the latter, which accounts for 75-80 percent of stroke cases, the vast majority of patients treated in time can be cured with minimal or no residual harm with the help of a catheter-based method, known as thrombectomy, which has become popular in recent years," explains Dr. Tamás Dóczi, academician, full professor at the UP Department of Neurosurgery and one of the developers of the concept.
Time is indeed a critical factor, and very little of it is available. Areas deprived of blood supply will sooner or later suffer irreversible damage, and there are only a few hours for a stroke patient or their relatives to call the ambulance, be taken to the hospital, have the necessary tests carried out and assessed, and then be transferred to a medical team with the appropriate skills. Every minute really does count, and although the previous three-hour time limit has been extended in recent years to up to 24 hours in some cases, everything must (or should) be done like clockwork. In addition, the chances are also determined by the health care system: for example, patients living near hospitals in Pécs or Kaposvár with a so-called neurointervention centre (of which there are 6 in the country, while there are 39 stroke centers to receive patients) have much better chances than those living farther away.
Trained radiologists are just as important as the staff performing the procedure. "The role of diagnostic imaging has also changed with the expansion of therapeutic options: whereas previously the primary aim was to distinguish between hemorrhagic and occlusive strokes, now a more precise analysis is required. We need to determine the location of the vascular occlusion, the extent of the damage, and the area that can be saved. These factors are assessed using the so-called ASPECTS system: in some cases, it may turn out that the intervention would not benefit the patient and would only cause unnecessary stress, while in other cases, it may turn out that it is worthwhile to perform it beyond the default time limit," says Dr. Péter Bogner, Head of the UP Department of Medical Imaging, the initiator of the concept.
This means that ideally, patients should be admitted 24 hours a day at a stroke centre by an experienced radiologist who will carry out a complex examination, evaluate it in a very short time and, if he or she deems it appropriate, recommend the patient to the nearest neurointerventional centre, while consulting with the specialist performing the thrombectomy. There is a lot that can go wrong in this complex process, resulting in either a patient not receiving care in time, or the procedure being performed on a patient whose condition will only be worsened by it. "We need high-quality, uniform diagnostics that can be used to make decisions within minutes and then transport patients because the availability of the equipment and specialists is for nothing if the patient does not arrive," says Tamás Dóczi, summing up the challenges.
Teleradiology instead of television
The team of experts from the Department of Neurosurgery, the Department of Neurology, the Department of Medical Imaging, and the Pécs Diagnostic Centre has developed a solution to this problem, which will allow the evaluation of the findings based on a uniform methodology and all patients admitted to a hospital with stroke in the region may have an equal chance. The solution is based on teleradiology, i.e., the sharing of imaging results between institutions. Although it sounds simple in theory, there are several problems to be solved in practice. High and stable network bandwidth is important because thousands of images of a single patient may be needed, resulting in gigabyte-size files. In addition, to get reliable results, images need to have a standard format so that all images taken with different devices can be assessed: a software called eRAD provides a unified platform for this.
Teleradiology allows the immediate sharing of findings; the next step is to assess them quickly and reliably. This is where artificial intelligence comes in, which is most powerful in tasks like these, involving large amounts of data but with unified parameters. The so-called e-Stroke Suite, developed by Brainomix, performs the assessment using machine learning algorithms optimized for the CT scans of stroke patients. While the speed of machine intelligence far exceeds that of human intelligence, the AI system can be used to eliminate the potential for error under different conditions - after all, as Péter Bogner puts it, you have to examine deformations that are barely visible to the human eye, where even the slightest change in lighting conditions can cause serious deviations, not to mention the typical problems of acutely acquired images (such as head movements), which the software automatically compensates for.
Of course, software using artificial intelligence is expensive, which makes it difficult to access. However, Péter Bogner's idea has led to the development of a network version that allows stroke cases arriving in Pécs, Szekszárd, Baja, Nagykanizsa, Zalaegerszeg and Szombathely to be assessed in a uniform and quick way, and then, thanks to the smooth cooperation with the National Ambulance Service, they are sent as quickly as possible to the Kaposvár or Pécs center where the thrombectomy can be performed and prepared staff receives them.
A truly smart investment
The results of the initiative, launched in 2019, are dramatic. While nationally, only around 10 percent of patients eligible for thrombectomy undergo the procedure in time, the figure is around 50 percent in South Transdanubia, which is close to the Western European 70 percent rate. However, there is still room for improvement, and significant developments are planned in the project, which will be transferred to the Translational Neuroscience National Laboratory soon.
One of these developments is to further refine diagnostics to ensure that a higher proportion of patients may benefit from the intervention in time and that those whose condition cannot be improved by the intervention are screened out, as unnecessary interventions tie up valuable resources, in addition to the fact that thrombectomy may even worsen their condition. The second development is to extend the concept to the national level so that it is available to every stroke centre in Hungary and to the patients arriving there. "This is how equal opportunities are provided: the same universal knowledge is available to doctors at a small-town hospital or at a Budapest clinic," says Péter Bogner.
The annual cost of the system - the software license fee and the price of the server - is around HUF 20 million (paid for the first two years by the Pécs Diagnostic Centre), which cannot be compared to the cost of one stroke patient with severe permanent damage to their health dropping out of the workforce and the cost of their care. "The current NEAK guidelines on the inclusion of new therapies and drugs in social insurance state that the cost of gaining a so-called 'useful year' should not exceed HUF 20 million. With this concept, we can help thousands of patients with a similar level of expenditure, and South Transdanubia is just the beginning: the healthcare government provided funding for the expansion to the central region in 2020 through the tender titled Transnational and innovation development of national institutes," summarized Tamás Dóczi.
The method can also work for other diseases and diagnostic tests as well: the Hunchest2 program was another initiative in Pécs, which networked lung cancer screening and assessed the results using AI, and screened thousands of high-risk patients.
This article was published on hvg.hu on 25 October 2022.