AGH UST researcher investigates diffusion signal in MRI brain scans even closer

The diffusion tensor model needs modifications

Piotr Włodarczyk, Centre for Communication and Marketing: How does your research change the knowledge about diffusion occurring in the brain and its imaging methods?

Dr Eng. Tomasz Pięciak: Until now, the literature has claimed that free diffusion occurs mostly in cerebrospinal fluid. Only some papers described its presence in white matter. Furthermore, only one research article suggested that the free water volume fraction [one of the parameters measured in diffusion tensor imaging – ed.], related to free diffusion in extracellular space, increases with age. What was lacking, though, was broad research to corroborate these findings. The investigation I conducted not only allowed me to confirm the matter, but also proved that the level of free water volume fraction varies depending on the area of the brain and increases considerably starting in early old age. Additionally, the research shed light on another quite important issue.

Until now, it was believed that the parameters obtained using standard diffusion tensor imaging changed in the ageing process, which had been interpreted in the literature in a number of ways. I was able to demonstrate that the observed changes in the diffusion tensor are, in reality, the result of the change in the aforementioned free water volume fraction. After having inspected it even closer, that is, considering this component in modelling the diffusion tensor, the changes in the diffusion tensor parameters over the course of life prove to be extremely subtle.

What signal processing methods did you have to use to obtain a more clear diffusion image than before?

We introduced a numerical method, verified earlier on in silico data [computer-generated data –ed.], and then on real data from patients. Moreover, a special technique of data analysis proved necessary, which allowed us to correctly perform statistical modelling of the registered changes. We performed a quantile regression analysis – it’s a particular statistical method that, to put it simply, allows us to model data in which the variance changes with the response variable – in this case, age. In relation to the example I am describing: the quantity parameter variability within the area of white matter in the brain between individual test subjects observed in young age is small and increases as the person ages.

Based on your findings, how does free diffusion change in white matter with age?

Depending on the area of the brain, for people between 70 and 80, the changes range from several to several dozen per cent in relation to the values registered for 20-30 year olds who were our reference group. We can say that these changes are quite significant. In my research, I also managed to determine that the free water volume fraction shows increasing inter-patient variability with respect to age.

How did you choose the research sample and what did the data registration process look like?

We tested 287 people in good health. Their condition was verified by comprehensive tests performed by a team of Norwegian doctors and psychologists, starting with the visual examination of brain scans by a radiologist, ending with IQ tests and a lifestyle interview. We excluded people with previous head trauma, neurological and psychological disorders, and those who admitted to taking drugs and regular alcohol consumption.

Each person who participated in the investigation was scanned once; additionally, 99 people were scanned again after ten to thirty months. Due to that, we obtained data on how the values of microstructural brain parameters, including the free water volume fraction and diffusion tensor parameters after accounting for the free water component, change with age.

New tool for neuroscientists

What are the perspectives that arise from the results of this research?

Knowing how the microstructural brain parameters change in healthy people with age, we can begin to study the deviations from it, that is, generate similar charts for people suffering from various neurodegenerative diseases, such as Alzheimer’s or Parkinson’s. Comparing them with the charts obtained for healthy people, we will be able to capture the onset of the pathology in the brain. This is extremely valuable because even in magnetic resonance diffusion imaging, where we use quantity parameters, the changes we register are very subtle in relation to various diseases and therefore difficult to detect. Another example is migraine, in which capturing changes in the brain comparing episodic migraine and chronic migraine is extremely challenging.

Does this mean that the much clearer diffusion image that you managed to obtain thanks to your research makes the past findings disposable?

I wouldn’t put it like that, but certainly this is the time to review them. In light of new facts, I’m not saying that a portion of the results wouldn’t have to be verified even closer. It might even turn out that, in some cases, we will arrive at reverse conclusions. This is frustrating, but this new investigation will definitely introduce a touch of uncertainty and excitement in academic circles.

What was the response of the scientific and medical world to these new findings?

They were first presented in May and June 2022 during the ISMRM congress in London and the Iberian ISMRM in Lisbon, where they were received with attention. Many people, not only specialists in magnetic resonance imaging but also neurology and related fields, expressed considerable interest in these findings in the context of research they carry out themselves. Thanks to this even closer inspection, they have the opportunity to re-evaluate the data collected in their research projects without the need to scan their patients again.

Currently, I also receive communication from psychologists and neuroscientists who study various brain disorders that may be captured by means of diffusion magnetic resonance imaging and who are interested in a cooperation in this field. As engineers, we provide tools for neuroscientists and they are the ones who offer us suggestions on what they would like to see and use in their practice. This generates other engineering problems to solve, often very complicated ones.

Since the presentation in Lisbon, my research has markedly advanced, and in the near future, I will be thrilled to present the most recent results on the ISMRM conference in Toronto in June. These new findings will shed some light on the microstructural changes in the brain that occur during the lifetime of an adult person, taking into account the free water component, going beyond diffusion tensor imaging.

Repeatability issues and lack of experts

Do the signal processing methods you are focusing on scientifically have a chance of having a more pronounced appearance in clinical practice?

From my conversations with medical circles in Poland, I know that this is something that doctors are interested in. Unfortunately, it takes years of practical experience to know how to correctly interpret such data. There are not that many centres in our country that deal with these techniques. In the United States, Great Britain, France, Benelux countries, or Germany, there are many more people who actively work on such data and have the necessary knowledge.

The other issue, this time global, is the level of repeatability of these diagnostic tests. It might turn out that the quantity parameters estimated on the basis of data collected in different facilities on the same type of scanner but with different settings may come out slightly off. Add to that the discrepancies that result from the particular specifications of different equipment producers. In the latter case, the deviations might even oscillate between ten and twenty per cent.