Rosita Shishegar has data on the brain

July 27th, 2023

Dr Rosita Shishegar and her colleagues are using data science to facilitate Alzheimer’s disease research.

Rosita Shishegar works at CSIRO’s Australian e-Health Research Centre.

Want to find out what motivates Rosita and how she created the largest Alzheimer’s disease dataset in the world? We explore the fascinating life and work of Rosita Shishegar in this deep dive republished here from CSIRO.

Meet Rosita Shishegar, a research scientist at our Australian e-Health Research Centre.

She’s using digital technology to improve the health and wellbeing of Australians. Her work in data science is facilitating research on neurodegenerative disorders such as Alzheimer’s disease.

Rosita has always had a head for numbers. But it’s not data that inspires her work: it’s the people behind it.

A soft spot for brains

Rosita didn’t expect to find herself studying brain disorders and collaborating with neuroscientists.

After completing a diploma in maths and a bachelor’s degree in engineering, she studied a master’s degree in electrical engineering. Her minor was bioelectronics, so her supervisor suggested she engage in a study of the brain. It involved looking at brain MRI scans.

“As an engineer and someone with a background in maths, I was a bit scared of the inside the human body at that stage. I convinced myself that what was on the scans was just a shape; I was studying the shape of a ‘thing’ that just happened to be inside the brain,” Rosita says.

Yet, she found herself leaning into the human aspect of her work. “I found that I enjoyed studies of the brain. I have a curious mind, and there are a lot of mysteries about the brain and human beings,” she says.

Neurodegenerative disorders are one area where there is still much to learn. After her PhD, which focussed on foetal brain development, Rosita studied Huntington’s disease and Friedrich’s ataxia, two rare neurodegenerative disorders.

“I was, and still am, motivated by wanting to create change and make people’s lives better,” she says.

[Music plays and several technology related images appear inside a circle on the screen. The CSIRO logo appears from the circle. The vision cuts to Pierrick Bourgeat, CSIRO Research Scientist, who is seated in a loungeroom environment on a cushioned chair with soft lighting.]

PB: Alzheimer’s is a very complex disease and it’s very hard to find a cure because there’s a lot of things we still don’t really know.
Screen cuts to Rosita Shishegar, CSIRO Research Scientist. Rosita is being interviewed via Zoom and is pictured sitting at her home desk.
RS: We are living longer and understanding diseases such as Alzheimer’s disease that is related to age is very critical to understand how it works and how we can prevent that.
Screen cuts to Pierrick Bourgeat.
PB: So for a healthy functioning brain, you need to have all your neurons to be healthy. With Alzheimer’s disease, you’ve got first the protein called amyloid, which starts to accommodate in your brain. Once it reaches a certain threshold, what we believe is that the tau protein that starts to accumulate. Those tau proteins are very toxic to neurons and they basically start to kill them off, which basically leads to cognitive decline or loss of memory. And at that stage that’s usually when the disease is being diagnosed.
Screen cuts to Filip Rusak, CSIRO Research Scientist.
FR: In Alzheimer’s disease research, one of the biomarkers is the thickness of the brain cortex. Brain cortex is a thin layer of grey matter that sits on the top of white matter and is used for thinking and all cognitive functions. As we get older, the cortex is getting thinner and thinner due to the neurodegeneration and it’s shrinking, which is a part of the normal aging process. Up until now, we could not measure it accurately because we can measure it only once the person dies and even then, these measurements are not very accurate because our brain starts to shrink as soon as we die.
Screen cuts to Rosita Shishegar.
RS: For different purposes of Alzheimer’s disease early detection or distinguishing risk factors for the disease, we need to have a lot of data. So one of my research aims these days is to provide a huge data set for artificial intelligence. When I think of AI, I think of my 16 months old baby girl that she’s just learning about the world. A few days ago, I realized that she’s pointing to an elephant on her mat, calling it a dog. And it reminded me of AI and the importance of data set. This little girl, she called a dog. And AI is the same.
Screen cuts to Filip Rusak
FR: In our research, we used generative AI in order to create fake brain MRI images with accurate cortical thickness measurements.
Screen cuts to Pierrick Bourgeat.
PB: My research is all about image analysis, so image quantification, and how to use software and algorithm to automate the extraction of information, the quantification of information from images. What we developed through the use of machine learning is a software that allows to do that special PET quantification without the need of an expensive MRI, which basically reduces the burden on the health system and allows to streamline the quantification of the PET images, making it more efficient.
Screen cuts to Filip Rusak.
FR: The advantage of using AI in Alzheimer’s disease research is to simulate brain MRIs with different brain conditions, which may not be as easy to acquire in reality. So, these fake MRIs gives us the ground truth, which enables us to evaluate different cortical thickness measurement methods in different brain regions and pick up the right tool for the job.
Screen cuts to Pierrick Bourgeat.
PB: So the current hypothesis is that amyloid will accumulate first and lead to TAR accumulation. If you can remove amyloid before TAR starts accumulating, you would basically remove the protein that kills the neurons and hopefully be able to prevent Alzheimer’s.
Screen cuts to Rosita Shishegar.
RS: Therefore, doing early detection and very early treatment is probably a key to find a cure. If we provide artificial intelligence with enough data, it can make predictions very accurate, much more accurate than human beings can do. And we already have large data sets in Australia, in US. And what I did is to create an algorithm that harmonize and align these data sets together. As a result, we created the largest Alzheimer’s data set around the world that can help us to realize and learn a lot about the disease.
Screen cuts to Filip Rusak.
FR: Accurate cortical thickness measurements are important. The first example would be in tracking the progression of Alzheimer’s disease in patients. The second reason is that we can test the efficacy of drugs in clinical trials aimed for combating Alzheimer’s disease.
Screen cuts to Rosita Shishegar.
RS: So we have done the hard work. And the next phase is to use this data set to, for example, look at genes and how they may affect the progress or onset of the disease.
Screen cuts to Pierrick Bourgeat.
PB: Some of the work we also are currently undertaking is looking at blood tests. And we’re actually using the PET quantification that is obtained from our software to evaluate how robust those blood tests are. And the very strong impact of that research will be once we can find some very robust and very sensitive blood tests, because then you’re going to be able to deploy those everywhere in memory clinics and be able to get early diagnosis without having a very expensive PET scan.
Screen cuts to Rosita Shishegar.
RS: It is absolutely vital to understand the brain and these diseases using the data sets that we have, because it can give us details about us human beings that we were not able to see and understand before.
Screen cuts to Pierrick Bourgeat.
PB: What is my favourite brain fact? I do like the finding that when we sleep our brain washes itself so basically your neurons starts to get out of the way and then all the fluid removes all the nasty proteins or things that were accumulated during the day. And some people actually think that disruption of sleep could basically lead to an impairment of that cleaning process, which could actually lead to the build-up of those proteins leading to Alzheimer’s. So, this is also a very exciting area of research. So, yes, I’m not a doctor, but people should really get some sleep.
Screen cuts to Filip Rusak.
FR: Brace looks like chestnut, right? I like chestnuts. No chestnuts, how did you, what’s the?
Director’s voice is heard but camera remains on Filip Rusak.
Director: Walnut.
Screen cuts to Filip Rusak.
FR: Walnut, yeah, yeah.
Screen cuts to Rosita Shishegar.
RS: Did you know that human brain has inspired one of the most popular artificial intelligence algorithms? This is called neural network. So next time that you admire artificial intelligence, remember that it has been inspired by human brain structure.

Rosita and two of her colleagues at our Australian e-Health Research Centre talk about the ways they are using artificial intelligence to diagnose and manage Alzheimer’s disease.

A passion for people

As Rosita discovered, a career in science is rewarding. But it is not without its challenges. “Sometimes, when things are hard, you can’t help but wonder, are we actually making a change in the world?” she says.

But spending time with people impacted by neurodegenerative disorders helps to dispel her doubts.

“Interacting with patients and seeing how grateful they are, how hopeful they are that science will make a change in their life, reminds me why we do what we do,” Rosita says.

Changes in Rosita’s personal life have also brought new meaning to her work.

“I lost both of my parents last year. Before this, I never truly understood what it was like to lose a loved one. Now I understand why the patients and their caregivers are so grateful for the work that we do. To lose a loved one is the greatest challenge of all,” Rosita says.

“If we can use digital technologies to find ways to prevent or even delay the onset of disease and give people more time with their families, that’s huge.”

Rosita and her daughter recently attended a conference on Alzheimer’s disease together.

A new era for Alzheimer’s research

Rosita’s family gained a special addition last year: her young daughter. Rosita is enjoying watching her 16-month-old grow and explore.

“My baby girl is just learning about the world. A few days ago, she pointed to an elephant on her mat and called it a dog. She hasn’t seen many elephants, so in her brain, every animal with ears is a dog,” Rosita says.

This sweet moment is etched into Rosita’s mind. It’s adorable, of course, but it’s also a fantastic analogy for the issues that come with using AI technologies.

“AI is the same as any human in how it learns; it can only make predictions based on the data it’s given. The more data we have, the more we can leverage AI technologies,” Rosita says.

Collecting data on Alzheimer’s disease is not only expensive, but hard on patients with Alzheimer’s disease and their caregivers.

“A way to solve this problem is using data we already have. We can bring together existing datasets from Australia and other countries. The issue is that different countries and different cohorts have different protocols for collecting their data,” she says.

That’s where Rosita’s work comes in. She created an algorithm that harmonises and aligns these datasets.

The result? The largest Alzheimer’s disease dataset in the world.

“With enough data, AI can make predictions very accurately. More accurately than humans, even. We can take advantage of the technology to learn more about the onset and progression of disease,” Rosita says.

An international team involving multiple research groups is now using the dataset to investigate Alzheimer’s disease risk factors and how they influence patient outcomes.

Rosita and her colleagues are excited about the future of their life-changing work.

“It’s incredible to understand the brain and these diseases more than we ever have before. I hope we can use this information to make a difference in the lives of Alzheimer’s disease patients and their loved ones,” she says.

Learn how we’re using data to detect and prevent chronic illnesses