My PhD project
Unless you are friend or family, I assume you are less interested in me and personal aspects of the project, and more interested in the research, in which case I suggest you skip down to “the project”.
Neurophysiological Markers for Autism - A MEG study
History
I started my PhD studies in the end of 2016. Originally, I had intended to finish my clinical internship and residency first, and then slowly get involved with a research group, with which I would then do my PhD. However, as I was working at the child psychiatry clinic in Borås and applying for internships, I stumbled upon a PhD position vacancy at the University of Gothenburg. For those who are interested in doing research, I can strongly suggest regularly checking university websites for vacancies as research assistants, or PhD students.
Since I have always been interested in a wide array of topics, I did not have a particular field in mind when looking for a PhD position. An idea that had been growing in my mind for several years, due to my interests in child psychology, neurobiology, and diagnostic radiology, was to somehow combine radiology and psychiatry as a clinician, and then to enter the academic field of psychiatric neuroradiology. To my surprise, the vacancy I found happened to be precisely that: A cross-disciplinary PhD in child psychiatry and neuroradiology / neurophysiology. The fact that the project happened to be focused on autism, the most interesting diagnosis of them all, and one I had been reading and thinking about frequently, was just the icing on the cake.
The PHD
Who?
I’m blessed enough to have four very complementary supervisors.
My main supervisor is Christopher Gillberg, specialist and professor of child psychiatry, the most productive autism researcher in the world, and a household name within the field.
Nouchine Hadjikhani, medical doctor and docent in radiology at Harvard Medical School, who has extensive experience with various neuroimaging methods, including MEG and interpretation of MEG analysis.
Justin Schneiderman, physicist and docent in clinical neurophysiology at Chalmers and the University of Gothenburg, who is developing an entirely new MEG-system and knows the hardware and software of MEG inside and out.
Sebastian Lundström, psychologist and docent in child psychiatry, who is an expert in the behavioural phenotype of autism (having completed his PhD within the same area as I did for my psychology bachelor), the scientific method and statistics (and a general sanity checker for my crazy ideas).
Suffice to say, I am well covered in all areas, from the clinical aspects of autism and how to study behavioural syndromes, through academic writing and methodology, to the nitty gritty of brain recordings, and data analysis and interpretation.
I should also mention two colleagues who have been instrumental for my PhD. Bushra Riaz, a biomedical engineer and now postdoc, who has been an incredibly patient private tutor, and introduced me to the world of programming and data analysis. And Elena Orekhova, a biologist and PhD of psychophysiology, who is an expert on neurophysiology and data analysis.
Where?
I’ve also been blessed with two offices that have very different internal cultures, areas of expertise, and academic methodologies.
My primary office is at the Gillberg Neuropsychiatry Centre in central Gothenburg. It is located just below the Child Neuropsychiatric Clinic, and houses dozens of experts on neuropsychiatry who are both clinicians (child psychiatrists, psychologists, special education professionals, and speech and language pathologists) and academics (ranging from PhD students to senior professors). I also have an office at MedTech West, located at the Sahlgrenska University Hospital, which is a collaborative platform filled with engineers, computer scientists, and medical scientists. For someone who has always had a special relationship with physics and the “hard sciences”, but got deprived of this throughout medical school, being part of this community and the local discussions/meetings/seminars has been amazing. I’m fairly certain that this is the reason for my newfound obsession with programming, which has opened more doors than i can count, and has allowed me to play around with intriguing theoretical/academic ideas in ways that would’ve otherwise been impossible.
The project
What?
The project, which is called “Neurophysiological markers for autism - A magnetoencephalographic study”, is quite a mouthful, so here is a deconstruction of what it means:
Neurophysiology – The study of how the nervous system works
Markers – Specific ways in which brains differ, for example how strongly they respond to a loud noise
Autism – A developmental condition which is characterised by social and behavioural problems
Magnetoencephalography – A method that records the magnetic fields that the brain generates when brain cells send signals to one another. Abbreviated as MEG.
How?
Participants – In essence, we are comparing a group of individuals who have been diagnosed with autism, with a group of, so called, neurotypical individuals with no diagnoses.
Methods – All participants have their brains scanned twice. First using MEG at the Karolinska Institute in Stockholm. This gives information about how signals are being sent throughout the brain. Then they do a brain MRI scan at the Sahlgrenska University Hospital. This gives information about the structure of the brain, and allows us to find the precise location within the brain where the activity from MEG is being recorded.
Experiments – We use three types of experiments: The first is the so-called resting state, where they just stare at a screen without any particular task to do. For the second one, we present them with faces and face-like objects. For the third, we show them fast-moving circles, as a way of testing their visual sensory perception.
Data collected – Besides the functional MEG data, and structural MRI data, we also collect additional data during the MEG scanning such as ECG, accelerometer data from fingers during button presses, breathing pattern through a respiration belt, and finally pupil size and movement, as well as viewing location using eye-tracking. Besides that, the participants fill out questionnaires that test autistic-like traits, anxiety, sensory perception, and emotional perception. Their intelligence is also tested.
Why?
The global aim is to better understand the neurobiology of autism. To achieve that, we have a couple of specific aims.
First, we are interested in how autistics process faces. We have looked specifically at activation in a brain area called the Fusiform Face Area. Perhaps their social difficulties stem from an inability to perceive faces the same way neurotypical individuals do. Preliminary results suggest that this is not the case, and that both groups process faces the same way, but with slight differences in the strength of brain activation.
Second, we are looking at a specific type of brain activity, called oscillations, within a particular frequency range, the gamma range (30-100Hz). This activity is thought to reflect the difference between excitation and inhibition in the brain. Studies have shown that autistics have problems with inhibition, and thus higher levels of activation in the brain. We have published a study (which you can find here: https://doi.org/10.1002/hbm.24469) where we showed that a feature of the gamma oscillations predicted which individuals would have a high sensitivity to visual stimulation.
Third, due to my specific interest in diagnostics and classification, I have also worked on ways to improve classification and diagnostics of autism. One way to do that was by developing a new statistical method for classification based on multiple parameters (so called multivariable classification). Since most such methods require expertise, and are not widely available, I wanted to make a method that is both simple and intuitive. It is so simple it can be used in excel, and the most complicated arithmetic one needs to know is how to take the square of a number, the rest is all addition, subtraction, multiplication, and division. I used the method to classify the participants’ brain patterns, whether they look more autistic or neurotypical, with almost 80% accuracy. You can find the paper here: https://doi.org/10.1002/mpr.1846.
Fourth, to be able to diagnose autism optimally, one must first understand it completely. Currently, no model exists that tries to explain how autism develops. So, one part of my PhD has been to develop such a model, and I have previously written about it here. The model consists of three factors that interact to cause the individual to behave in a way that is autistic. I used data from my project to measure those three factors in all the participants. I, then, used those measurements to classify them. This method resulted in an even higher accuracy, which was closer to 100% (although it is highly promising, it needs to be repeated in a larger sample). You can find the papers here: for neurodevelopmental disorders https://doi.org/10.31234/osf.io/mbeqh; for autism https://doi.org/10.3389/fpsyt.2021.767075.
When?
As I wrote in the beginning, I started my PhD in the end of 2016. I took a short detour, and spent 2017 doing my clinical internship. Soon after that the data collection for the project was completed. Now, in 2019, officially 2 years and 1 month into my PhD I am preparing my half time defence, which is scheduled in about three weeks. During these years I have published one paper as co-author, I have studied other courses in parallel at 100-150% and completed a Bachelor in psychology (I’m currently taking courses in pedagogy, economics, and computer science), I finished my clinical internship and just started a part-time clinical residency in radiology, I have two manuscripts under review, and the data analysis for the final sub-study is complete. Luckily, the project has a ton of data, and I have just as many ideas for things to investigate, so even after this final paper is done, which completes the PhD, there will be more to do. Stay tuned…