Parkinson’s disease affects over 8.5 million people globally. Early diagnosis remains a challenge, due to overlapping symptoms with other brain movement disorders and the scarcity of diagnostic tools. Edward Chen, an MSc in Applied Computing (MScAC) program student at the University of Toronto, is working on solving this. During his internship at BRAIN-TO, Chen developed an AI-driven magnetic resonance imaging (MRI) analysis method that can automatically identify unique brain biomarkers associated with Parkinson’s disease, marking a step forward in personalized, timely diagnosis.
From Research to Real-World Impact
Chen’s work is set to make a tangible difference in the diagnosis and management of movement disorders. By offering a non-invasive, AI-driven tool for Parkinson’s biomarker detection, clinicians will be able to identify patients earlier and more accurately, enabling timely interventions and improving patient outcomes.
“I am a firm believer that implementing AI technologies in clinical settings worldwide will yield significant benefits, and I feel privileged to contribute to this transformative shift,” says Chen. “My experience at BRAIN-TO has allowed me to collaborate with esteemed colleagues like Dr. Sriranga Kashyap and Dr. Paula Alcaide Leon, whose generous support has been pivotal to my success. I have gained invaluable insights from them and am eager to make even greater contributions to this field.”
The methodology’s flexibility opens doors to broader applications, such as analysing other neurodegenerative disorders like amyotrophic lateral sclerosis (ALS). With radiomics and AI integration, Chen’s research showcases how cutting-edge technology can redefine diagnostics in neurology.
Pioneering Neuroimaging Research for Brain Health
Focusing on the substantia nigra, a midbrain region linked to Parkinson’s disease, Chen utilized advanced imaging techniques to extract biomarkers that can differentiate between people living with Parkinson’s disease and healthy individuals. Toronto Western Hospital provided high-resolution, multi-modal datasets acquired with a 3 tesla MRI (that is, an MRI device with a magnetic field strength of 3 tesla). With these datasets, Chen deployed and fine-tuned a pre-trained neural network model known as SWI-CNN — an ensemble of architectures including 3D U-Net and V-Net — for segmenting the substantia nigra. A three-layer convolutional neural network (CNN) then classified the segmented data, achieving 92% accuracy and an F1-score of 0.93.
To further enhance the model, Chen incorporated radiomics — a method for analyzing features from medical imaging — and began exploring direct training on raw MRI data to streamline the diagnostic workflow. The project’s next steps include developing a clinician-friendly model that reduces processing overhead and adapts seamlessly into real-world healthcare settings.
By optimizing quantum circuits, Wang’s project expands the capabilities of quantum processors, positioning them as an increasingly viable tool within the high-performance computing landscape alongside CPUs, GPUs, and TPUs. This work broadens the potential applications of quantum computing, making it a more practical option for machine learning practitioners and beyond, bringing Xanadu’s vision of accessible, impactful quantum technology closer to reality.
A Collaborative Environment to Bridge Research and Clinical Impact
Located at the University Health Network in Toronto, BRAIN-TO Lab is a multi-disciplinary research group working on cognitive neuroscience and advanced MRI. The lab is located at the Toronto Western Hospital, a top neurosurgical centre, where it applies innovations in neuroimaging directly to patient care. By focusing on movement disorders such as Parkinson’s disease, the lab aims to bridge the gap between fundamental MRI research and clinical applications, with a strong commitment to impacting patient outcomes.
Chen’s research was guided by experienced supervisors at BRAIN-TO Lab and University of Toronto, combining expertise in physics, engineering, and clinical neuroscience. Toronto Western Hospital, with its neurosurgical and movement disorder care facilities, offers a prime environment for this type of applied neuroimaging research, especially in conditions like Parkinson’s disease where early intervention can improve patient outcomes.
In this applied-research internship, Chen worked under the supervision of Dr. Kâmil Uludağ at BRAIN-TO and Professor Lueder Kahrs, from the University of Toronto’s Institute of Biomedical Engineering.
“Our goal isn’t just to advance technology; it is to improve lives by turning research into actionable solutions for patients,” says Uludağ. “While our current focus is Parkinson’s disease, this methodology has the potential to dramatically improve how we approach all neurodegenerative disorders. Seeing our AI model succeed in identifying Parkinson’s biomarkers is incredibly rewarding — it’s a glimpse into what the future of neurology could look like. By integrating radiomics with AI, we are opening doors to a new era of personalized medicine.”
A Step Toward Accessible MRI-Based Diagnostics
The BRAIN-TO team is optimistic that this AI-based model will ease integration into clinical practice, bringing a valuable tool to neurologists and radiologists alike. As this model evolves to operate on raw MRI data, it has the potential to improve the detection process by reducing post-processing time, enhancing clinical decision-making, and delivering earlier, personalized care for patients with Parkinson’s disease and other movement disorders.
This internship project exemplifies how academic-hospital partnerships in AI and neuroimaging can drive innovations that are both highly technical and deeply relevant to patient care. As advancements like these continue, they move us closer to a future where timely, precise brain diagnostics are readily accessible for all.
