Amir Dehsarvi

I am a machine learning/deep learning scientist, with ample experience working within various research positions at The Ludwig-Maximilians-University Munich Medical Center, The University of York, The University of Aberdeen, and at The Trinity College Dublin, in addition to working in diverse biomedical companies (Chief Technology Officer at ClearSky Medical Diagnostics Ltd., a machine learning and image processing engineer at smartR.ai, and as a Professional Engineer at My Therapy Tools Ltd. – EU Horizon 2020). My research focuses on the application of machine learning/deep learning for the analysis of biomedical (brain imaging, movement, speech, etc.) data for disease diagnosis/target identification and validation (Parkinson’s disease, Alzheimer’s disease, autism, depression, etc.), using different types of large, complex, biomedical datasets (e.g., brain imaging, movement, and speech), in which I lead the development of novel end-to-end analyses. In my current position at the LMU Hospital, I am working with Dr. Franzmeier on developing cutting-edge pre-processing and processing pipelines for the analysis of multimodal brain imaging (MRI, fMRI, PET, DTI, EEG, etc.).

Additionally, I have been working with Prof. Smith, Dr. Moriarty, Dr. Paton, and Dr. Dutta on a project focusing on objective assessment of depression from rsfMRI brain scans from a UK-Biobank cohort utilising white-box machine learning. Further, I have been working with Dr. Waiter and Prof. Basu on the analysis of cognitive and multimodal brain imaging data to investigate the underlying mechanisms of rheumatoid arthritis related fatigue in the brain. Moreover, I have been involved with Dr. de Looze and Prof. Reilly in the analysis of a combination of speech and brain imaging features for the classification of Alzheimer’s disease patients from patients with mild cognitive impairment, and from healthy participants, aiming to identify speech markers and their underlying neural correlates (brain structure and functional connectivity). Furthermore, under the supervision of Prof. Smith, my doctoral thesis examined the classification of resting state fMRI data (timeseries analysis and DCM analysis) and movement data, applying different classification techniques (Cartesian Genetic Programming, Artificial Neural Networks, Support Vector Machines, etc.) on extracted features, including validation and k-fold cross-validation methods as well as data balancing techniques. This research focused on the diagnosis and monitoring of diseases (Parkinson’s disease, autism, etc.), including the classification of Parkinson’s disease patients from healthy controls with an accuracy of 92%. I have also classified open resting state fMRI data for participants following treatment with Modafinil versus a placebo.