Prof. Dr. Alpay ÖZCAN
MRI as a Medical Imaging Modality: From MR Compatible Robotics to Modelling Diffusion and MR based AI in Cancer Imaging
About the Seminar:
MRI’s superior soft tissue contrast with its unrestricted slice positioning and orientation is extremely advantageous for real time image guided interventions. However, unlike CT, MR scanner’s gantry is highly restrictive for placing the interventionist near the patient. The prototype of our MR compatible robotic system addressing access restrictions will be described.
In general, for successful interventions as well as clinical decisions, MR needs to accurately determine and locate malignancies before, during and after the interventions.
For this aim, diffusion MR has provided significant advances in attaining accuracy for various pathologies and disorders with the now very popular diffusion tensor imaging (DTI) methodology. In this part of the seminar, starting from the mathematical necessary conditions for obtaining a unique solution in estimating the diffusion tensor, strategies for minimizing the effect of imaging gradients in DTI gradient schemes and subsequently the development of a new model for diffusion MR signal, the so-called Complete Fourier Direct MR (CFD-MR), will be presented. Furthermore, like DTI, MRI susceptibility mapping is currently in the process of establishing a foothold among MR modalities. After the first implementation of quantitative susceptibility mapping (QSM) in Turkey in July 2017, recently obtained preliminary results on neurodegeneration with brain iron accumulation from a cohort with genetic mutations will be presented. Adding new imaging modalities create valuable means for improving accuracy.
Prostate MR is a good example where adding diffusion MR to the established T2 weighted imaging during the last decade improved the distinction of the benign prostatic hyperplasia from aggressive tumors. However, increasing the number of modalities raises the concern of inter- and intra-observer variability. Recently developed Interactive Feature Space Explorer©, which will be shown, resolves this problem by associating multi-dimensional MR tissue features and anatomic locations, and vice versa.
With the information type changes and/or dimensionality augmentation artificial intelligence methods become necessary. Our recent work on determining immunohistochemical subtypes from Breast Imaging Reporting and Data System (BIRADS) as well as predicting molecular groupings of glioblastoma from anisotropy indices using machine learning methods will be presented.
About the Speaker:
Alpay Özcan received the B.S. in Electrical and Electronics Engineering and the B.S. degree in Mathematics with honors from Boğaziçi University, Istanbul, Turkey, the M.Sc. degree with distinction in Electrical Engineering (Control Systems Section) from Imperial College, London, UK, the M.S. degree and the D.Sc. degree in Systems Science and Mathematics from Washington University in St. Louis, USA.
He was a research assistant and a systems administrator at Washington University during his graduate studies where he also served as the co-director of the Robotics and Computation Laboratory. His doctoral thesis investigated the calculation of the feasibility boundary for electrical power systems modeled by differential-algebraic equations. He remained in the department of Systems Science and Mathematics as a research associate working for the DARPA Joint Force Air Component and Commander program on modeling command and control of military operations using sequential game theoretical methods. He then joined the Biomedical MR Laboratory, at Washington University in Saint Louis and was subsequently promoted to research assistant professor. His administrative responsibilities covered the information technology structure of the laboratory.
During his tenure, he collaborated on building of a prototype 7-degrees of freedom MR compatible robotic system, optimizing diffusion tensor imaging acquisitions.
He created a new model of the diffusion MR signal and developed a novel methodology for multimodal MR imaging with applications to prostate cancer while participating in large scale research projects on child brain development and on multiple sclerosis.
He later joined Arlington Innovation Center: Health Research, Virginia Polytechnic Institute and State University where he conducted research for the multinational (US and South Korea) Neuroperformance imaging project supported by the US Army. At the same time, he was a guest researcher at the Molecular Imaging Program at National Cancer Institute working on multimodal prostate imaging using his methodology as well as with Memorial Sloan Kettering Cancer Center focusing on renal cancer imaging.
On January 2016, Dr. Özcan joined Acıbadem Mehmet Ali Aydınlar University in Istanbul as an associate professor. He is the founder and director of the Biomedical Imaging Research and Development Center. While continuing his research on diffusion MR modeling including intravoxel incoherent motion imaging in liver, he is also working on radiogenomics of glioblastoma and of breast cancer using artificial intelligence, neurodegenerative brain iron accumulation with quantitative susceptibility mapping MR, multimodal MRI of Behçet disease, multiple sclerosis, amyotrophic lateral sclerosis as well as theranostic device development. Additionally, he is particularly interested in applying systems theory to fMRI and more broadly in neuroscience.