Sepand Soltani

Title: Estimation of the derived input function from dynamic PET/MRI imaging ⋄ Implementation of a Bayesian inference algorithm (MCMC) for partial volume correction to accurately estimate input functions from PET/MRI data.

⋄ Development of a robust automatic artery segmentation algorithm on TOF-MRA images using Python to eliminate the need for manual delineation. ⋄ Acceleration of computation for large datasets via parallel processing and JIT compilation to optimize Python pipelines. ⋄ Model validation using realistic Monte Carlo PET simulations and multiple experimental datasets.

⋄ Integration of a Python interpreter via pybind11 to enable the execution of ML inference models directly on real-time C++ data streams. ⋄ Implementation of real-time digital signal processing (DSP) algorithms for feature extraction for AI analysis and visual representation. ⋄ Development of high-performance custom QML components, optimized for rendering massive datasets at high frequency at 60 FPS without blocking the interface. ⋄ Design of a robust multithreaded data acquisition backend to interface with low-level PCIe hardware drivers.

⋄ Design of a medical image visualization tool with PyQt, VTK, and ITK, integrating features such as colormaps, thresholding, segmentation, and annotation. ⋄ Implementation of a workflow interface in Qt, allowing users to dynamically build and execute image processing pipelines. ⋄ Collaboration with the R&D team to integrate AI algorithms into the pipeline.