Jakub Smid, Bc
Student
Department of Patophysiology, Second Faculty of Medicine, Charles University
Scientific background
During my undergraduate studies, I focused on electronic circuits and programming – from programming single-chip microcomputers in C to programming applications with user interfaces in Java and C# to programming simple artificial intelligence architectures and signal processing in Python. I gained this knowledge by taking elective courses such as Embedded Systems, Circuit Technology, or Vision for Robotics.
As a part of my Bachelor thesis, I worked on speaker’s emotion recognition from voice, during which I was analyzing the audio signal and matching it to one of six emotions using simple neural networks that are used for image classification.
During my Master’s program, I expanded my knowledge in circuits and signal processing by completing Medical Technology, Neuroinformatics, and Computer Vision Methods.
In addition to the above, I studied Space Missions and Space Engineering and now I am working in space research at J.E. Purkyně University in the KOSMOW team, which is involved in mapping and analyzing astronauts’ social relationships and isolation experiments.
Research interests
Currently, I am writing my Master’s thesis titled Automatic segmentation of ischemic lesions in stroke, in which I analyze and develop image processing – specifically algorithms for stroke segmentation from two modalities taken by MRI.
As a part of my work, I am validating state-of-the-art deep learning methods with a teacher based on convolutional neural networks. I experimentally validate the DeepMedic and nnUNet methods and optimize them for test data provided by Second Faculty of Medicine. In addition, I modify the popular neural network architecture U-Net and investigate how different methods of information fusion from modalities affect the overall segmentation accuracy after training the segmentation model.
Selected publications
More information