Projects

Brainhack Zagreb 2018 Projects

[Project name]

[Names of project proposers]

[Short description]

Deep Learning based Automatic Brain Tumor Segmentation

Domagoj Pluščec (FER), Tomislav Lipić (RBI)

Quantitative analysis of brain MRI scans is prerequisite for analysing of many neurological diseases and conditions and it relies on accurate segmentation of structures of interest. Our focus is segmentation of gliomas in pre-operative MRI scans and prediction of patient overall survival from pre-operative scans which are main challenges from BRATS competition. Proposed tasks are evaluation of existing end-to-end segmentation pipelines based on U-Net model and to survey other deep learning based methodologies such as 3D CNN, anisotropic cascading CNN, multi-path CNN.

EEG during solving cognitive problems

doc.dr.sc. Ana Sušac, Andrea Matić

Analyse EEG recorded during solving cognitive problems. Determine the best method for analyzing spatial distribution of spectral density of power in different frequency ranges and implement it. Compare the results obtained for different types of tasks and different groups of respondents.

Classification of hand motor imagery

Ivan Franjić, Luka Martinez

The goal of this project is to try out some feature extraction and classification methods applied to EEG signals, more precisely hand motor imagery signals. We use publicly available datasets. We decided to use BCI Competiton IV dataset 2b where signals were recorded on 9 subjects and there is 1840 signals for each hand.