The project is divided in five points, involving the IGG for the clinical and data collection phase, and DIBRIS and DIMA for the development of software that supports the non-invasive localization of the EZ. To this end, for each patient referred to IGG, (S1) neurophysiologists will record the neurological signal through high density EEG (HDEEG), totally non-invasive. Based on the information collected in the clinical analysis phase, patients will be divided into homogeneous groups.
When required by clinical needs, the patient will be sent to Niguarda Hospital (Milan) for the implantation of SEEG electrodes. The recordings with HDEEG will first be used to identify functionally connected brain regions (S2). Connectivity information can in fact be used as a constraint in the definition of a neural model and in the study of signal propagation; in this way, they constitute a tool to improve the detection of epileptic patterns in the signal, which will be done through the use of innovative methods of signal analysis and machine learning techniques (S3).Identifying and isolating epileptic patterns in the HDEEG trace is in turn essential to be able to localize with good accuracy the neural source that generated them; for this purpose Bayesian inversion techniques will be used (S4). The reliability of EZ localization will be tested against SEEG data.
Finally, the project will transfer the developed tools to the clinical routine (S5), following a thorough validation on the data collected during the whole project. The validation will be performed using retrospective data provided by Niguarda, and prospectively through the analysis of data acquired by IGG during the project.
Total funding: 167keur
Funder: Compagnia di San Paolo
Annalisa Barla: Principal Investigator
Gabriele Arnulfo: Co-PI
Alberto Sorrentino: Co-PI
D’Amario, Vanessa, et al. “Multi-task multiple kernel learning reveals relevant frequency bands for critical areas localization in focal epilepsy.” Machine Learning for Healthcare Conference. PMLR, 2018.
D’Amario, Vanessa, et al. “Classification of Epileptic Activity Through Temporal and Spatial Characterization of Intracranial Recordings.” International Meeting on Computational Intelligence Methods for Bioinformatics and Biostatistics. Springer, Cham, 2018.
Arnulfo, Gabriele, et al. “Long-range phase synchronization of high-frequency oscillations in human cortex.” Nature communications 11.1 (2020): 1-15.
Conferences and seminars
Annalisa Barla KEYNOTE SPEAKER | “Python in medicine: signal processing learning and visualization of temporal data” | EuroSciPy 2018 |Trento Keynote speaker
Date: 31 Agosto 2018
Event: EuroSciPy 2018 The EuroSciPy meeting is a cross-disciplinary gathering focused on the use and development of the Python language in scientific research.
Host: Valerio Maggio
Abstract: Python in medicine: signal processing learning and visualization of temporal data. The analysis of high dimensional time series is nowadays of wide interest, not only in many fields of science but also for more practical and technical applications in industry. Python represents a convenient and easy way to: perform in optimal time mathematical operations, integrate signal processing tools import external libraries integrate the whole preprocessing procedure with learning tools visualize nicely/export through Bokeh for third parts which are not familiar to computers In this talk I will describe a joint work to make sense out of electroencephalography data for the identification of the epileptogenic area in focal epilepsy patients.