Titulo Estágio
On the development and real-time implementation of improved EEG/ECG features for epileptic seizure prediction
Área Tecnológica
Informática Médica
Local do Estágio
Laboratório de Computação Adaptativa do CISUC
Enquadramento
On behalf of the European Project EPILEPSIAE (http://www.epilepsiae.eu) the Adaptive Computation Group-CISUC developed seizure predictors based on several standard features from EEG/ECG, using computational intelligence methods (Artificial Neural Networks (ANN) and Support Vector Machines (SVM)). Until now some success were obtained with the state-of-the-art features, but clinical application needs the improvement of the actual results.
Objetivo
This proposal aims the study of new features based on single- and multi-electrode approaches that should improve the seizure prediction results. The new features should result in good seizure precursors, presenting at the same time low-computational cost that should enable real-time operation. Concerning real-time operation it is expected that the new features should enable seizure prediction in portable architectures such as tablets PCs (Ex. iPAD) and net PCs (Ex. Azus Eee PC). Portability is a major aspect concerning the development of a transportable device that should warm the patient for an incoming seizure in real-time.
Plano de Trabalhos - Semestre 1
1. State-of-the-art survey
• Objective: In the first phase the student should read about the existent features that are commonly applied for seizure prediction.
• Start: September 2011
• End: October 2011
• Expected Result: A report that should describe the state-of-the-art features, and their major benefits, disadvantages and limitations.
2. Study of new features based on both time, frequency and space domains
• Objective: The student should understand the on-going work on the laboratory and participate in the development of new features.
• Start: October 2011
• End: April 2012
• Expected Result: A report describing the new features or improvements on the on-development ones.
Plano de Trabalhos - Semestre 2
2. Study of new features based on both time, frequency and space domains
• Objective: The student should understand the on-going work on the laboratory and participate in the development of new features.
• Start: October 2011
• End: April 2012
• Expected Result: A report describing the new features or improvements on the on-development ones.
3. Test of the new features by using computational intelligence methods, in off-line mode
• Objective: The new features should be classified by using Artificial Neural Networks and Support Vector Machines. This should be the validation about the applicability of the new features for seizure prediction. This validation should be performed in a significant number of patients (the number of patients will be adjust according to the computational complexity).
• Start: March 2012
• End: April 2012
• Expected Result: A document reporting the performance statistics of the new features in a set of selected patients.
4. Implementation on a portable architecture
• Objective: The computation time of the new features should be investigated and optimizations should eventually be required in order to enable real-time seizure prediction in a portable architecture.
• Start: April 2012
• End: May 2012
• Expected Result: A software that should perform seizure prediction in real-time in a portable architecture.
5. Clinical Validation & Thesis writing
• The implementation should be tested in the Unidade de Monotorização em Epilepsia e Sono, at the Hospitais da Universidade de Coimbra. This will enable the clinical validation of the developed software by clinicians. This final task should end with the compilation of the thesis, based on the reports developed during the year and on the obtained results from the clinical test.
• Start: May 2012
• End: July 2012
Condições
The work is supposed to be developed on the Adaptive Computation Group Laboratories-CISUC, under normal logistic conditions. The student will be supported by the supervisors and by two PhD students.
Observações
Depending on the work quality, publications in international conferences may occur.
This work is carried out as part of the European Project Epilepsiae.
Orientador
César Teixeira
cteixei@dei.uc.pt 📩