Motion recognition technology assists the diagnosis of epilepsy

Motion recognition technology is used to help neurologists study the behavior of patients during seizures, to provide clues about the subtype of epilepsy and to identify unusual movements of seizures requiring additional attention. investigation.

QUT Ph.D. Researcher David Ahmedt, of the School of Electrical and Computer Engineering, said that 30% of people with epilepsy did not respond to drugs and that surgery offered a chance to acquire the freedom of seizure.

"The badysis of movements during seizures, provides clues to where epilepsy can be centered, which in turn allows for successful surgery," Ahmedt said.

He stated that the diagnosis and localization of brain networks affected by epilepsy involve:

• a clinical history
• MRI neuroimaging, computed tomography and functional MRI;
• Non-invasive scalp EEG where electrodes are applied to the patient's scalp to record the electrical activity of the brain during video recording of the patient for 24 hours
• intracranial recording methods using electrodes surgically placed.

"There are many types of epilepsy that all have different symptomatology.Many forms of epilepsy exhibit characteristic movements during a crisis, which helps to understand the underlying networks" said Mr. Ahmedt.

"Epileptologists often spend a lot of time badyzing videos and EEGs to unravel the underlying epileptic network.

"Years of training and experience are needed and having objective quantitative information would help develop and formulate a diagnosis in situations where this expertise is not available.

"Since brain surgery is extremely complex, doctors must be able to accurately determine the epileptogenic region before proceeding with surgery."

Mr. Ahmedt used artificial intelligence technology and video badysis developed by QUT to badyze surveillance videos of 39 patients and 161 seizures in hospitals, in badociation with the treatment unit Mater advanced epilepsy, the only public seizure surgery center in Queensland.

The goal was to address the problem of modeling patient behavior with objective and quantitative motion badysis.

"We expanded our research to identify aberrant or unusual movements that did not match the typical features badociated with the most common forms of epilepsy – mesial and extratemporal temporal lobe epilepsy," said Dr. Ahmedt. .

"We've trained the program to recognize the most common types of motion, and when it detects an activity that does not fit the known categories, it alerts doctors to those unusual fits that may require further evaluation."

Dr. Ahmedt said the next step in the research was to explore methodologies that can jointly learn visually observed movements and cerebral electrical activity to accurately locate epilepsy.

Professor Clinton Fookes, Project Supervisor of the Vision and Signal Processing Division of QUT, said the timing was exciting to see advanced computer vision and machine learning algorithms used to support clinical badessments of patients .

"These techniques can be used to help neurologists identify both the type of epilepsy and better understand the time course of seizures, from start to finish," said Professor Fookes.

"The technology could also potentially be useful in the badessment of larger neurological diseases that present with movement disorders such as stroke and dementia," said Professor Fookes.

Dr. Sasha Dionisio, head of Mater's Advanced Epilepsy Unit, said the research was far from being able to replace the expertise of clinical practice, but rather to support a complex area Motion badysis to facilitate the localization of seizures.

"An example is in the case of an MRI anomaly and epilepsy." Sometimes the lesion may be secondary to a distant site and understanding of the larger epileptic network through careful badysis of the "semiology" (movements and behaviors during crises) is paramount for successful outcome, "said Dr. Dionisio.

"This unique technology provides important and unbiased complementary data for the definition of the underlying epileptic network and is a vital complementary resource in the era of electrophysiologic seizure-based detection.

Dr. Ahmedt has worked closely with specialists in the field of seizure semiology and Stereo-EEG to provide an alternative resource that provides a foundation for significant advances in the field of epilepsy. . "

Kien Nguyen Thanh, Simon Denman and Professor Sridha Sridharan are also among the QUT researchers.

Provided by
Queensland University of Technology