Knowledge base: Warsaw University of Technology

Settings and your account

Back

Decision Support System for surgical treatment of Parkinsons disease

Konrad Andrzej Ciecierski

Abstract

Neurosurgery is one of the youngest branches of medicine. It has started to function as separate medical profession in the beginning of XX century. One of the notable pioneers in neurosurgery was Harvey Williams Cushing (1869 – 1939). In one of his most notable quotes, he stated that ’the operative part is the least part of the work’. This short sentence reveals the complexity faced by the neurosurgeons. The brain anatomy was known to some extent already during the renaissance in XV century. Some of Leonardo da Vinci sketches show dissections of various brain areas. Very little was known however about the brain physiology. Almost all the knowledge came from the observation of people with various head injuries. This all has changed with onset of modern neurology and neurosurgery. It quickly become apparent that brain is not an uniform organ and that its various areas are specialized towards specific functions. That is why when a brain surgery is going to be performed, a carefully planned surgical approach is selected. To properly map the brain regions that are to be operated or avoided during surgery most commonly the various neuro–imaging techniques are used. Among those most commonly used are CT and MRI . CT bases on x–rays and provides fast insight into the brain anatomy. However, the images acquired from the CT are of relatively low resolution and are ill suited for imaging of finer details of the brain anatomy. The MRI uses varying strong magnetic fields to produce its scans. It does not use x–rays and as such does not expose the brain to potentially harmful radiation. MRI also is capable of producing images with much grated resolution than the CT. Two features of the MRI helped it to become the favourite neuro–imaging technique. First one is based on MRI scanning technique called EPI [3] and allows to acquire single plane scan of the brain in time between 20 and 100 ms. Such fast scanning ratio allows for live imaging of living brain tissue. On this feature bases technique, called fMRI that identifies brain parts that have increased oxygen consumption – neural activity while patient is performing certain tasks. Thus, correlation between various tasks and brain regions engaged in their processing can be established. While the principles of EPI scanning method are known since 1977, the second MRI appliance has been discovered more recently. This technique called DTI by tracking the movement of water molecules, shows the tracts of the axonal bundles within the brain. This allows one to observe some brain structures in an indirect way. While some brain structures might not be distinguishable using conventional MRI, the axonal bundles leading to and from them are still tractable [4]. In some cases when aid provided by CT and MRI is still not sufficient, the electrophysiological recording can be used. Recording is then done intra–surgically using electrodes that are either placed on surface of the brain of inserted deep into it. Number of electrodes being used, their kind and placement depends on chosen neurosurgical approach and target. For deep brain recordings the micro-electrode recording is used. Another critical neurosurgical appliance connected with electrophysiological recording is electro-stimulation. In that case an electrode can be briefly switched from recording to specific small current generation. This procedure must be performed with caution as not to damage the delicate contacts of the microelectrode. In current most advanced appliances the stimulation and recording can even be performed simultaneously. Stimulation can be used for brain region identification (e.g. primary motor cortex) or as a way of treatment for symptoms of various diseases. In the treatment approach, electrode(s) are usually permanently implanted into specific regions of patients brain. Computational analysis of the signals recorded from a patient’s brain can in many cases identify brain structure from which they were recorded. It is possible, because different brain sub-structures have varied morphology and thus different physiology and electrical activity. Ability to locate various – often deep placed – structures with micro-electrodes provide neurosurgeons with additional information that can be used for precise navigation during brain surgery. This allows brain surgeries – out of necessity often done without general anesthesia – to be performed in shorter time and also decreases the risk of medical complications. One of the diseases that can be treated with micro-electrode recording and stimulation is the Parkinson Disease (PD). During deep brain stimulation (DBS11) treatment for Parkinson Disease, the target of the surgery is a deeply in brain placed structure called the Subthalamic nucleus (STN). This anatomical structure is small (9 x 7 x 4 mm) and often not well visible using CT or simple MRI. Because of that, a multi-electrode micro recording systems can be used intra surgically for precise localization of the target nucleus. This thesis presents autonomic system that analyzes recordings from micro-electrodes and provides localization of the STN nucleus. Base of the system consists of different computational approaches. Some of them focuses on electrical activity of several neurons closest to the electrode's recording tip. Others look for specific characteristics in background noise coming from the tissue surrounding the electrode. On the results of those analyzes a Deep Brain Stimulation classifier is defined. Classifier can be used for identification whether given recording comes from the STN or not. Cross validation of the classifier yields accuracy exceeding 95 percent. Computer software based upon solutions described in this thesis has already been used during surgeries in the environment of the operation theater.
Record ID
WUT393355
Diploma type
Doctor of Philosophy
Author
Title in Polish
System wspomagania decyzji w chirurgicznym leczeniu choroby Parkinsona
Title in English
Decision Support System for surgical treatment of Parkinsons disease
Language
(en) English
Certifying Unit
Faculty of Electronics and Information Technology (FEIT)
Discipline
information science / (technology domain) / (technological sciences)
Status
Finished
Start date
27-03-2012
Defense Date
20-05-2014
Title date
27-05-2014
Supervisor
Internal reviewers
External reviewers
Andrzej Skowron Andrzej Skowron,, Undefined Affiliation
Honored
yes
Pages
164
Keywords in English
Parkinson’s Disease, DBS, STN, DWT, PCA, FFT, LFB, HFB, RMS, filtering, spike detection, spike discrimination, clustering, classification
Abstract in English
Neurosurgery is one of the youngest branches of medicine. It has started to function as separate medical profession in the beginning of XX century. One of the notable pioneers in neurosurgery was Harvey Williams Cushing (1869 – 1939). In one of his most notable quotes, he stated that ’the operative part is the least part of the work’. This short sentence reveals the complexity faced by the neurosurgeons. The brain anatomy was known to some extent already during the renaissance in XV century. Some of Leonardo da Vinci sketches show dissections of various brain areas. Very little was known however about the brain physiology. Almost all the knowledge came from the observation of people with various head injuries. This all has changed with onset of modern neurology and neurosurgery. It quickly become apparent that brain is not an uniform organ and that its various areas are specialized towards specific functions. That is why when a brain surgery is going to be performed, a carefully planned surgical approach is selected. To properly map the brain regions that are to be operated or avoided during surgery most commonly the various neuro–imaging techniques are used. Among those most commonly used are CT and MRI . CT bases on x–rays and provides fast insight into the brain anatomy. However, the images acquired from the CT are of relatively low resolution and are ill suited for imaging of finer details of the brain anatomy. The MRI uses varying strong magnetic fields to produce its scans. It does not use x–rays and as such does not expose the brain to potentially harmful radiation. MRI also is capable of producing images with much grated resolution than the CT. Two features of the MRI helped it to become the favourite neuro–imaging technique. First one is based on MRI scanning technique called EPI [3] and allows to acquire single plane scan of the brain in time between 20 and 100 ms. Such fast scanning ratio allows for live imaging of living brain tissue. On this feature bases technique, called fMRI that identifies brain parts that have increased oxygen consumption – neural activity while patient is performing certain tasks. Thus, correlation between various tasks and brain regions engaged in their processing can be established. While the principles of EPI scanning method are known since 1977, the second MRI appliance has been discovered more recently. This technique called DTI by tracking the movement of water molecules, shows the tracts of the axonal bundles within the brain. This allows one to observe some brain structures in an indirect way. While some brain structures might not be distinguishable using conventional MRI, the axonal bundles leading to and from them are still tractable [4]. In some cases when aid provided by CT and MRI is still not sufficient, the electrophysiological recording can be used. Recording is then done intra–surgically using electrodes that are either placed on surface of the brain of inserted deep into it. Number of electrodes being used, their kind and placement depends on chosen neurosurgical approach and target. For deep brain recordings the micro-electrode recording is used. Another critical neurosurgical appliance connected with electrophysiological recording is electro-stimulation. In that case an electrode can be briefly switched from recording to specific small current generation. This procedure must be performed with caution as not to damage the delicate contacts of the microelectrode. In current most advanced appliances the stimulation and recording can even be performed simultaneously. Stimulation can be used for brain region identification (e.g. primary motor cortex) or as a way of treatment for symptoms of various diseases. In the treatment approach, electrode(s) are usually permanently implanted into specific regions of patients brain. Computational analysis of the signals recorded from a patient’s brain can in many cases identify brain structure from which they were recorded. It is possible, because different brain sub-structures have varied morphology and thus different physiology and electrical activity. Ability to locate various – often deep placed – structures with micro-electrodes provide neurosurgeons with additional information that can be used for precise navigation during brain surgery. This allows brain surgeries – out of necessity often done without general anesthesia – to be performed in shorter time and also decreases the risk of medical complications. One of the diseases that can be treated with micro-electrode recording and stimulation is the Parkinson Disease (PD). During deep brain stimulation (DBS11) treatment for Parkinson Disease, the target of the surgery is a deeply in brain placed structure called the Subthalamic nucleus (STN). This anatomical structure is small (9 x 7 x 4 mm) and often not well visible using CT or simple MRI. Because of that, a multi-electrode micro recording systems can be used intra surgically for precise localization of the target nucleus. This thesis presents autonomic system that analyzes recordings from micro-electrodes and provides localization of the STN nucleus. Base of the system consists of different computational approaches. Some of them focuses on electrical activity of several neurons closest to the electrode's recording tip. Others look for specific characteristics in background noise coming from the tissue surrounding the electrode. On the results of those analyzes a Deep Brain Stimulation classifier is defined. Classifier can be used for identification whether given recording comes from the STN or not. Cross validation of the classifier yields accuracy exceeding 95 percent. Computer software based upon solutions described in this thesis has already been used during surgeries in the environment of the operation theater.
PKT classification
4100
KBN classification
28 Informatyka
EU classification
80-30
Thesis file
Request a WCAG compliant version
Citation count
5

Uniform Resource Identifier
https://repo.pw.edu.pl/info/phd/WUT393355/
URN
urn:pw-repo:WUT393355

Confirmation
Are you sure?
Report incorrect data on this page