Optimal Patient Specific Electro Convulsive Therapy Simulator

Electroconvulsive therapy (ECT) is a treatment for cases of severe depression. The therapy involves inducing epileptic seizures in the patient by means of electrical currents applied to the scalp. ECT is typically administered to patients who do not respond to drugs. The risk associated with the treatment is on the same level as general anaesthesia, but patients typically also experience short term cognitive side effects. The induced seizures are generally believed to be important to the therapeutic effect, but there is no consensus of how ECT works. Recent imaging studies of the brain of ECT patients taken before and after a course of ECT treatment show volume changes in the brain after the treatment. The volume changes may be caused by for example growth of new cells, new connections between neurons or new blood vessels. It is not clear that the volume changes are related to the treatment outcome. The seizure can be measured more directly by placing electrodes on the scalp to measure fluctuations in the electrical potential. This is called an electroencephalography (EEG). Neuronal activities create fluctuations in the electrical potential. During a seizure, these fluctuations are noticeably different from the resting state. The EEG only measures the electrical potential on the scalp. As the electrodes are placed some distance from the brain, and there is plenty of bone and tissue in between, it will only tell us the behaviour of a group of neurons, not every single one. During a treatment an EEG is taken on both sides of the head to make sure the seizure involves both hemispheres of the brain. The ECT treatment may have an anticonvulsant effect. This means patients can experience shorter seizures after successive treatments and it that more electrical current is needed to induce a seizure. The anticonvulsive effect is caused by chemical changes in the brain that may also be related to a positive outcome of the treatment. The EEGs are not in general stored digitally but are printed on strips of paper. We have written a computer program for digitising such paper strips. We will use this program to investigate the quality of the paper strips and whether they are suitable to be used in research. If they are good enough, we will use them to determine whether the EEG can be used to determine whether the patient needs more electrical current in future treatments and if this can be related to volume changes in the brain. There are many well known models of the electrical field in neurons. These models describe changes in the electrical field due to ions going through ion channels in the cell membrane. The behaviour of the proteins that constitute the ion channels, and consequently the behaviour of the neurons, depend on the chemical environment in the brain. We investigate whether these comparatively simple models can explain which areas of the brain are most active during an epileptic seizure. An alternative hypothesis is that one has to take the folding of the brain and the orientation of the cells into account. To answer these questions we run big simulations on the supercomputer Saga where we compute the effect neurons has on the electrical field in the brain, and how it changes through the brain and the skull. For us to have confidence in the model, we should compare its predictions to the EEG, and look for the same basic patterns. Our goal is to maximise the efficacy of the treatment and reduce side effects, by taking patient specific characteristics into consideration. Our model will allow us to do that.

Prosjektets deltakere har utvidet sitt kompetanseområde og sitt faglige nettverk gjennom prosjektet, Spesielt har PhD kandidaten i prosjektet utviklet et nytt samarbeid med en faggruppe i Bergen.

Elektrokonvulsiv behandling (ECT) brukes på pasienter med alvorlig depresjon. Metoden går ut på å påføre pasientene elektriske sjokk slik at de får epileptiske anfall. Resultatene er til dels svært gode - ECT er ansett som den mest effektive akutte behandling ved en alvorlig depressiv episode, men kan også ha alvorlige bivirnkinger. Det er derfor viktig å tilpasse strømpulsen til hver enkelt pasient. I prosjektet skal det utvikles en numerisk simulator for individualisert simulering av elektrosjokk i hjernen. Dette bygger på en presis geometrisk representasjon som baseres på segmenterte volumetriske bilder (T1 vektet magnetiske resonansbilder kombinert med diffusjons-tensoropptak for å inkludere anisotropi i nervefiberbaner). Dessuten skal den bygge på oppdaterte modeller av aksjonspotensialet til hjerneceller (med lokale variasjoner) og den skal i utgangspunktet bygge på Bidomene modellen. Modellen skal kalibreres mot reelle data hentet fra pasienter og det skal evalueres om modellen kan gi prediksjoner med tilfredsstillende presisjon. Dersom kalibreringen ikke leder til tilfredsstillende prediksjoner, vil vi arbeide videre med nyere varianter av Bidomene modellen for å søke høyere nøyaktighet. Siden siktemålet med prosjektet er å legge det faglige grunnlaget for et kommersielt produkt, vil effektivitet i beregningen tillegges stor vekt. I de tilsvarende beregningene i hjertet er dette en meget stor utfordring, men veldig tidlige analyser gjennomført av Expert Analytics antyder at regnetid vil være mindre problematisk for simulering av elektrosjokk i hjernen. Det er imidlertid for tidlig å fastslå dette. Resultatene skal presenteres i internasjonalt anerkjente vitenskapelige fora. PhD-kandidaten skal utdannes til å kunne drive selvstendig forsknings- og utviklingsarbeid innen avansert numerisk simulering på det nivået som er nødvendig for å arbeide i Expert Analytics.