8491 KB160 KB0 KB2023Jazbinšek, VojkoMarhl, UrbanV, 94 f., 31 cmCOBISSID:140313859URN:URN:NBN:SI:doc-06HAFB1TslU. Marhlvisokošolska delabiomagnetismbiomagnetizemdisertacijedissertationsfunctional neuroimagingfunkcionalno nevroslikanjemagnetoencefalografijamagnetoencephalographyMagnetometrioptically pumped magnetometersoptični magnetometriUniverzitetna in visokošolska delaUporaba optičnih magnetometrov v magnetoencefalografiji| doktorska disertacija|Magnetoencephalography (MEG) is a non-invasive technique in neuroscience that determines activity in the brain by measuring magnetic fields near the head. The main drawback of the standard MEG system is the use of liquid helium to cool the superconducting SQUID magnetometers. In the last decade, commercial Optically Pumped Magnetometers (OPM) have appeared on the market, which have slightly lower sensitivity than SQUID magnetometers but do not require cooling. Therefore, they are suitable for magnetic field measurements in MEG. As this new generation of sensors operates near room temperature, they can be placed closer to the individual's head. Theoretical studies show that this reduction in distance significantly increases the signal-to-noise ratio (SNR), which is expected to increase spatial resolution. In practice, OPM sensor systems are not yet as sophisticated as SQUID systems. We are limited by the smaller number of sensors; the sensors need better magnetic shielding and have a slightly worse SNR than SQUID-MEG systems. In our work, we want to verify the performance and usability of an OPM-MEG system compared to a SQUID-MEG system. With both systems, we performed measurements of auditory stimulation on eight subjects. We show that a double-layered magnetically shielded room with active shielding is suitable for measurements with the new OPM sensors. For the characteristic M100 auditory response, we solved the inverse problem by fitting a current dipole. The locations and strengths of the fitted current dipole are very similar for both systems. Comparing two different systems is not trivial. In our work, we present a method that can be used to transform and compare data between different types of MEG systems. The method first reconstructs the source for one MEG system and then calculates the magnetic fields on the other MEG system using a forward model. Using the measurements and computer simulations of auditory evoked fields, we have verified how different reconstruction methods and different components of the OPM sensors affect the transformation results. We found out that the transformation errors are comparable between the two systems, although we have fewer sensors in the OPM-MEG system than in the SQUID-MEG system. Some commercial OPM sensors allow to measure with multiple orthogonal components simultaneously. We performed a numerical simulation for real head geometries of 8 subjects, where we simulated random sources within the cortex. To the simulated magnetic fields, we added two types of noise: spontaneous brain activity noise and ambient noise. We verified the performance of different sensor components of both OPM-MEG and SQUID-MEG systems by solving the inverse problem (current dipole fitting) for the noisy data. We showed that for both MEG systems the normal component has the largest values of magnetic fields, and consequently the smallest localization error. Additionally, we showed that the combination of several orthogonal components of the OPM sensors only makes sense when we have a small OPM sensor count. Current prices for commercial OPM sensors are around a few thousand Euros per sensor. If we want to cover the whole scalp area, like SQUID-MEG systems, we need at least 50 OPM sensors. This requires a significant financial investment. OPM-MEG systems, therefore, do not achieve full coverage in most laboratories. In this thesis, we apply a method that determines the optimal placement of sensors based on a statistical comparison of measurements. We show that the optimal placement of a limited number (15) of OPM sensors can better reconstruct brain activity, compared to the random placementMagnetoencefalografija (MEG) je neinvazivna tehnika v nevroznanosti, ki na podlagi merjenja magnetnega polja v bližini glave določa aktivnost v možganih. Glavna pomanjkljivost standardnega sistema MEG je uporaba tekočega helija za hlajenje superprevodnih magnetometrov SQUID. V zadnjem desetletju so se na trgu pojavili komercialni optični magnetometri (OPM), ki imajo nekoliko manjšo občutljivost kot magnetometri SQUID, vendar ne potrebujejo hlajenja in so prav tako primerni za meritve magnetnega polja v MEG. Ker ta nova generacija senzorjev deluje pri temperaturah bližje sobni temperaturi, jih lahko postavimo bližje glavi. Dosedanje teoretične raziskave kažejo, da ta premik znatno poveča razmerje signal-šum (SNR), zaradi česar pričakujemo večjo prostorsko ločljivost določanja aktivnosti v možganih. Trenutni sistemi MEG s senzorji OPM še niso tako razviti kot standardni sistemi SQUID. Pri meritvah smo pogosto omejeni z manjšim številom senzorjev, senzorji potrebujejo boljšo magnetno zaščito ter imajo malenkost slabši SNR kot senzorji SQUID. V našem delu smo preverili zmogljivost in uporabnost novejšega sistema OPM-MEG v primerjavi s sistemom SQUID-MEG. Z obema smo izvedli meritve zvočne stimulacije na osmih osebah. Pokazali smo, da je dvoslojna magnetno zaščitena soba z aktivno zaščito primerna za meritve z novimi senzorji OPM. Za značilni slušni odziv M100 smo rešili inverzni problem, tako da smo prilagodili tokovni dipol. Lokaciji in amplitudi prilagojenih tokovnih dipolov sta podobni za oba sistema. Primerjava dveh različnih sistemov ni preprosta. V našem delu predstavimo metodo, s katero lahko transformiramo in primerjamo podatke med različnimi sistemi MEG. Metoda najprej rekonstruira izvor za en sistem MEG in nato z uporabo direktnega modela izračuna magnetna polja na drugem sistemu MEG. Na primeru meritev zvočne stimulacije in računalniških simulacij smo preverili, kako različne metode rekonstrukcije ter različne komponente senzorjev OPM vplivajo na rezultate transformacije. Ugotovili smo, da so napake transformacij primerljive med sistemoma, kljub temu da imamo v sistemu OPM-MEG na voljo manj senzorjev kot v sistemu SQUID-MEG. Nekateri komercialni senzorji OPM omogočajo, da lahko hkrati merimo z več ortogonalnimi komponentami. Naredili smo numerično simulacijo za realne geometrije osmih oseb, kjer smo simulirali naključne izvore znotraj korteksa. Simuliranim magnetnim poljem smo dodali dve vrsti šuma: šum spontane možganske aktivnosti ter šum okolice. Preverili smo uspešnost različnih komponent sistemov OPM-MEG in SQUID-MEG, tako da smo za šumeče podatke rešili inverzni problem (prilagajanje tokovnega dipola). Pokazali smo, da ima normalna komponenta pri obeh sistemih MEG največjo vrednost magnetnega polja in posledično tudi najmanjšo lokalizacijsko napako. Uporaba več ortogonalnih komponent senzorjev OPM je smiselna le, kadar imamo majhno število senzorjev. Trenutne cene komercialnih senzorjev OPM znašajo približno nekaj tisoč evrov na senzor. Če želimo podobno kot pri sistemih SQUID-MEG pokriti celotno območje lasišča, potrebujemo vsaj 50 merilnikov OPM, kar zahteva precejšnji finančni vložek. Sistemi OPM-MEG zato v večini laboratorijev ne dosežejo popolne pokritosti. V tej nalogi uporabimo metodo, ki na podlagi velike baze meritev s statistično primerjavo določi optimalno postavitev senzorjev. Pokažemo, da lahko z optimalno postavitvijo omejenega števila (15) merilnikov OPM bolje rekonstruiramo možgansko aktivnost v primerjavi z naključno postavitvijoTEXTvisokošolska delatheses and dissertationsSlovenian National E-content AggregatorNational and University Library of SloveniaUniverza v Mariboru, Fakulteta za naravoslovje in matematiko