Primerjava rezultatov analize funkcijske magnetne resonance z različnimi programskimi orodji

Demšar, Nina (avtor), Repovš, Grega (mentor), Vovk, Andrej (komentor)

Funkcijska magnetna resonanca (fMR) je metoda slikanja aktivnosti možganov, ki temelji na kompleksni analizi podatkov, zbranih s pomočjo magnetnoresonančnega tomografa. Zahtevno analizo omogočajo programska orodja, kot so AFNI, FSL in SPM. Ker se analiza ponavadi opravi le v enemu izmed teh orodij, se postavi vprašanje, ali bi dobili enake rezultate, če bi podatke analizirali tudi z drugimi orodji. V magistrski nalogi smo primerjali dva fMR eksperimenta, ki smo ju analizirali s tremi različnimi programskimi orodji; po privzetih nastavitvah FSL-ja in SPM-ja, ter po priporočenih korakih za AFNI, ki nima privzetih nastavitev. Naša raziskava je bila del večje študije o hiperglikemiji, mi pa smo uporabili podatke 16 zdravih otrok in mladostnikov, ki so v glavni študiji predstavljali kontrolno skupino. Otrokom smo v študiji prikazali slike zdrave in nezdrave hrane ter slike, nepovezane s tem, mi pa smo želeli opazovati njihov čustven odziv. Pri prvem eksperimentu, kjer smo analizirali nalogo, ki ima z dogodkom povezan eksperimentalni načrt, smo primerjali čustven odziv na slike, ki smo jih klasificirali kot negativni ali nevtralni dražljaj. Pri nobenem od programskih orodij se ni pokazala razlika med eksperimentalnima pogojema, torej so bili rezultati različnih programskih orodij skladni. Pomanjkanje razlik med aktivacijama je smiselno glede na vedenjske rezultate, ki niso pokazali razlik v reakcijskem času med pogojema. Pri drugem eksperimentu, kjer smo analizirali nalogo z blok eksperimentalnim načrtom, smo primerjali pogoja opazovanja slik nezdrave hrane in nevtralnih slik. Tu se je pokazala velika variabilnost rezultatov tako v viških razlik med aktivacijama pod različnima pogojema kot v samih področjih teh razlik. Za rezultate naše raziskave je več različnih razlag, saj je med programskimi orodji veliko razlik v postopku analize. Predvsem smo poudarili pomen različnih pristopov do registracije in normalizacije ter poprave za multiple primerjave. Čeprav ti rezultati postavljajo pod vprašaj zanesljivost metode funkcijske magnetne resonance, na njihovi podlagi še ni primerno delati zaključkov. To je šele začetek eksploracije tega področja in za celostno razumevanje razlik, ki nastajajo, bo potrebnih še veliko drugih primerjav.

• Jezik: Slovenski jezik
• Ključne besede: funkcijska magnetna resonanca
• Vrsta gradiva: Magistrsko delo/naloga
• Tipologija: 2.09 - Magistrsko delo
• Organizacija: PEF - Pedagoška fakulteta
• Leto izida: 2020
• PID: 20.500.12556/RUL-119994
• COBISS.SI-ID: 28100355
• Datum objave v RUL: 21.09.2020

Sekundarni jezik

• Jezik: Angleški jezik
• Naslov: Results of functional magnetic resonance imaging analysis with different software tools - a comparison

Izvleček

Functional magnetic resonance imaging (fMRI) is a method of capturing brain activity, which is based on a complex analysis of data that we get with a magnetic resonance machine. The demanding analysis is done in a software package such as AFNI, FSL and SPM. Since the analysis is usually done in only one of these software packages, we start to wonder if we would get the same results, if they were analyzed with different packages. In this Master's thesis, we compared two fMRI experiments which we analyzed in three different ways, depending on the software package; using the default settings of FSL and SPM and the suggested steps for AFNI, which does not have default settings. Our research was a part of a bigger study on hypoglycemia and we used the data of 16 healthy children and adolescents, that were the control group in the main study. The children were presented pictures of healthy and unhealthy food, as well as unrelated images. We then observed their emotional response. In one experiment, where we analyzed a task with an event-related experimental design, we were comparing the negative and neutral conditions of observing images. These did not show any difference in the brain activation between conditions in any of the software packages showing consistency. The lack of activity makes sense when compared to the behavioral results, where there was no difference in reaction times. In the other experiment, where we analyzed a task with a block experimental design, we were comparing the conditions of observing pictures of unhealthy food and neutral images. In this task we saw a large variability of the results of peaks of the difference between the conditions, as well as areas associated with them. There are many explanations for the results of our study, since there were many differences in the analysis pipelines. It is important to point out the different approaches the software packages have towards registration, normalization and multiple comparison corrections. Though these results question the reliability of the fMRI method, it is not appropriate to make conclusions based on them yet. This is only the beginning of the exploration of this area and many other comparisons will have to be done for the full understanding of the differences that happen.

• Ključne besede: functional magnetic resonance