Nadgradnja modela nevronskih mas za simuliranje možganske plastičnosti

Prestor, Adam (author), Demšar, Jure (mentor)

Raziskovanje in dobro razumevanje možganske plastičnosti ima lahko velik vpliv na zdravljenje raznih možganskih obolenj, oziroma na izboljšanje okrevanja po fizioloških poškodbah možganov. Z modelom nevronskih mas je mogoče simulirati obnašanje velikih nevronskih omrežij, kot so možgani, ter simulirati njihovo plastičnost. Ena od pomanjkljivosti tovrstnih modelov je, da so obstoječi modeli še vedno precej nestabilni. V naši nalogi smo nadgrajevali obstoječi model, ga preučili in poskusili dodatno stabilizirati. Pokazali smo, kako ključni parametri modela vplivajo na njegovo delovanje in kako lahko s primerno kalibracijo teh parametrov dobimo biološko relevantne konektome. Za dodatno stabilizacijo smo se osredotočili na stabilizacijo sinhronizacijske komponente plastičnosti. Uporabili smo tri različne pristope -- prilagajanje časovnega obdobja, uporabo kovariance za izračun sinhronizacijskega faktorja in implementacijo nevronske oscilacije. Pokazali smo, da lahko s prilagajanjem časovnega obdobja dodatno stabiliziramo model in ohranimo biološko relevantnost dobljenega konektoma. Ostali dve metodi sicer ponujata dodatno stabilizacijo, a ne ohranjata biološke relevantnosti.

• Language: Slovenian
• Keywords: Nevrogeneza, možganska plastičnost, model nevronskih mas, nevronska oscilacija
• Work type: Master's thesis/paper
• Organization: FRI - Faculty of Computer and Information Science
• Year: 2024
• PID: 20.500.12556/RUL-164766
• Publication date in RUL: 11.11.2024

Secondary language

• Language: English
• Title: Improving a neural mass model that simulates brain plasticity

Abstract

Researching and understanding brain plasticity could have a huge impact on future treatment of various brain diseases and recovery after severe brain trauma. By using neural mass models we are able to simulate behaviour of neural networks we find in human brain. The same models can also be applied to simulate brain plasticity. In our assignment we used a pre-existing model, studied it and tried to improve it through additional stabilisation. We showed how model's key parameters affect its performance and how we can fine tune them to get biologically relevant connectomes. We focused on stabilising the model by exploring and fine-tuning the synchronization component of plasticity. We explored three different approaches -- adjusting the time period, using covariance to calculate synchronisation factor and implementing neural oscillation. We showed that we can achieve higher level of stability with proper adjustments of the time period while also preserving biological relevance of resulting connectomes. The other two methods can offer even higher levels of stability but they unfortunately do not preserve the biological relevance.

• Keywords: Neurogenesis, brain plasticity, neural mass model, neural oscillation