Skupni doktorski študij s KU Leuven
V začetku leta 2021 je bil sklenjen sporazum o skupnem doktorskem študiju med UL in KU Leuven. Seznam predlogov raziskovalnih tem je bil študentom predstavljen v objavljenem oglasu za COGDEC Poletno šolo 2021 in je bil posredovan tudi na nekatere druge fakultete UL (Filozofska fakulteta, Fakulteta za elektrotehniko, Fakulteta za matematiko in fiziko).
Prvi trener iz UL je bil sprejet na skupni doktorski študij aprila 2021, drugi je bil sprejet junija 2021. Julija 2021 se je za skupni doktorat prijavil tudi prvi mladi raziskovalec, udeleženec COGDEC Poletne in Zimske šole 2020. Kot rezultat prejšnjega odličnega sodelovanja, mu je KU Leuven za doktorat oktobra 2021 ponudila polno štipendijo.
Skupni predlogi tem za doktorsko disertacijo
UL je skupaj s KU Leuven pripravila 6 predlogov tem za doktorsko disertacijo. Udeležencem, ki jih zanima skupni doktorski študij (Joint PhD) omogočamo kritje stroškov 10-dnevnega obiska v Laboratoriju za Neuro- in Psihofiziologijo na KU Leuven in možnost dogovora o skupnem raziskovalnem delu.
Pozor:
V okviru projekta COGDEC, kot rezultat omogočenega 10-dnevnega obiska, niso zagotovljena sredstva za kritje samega doktorskega študija ali kakršne koli druge oblike raziskovalnega sodelovanja. Za dodatne informacije nam pišite na: cogdec@mf.uni-lj.si
Topics:
Effects of confounds on EEG-based biomarkers of mild cognitive decline
Keywords: EEG, mild cognitive impairment, confounds
Abstract:
According to the 2017 report of Global Voice of Dementia around 50 million patients worldwide are diagnosed with dementia of which Alzheimer's disease (AD) is the largest group. As pharmacological treatments of AD is still beyond reach, considerable effort is currently put into early detection with goal to postpone institutionalization. Currently there are few promising EEG signatures that could differentiate between healthy individuals and the ones with mild cognitive impairment (MCI), however it is still not clear how much of this differentiation can be attributed to the individual characteristics of the study participants. The proposed PhD will address this issue by charting the relationships between these candidate EEG signatures and a number of confounding factors that can influence the clinical manifestation of AD. This way, we can advance our stride towards the objective electrophysiological biomarker for early detection of cognitive decline.
Studying cognitive reserve with EEG in patients with mild cognitive impairment
Keywords: mild cognitive impairment, cognitive reserve, EEG
Abstract:
It has been shown that cognitive reserve (CR) can postpone the clinical manifestation of cognitive decline and maintain individual’s normal functioning even in presence of certain degree of brain impairment. However, once individual exhausts his/her cognitive reserve and enters the stage of decompensation, the, patients present with a cognitive decline that in short time leads to dementia. The role of CR in mild cognitive impairment (MCI) and the value of its compensation mechanisms is less clear. The current PhD will investigate the role of CR in evolution of clinical and electrophysiological changes in patients with MCI and assess if the exhaustion of CR happens already in the stage of MCI or it lasts until patients enter the dementia stage.
Using neural networks to uncover hidden relations in cognitive decline-related EEG data
Keywords: DNN, explainable AI
Abstract:
Deep neural networks (DNNs) have witnessed modest successes in cognitive neuroscience as they offer only limited capabilities to account for prior knowledge. In recent years, the interest in explainable and interpretable AI has grown tremendously with a plethora of algorithms capable of explaining what the DNN has extracted from the data (“features”). This proposal aims at using first shallow neural networks, and in a later stage deep neural networks, to identify and gain insight into the relevance of extracted EEG features of patients suffering from cognitive decline.
Using functional connectivity brain networks as a biomarker of cognitive impairment
Keywords; Source localization, biomarker, functional connectivity
Abstract:
It has been shown that cognitive impairment coincides with changes in functional brain networks. However, due to restrictions in accuracy and fine-tuning of functional brain networks, the connection to cognitive impairment has only been done with static brain networks. With recent advances in source localization accuracy and sparsity, it has become possible to construct more accurate and finer functional connectivity brain networks. This PhD aims to build on these recent advances to gain insight into dynamic time-varying functional connectivity. In a second stage, these time-varying functional connectivity maps will be used to investigate the link to cognitive impairment.
Using Block Term Tensor Regression with source reconstructed EEG to investigate finger dexterity
Keywords; Source localization, BCI, Block term decomposition
Abstract:
Tensor-based techniques, like Block Term Tensor Regression (BTTR) for Brain Computer Interfaces (BCIs) are believed to better account for the complex structure of brain signals compared to conventional techniques. Source localization from EEG is a notoriously difficult inverse problem but essential to identify the brain regions implicated in information processing at high temporal resolution. The aim of the PhD is to investigate the effect of source reconstruction on the performance of decoding finger movement using BTTR.
EEG-based biomarkers of early mild cognitive impairment
Keywords: MCI, biomarkers, machine learning, signal processing
Abstract:
Faced with an ageing population, many western countries are witnessing an increased risk of their elder population to develop mild cognitive impairment (MCI), which eventually could develop into dementia. At this point, the used techniques to detect MCI are invasive, laborious, expensive or require skilled staffing, all of which renders them unsuitable for preemptive screening. This PhD will focus on machine learning and signal processing techniques, applied to public datasets of MCI and dementia patients and healthy controls, to arrive at an EEG-based biomarker with which the risk of MCI can be assessed.
Skupni članki
Trenerji in ostali raziskovalci iz UL, ki so vključeni v project COGDEC, aktivno sodelujejo z raziskovalci iz MUG in KU Leuven. Eden od trenerjev je sodeloval med raziskovalnim delom in pripravo članka Electrophysiological Proxy of Cognitive Reserve Index, ki je bil julija 2021 objavljen v reviji Frontiers in Human Neuroscience (povezava). Članek je imel v prvih treh mesecih po objavi več kot 900 ogledov.
Drugi trener iz UL in ostali raziskovalci, ki so vključeni v projekt COGDEC, so sodelovali pri raziskovalnem delu in pripravi članka Lay Public View of Neuroscience and Science-based Brain Health Recommendations in Slovenia, ki je bil julija 2021 objavljen v reviji Frontiers in Public Health (povezava). Članek je imel v prvih treh mesecih po objavi več kot 950 ogledov.
Ena od mladih raziskovalk je, v sodelovanju z Elviro Khachatryan (KU Leuven), delala na preglednem članku o predklinični Alzheimerjevi bolezni, slikovnih metodah ter vplivu zaščitnih dejavnikov in dejavnikov tveganja. Avtorji so pregledali 1197 člankov, po štirih pregledih so v končni izbor vključili 193 člankov, od katerih je bilo 33 naknadno izključenih zaradi razhajanj med izsledki. Preostalih 160 člankov je bilo razvrščenih v 9 različnih kategorij in nato povzetih v preglednem članku. Avtorji trenutno pregledujejo in urejajo besedilo ter izboljšujejo jasnost in jedrnatost vsebine članka. Pregledni članek je trenutno v zaključni fazi in bo predvidoma oddan jeseni 2021.
Skupni projekti in prijave na razpise
KU Leuven in UL se s partnerji iz Švedske, Nizozemske in Nemčije prijavljata na razpis HORIZON-EIC-2021-PATHFINDER, s predlogom projekta »DeepCognition: Revolutioising Assessment of Cognitive Reserve and Cognitive Decline with Model-Based Deep Learning Methods« (Challenge driven call, Section III, Tools to measure & stimulate activity in brain tissue). Predlog projekta bo oddan v oktobru 2021.
MUG in UL sta sodelovali na skupni prijavi na razpis Javne agencije za raziskovalno dejavnost Republike Slovenije (ARRS) in Avstrijske znanstvene fundacije (Austrian Science Fund - FWF), s projektom z naslovom "Late consequences of COVID19 on brain function and structure". Vloga je bila vložena marca 2021 in je še v pregledu.
Poleti 2021 sta MUG in UL začela delati na novem skupnem projektu, ki se osredotoča na analizo prenosa magnetizacije v možganih bolnikov z multiplo sklerozo. Predhodni rezultati kažejo, da bo ta projekt osnova za prijavo naslednjega skupnega projekta.
Sodelovanje na drugih mednarodnih projektih
Leta 2021 se je začelo sodelovanje z raziskovalci v okviru projekta “Precision Medicine Interventions in Alzheimer’s Disease” (PMI-AD). PMI-AD je tekoči projekt v UKC Ljubljana in je financiran iz programa EU Joint Programme – Neurodegenerative Disease Research (JPND), največje globalne raziskovalne pobude za spopadanje z izzivi na področju nevrodegenerativnih bolezni. Projekt PMI-AD se ukvarja z Alzheimerjevo boleznijo ter vnetjem in propadom sinaps, ki so povezani s premorbidnimi genetskimi predispozicijami, proteomskimi spremembami v cerebrospinalni tekočini in krvi ter razlikami v magnetnoresonančnih slikah možganov. V projektu PMI-AD bodo predvidoma sodelovali tudi trenerji iz UL, ki so na MUG pridobili nova znanja in tehnike na področju magnetnoresonančnega slikanja v okviru projekta COGDEC.
Maja 2021 je UL začela z delom na projektu H2020 Alliance4Life-Actions (povezava). Člani projekta COGDEC se dogovarjajo o možnostih skupne organizacije izobraževanj, delavnic, okroglih miz in srečanj s predstavniki industrije.