ULTRACEPT Experienced Researcher Dr Vassilis Cutsuridis is a Senior Lecturer in Computer Science, and a member of the Machine Learning research group at the University of Lincoln. He recently attended the GEM Conference 2023 Generative Episodic Memory: Interdisciplinary perspectives from neuroscience, psychology, and philosophy. The conference took place 12th to 14th June 2023 in Bochum, Germany.

This conference is organized and funded by the DFG-funded research group FOR 2812 “Constructing scenarios of the past: A new framework in episodic memory”. Episodic memories are widely regarded as memories of personally experienced events. Early concepts about episodic memory were based on the storage model, according to which experiential content is preserved in memory and later retrieved. However, overwhelming empirical evidence suggests that the content of episodic memory is – at least to a certain degree – constructed in the act of remembering. Even though very few contemporary researchers would oppose this view of episodic memory as a generative process, it has not become the standard paradigm of empirical memory research. This is particularly true for studies of the neural correlates of episodic memory. Further hindering progress are large conceptual differences regarding episodic memory across different fields, such as neuroscience, philosophy, and psychology. This interdisciplinary conference therefore aimed to bring together researchers from all relevant fields to advance the state of the art in the research on generative episodic memory.

Dr Cutsuridis presented his research poster ‘Memory retrieval enhancement in a CA1 microcircuit model of
the hippocampus’ 

Abstract

Memory retrieval is important in how the already stored information can be accessed. Improving it would help in developing strategies for preventing memory loss. We selectively scaled excitatory and inhibitory responses of key CA1 neurons to evaluate memory retrieval as a function of stored patterns, pattern interference, contexts, network size, and engram cells in a computational circuit model of the hippocampus. Model excitatory and inhibitory cells fired at specific phases of a theta oscillation imposed by an external inhibitory signal targeting only inhibitory cells, which inhibited compartments of excitatory cells. Sensory and contextual inputs targeting cell dendrites caused cells to fire. Simulation results showed scaling of excitatory synapses in proximal but not basal dendrites of bistratified cells inhibiting pyramidal cells made retrieval perfect. Scaling of inhibitory synapses in pyramidal cells made retrieval worst. Decreases in the number of memory engram cells improved memory retrieval in a pathway-dependent way. Increases in network size and stored patterns had a minimal effect on memory retrieval. Memory interference had a detrimental effect on memory retrieval, which was reversible as the number of engram cells decreased. Changes in contextual information made memory retrieval worse confirming previous evidence that more familiar context facilitates memory retrieval.

ULTRACEPT researcher Dr Vassilis Cutsuridis is a Senior Lecturer in Computer Science, and a member of the Machine Learning research group at the University of Lincoln. He recently attended the 16th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2023) to present his paper titled ‘Machine Learning Algorithms for Mouse LFP Data Classification in Epilepsy’. The conference took place 16th to the 18th February 2023 in Lisbon, Portugal.

The purpose of BIOSTEC is to bring together researchers and practitioners, including engineers, biologists, health professionals and informatics/computer scientists, interested in both theoretical advances and applications of information systems, artificial intelligence, signal processing, electronics and other engineering tools in knowledge areas related to biology and medicine. BIOSTEC is composed of five co-located conferences, each specialized in a different knowledge area.

Abstract

Successful preictal, interictal and ictal activity discrimination is extremely important for accurate seizure detection and prediction in epileptology. Here, we introduce an algorithmic pipeline applied to local field potentials (LFPs) recorded from layers II/III of the primary somatosensory cortex of young mice for the classification of endogenous (preictal), interictal, and seizure-like (ictal) activity events using time series analysis and machine learning (ML) models. Using the HCTSA time series analysis toolbox, over 4000 features were extracted from the LFPs after applying over 7700 operations. Iterative application of correlation analysis and random-forest-recursive-feature-elimination with cross validation method reduced the dimensionality of the feature space to 22 features and 27 features, in endogenous-to-interictal events discrimination, and interictal-to-ictal events discrimination, respectively. Application of nine ML algorithms on these reduced feature sets showed prei ctal activity can be discriminated from interictal activity by a radial basis function SVM with a 0.9914 Cohen kappa score with just 22 features, whereas interictal and seizure-like (ictal) activities can be discriminated by the same classifier with a 0.9565 Cohen kappa score with just 27 features. Our preliminary results show that ML application in cortical LFP recordings may be a promising research avenue for accurate seizure detection and prediction in focal epilepsy.