Browsing by Author "Kemere, Caleb T."

Now showing 1 - 7 of 7
  • Thumbnail Image
    Item
    Automating multi-parameter engineering of protein-based sensors of neural electrical activity
    (2020-08-31) Liu, Zhuohe; Kemere, Caleb T.; St-Pierre, Francois
    Engineered voltage-sensitive fluorescent proteins, termed genetically encoded voltage indicators (GEVIs), are emerging sensors for noninvasive microscopy of neural activity. However, the sensitivity and kinetics of existing GEVIs are often not sufficient for accurately reporting fast voltage dynamics in vivo, and they are not optimal for long-term recording due to low brightness and lack of photostability. A system for rapidly evaluating new variants across all performance characteristics is critically needed to accelerate GEVI development progress. This work reports an automated platform that screens libraries of GEVIs in a high-throughput 96-well plate format. This platform quantifies sensitivity, kinetics, brightness, and photostability to identify promising GEVI candidates. Using this platform, a faster and more sensitive indicator, JEDI-1P, was identified and validated in vitro and in behaving zebrafish. The platform is anticipated to optimize versatile biosensors that excel in deep-tissue imaging and promote understanding of neural computation and neurological diseases.
  • Thumbnail Image
    Item
    Bayesian Methods for Learning Analytics
    (2014-06-30) Waters, Andrew; Baraniuk, Richard G.; Kemere, Caleb T.; Vannucci, Marina
    Learning Analytics (LA) is a broad umbrella term used to describe statistical models and algorithms for understanding the relationship be- tween a set of learners and a set of questions. The end goal of LA is to understand the dynamics of the responses provided by each learner. LA models serve to answer important questions concerning learners and questions, such as which educational concepts a learner understands well, which ones they do not, and how these concepts relate to the individual question. LA models additionally predict future learning outcomes based on learner performance to date. This information can then be used to adapt learning to achieve specific educational goals. In this thesis, we adopt a fully Bayesian approach to LA, which allows us both to have superior flexibility in modeling as well as achieve superior performance over methods based on convex optimization. We first develop novel models and algorithms for LA. We showcase the performance of these methods on both synthetic as well as real-world educational datasets. Second, we apply our LA framework to the problem of collaboration– type detection in educational data sets. Collaboration amongst learners in educational settings is problematic for two reasons. First, such collaboration may be prohibited and considered a form of cheating. Detecting this form of collaboration is essential for maintaining fairness and academic integrity in a course. Finally, collaboration inhibits the ability of LA methods to accurately model learners. We develop several novel techniques for collaboration–type detection where we not only identify collaboration in a statistically principled way, but also classify the type of collaborative behavior.
  • Thumbnail Image
    Item
    Engineering Deep Brain Stimulation as a Treatment for Parkinson's Disease: from Models to Materials
    (2014-04-25) Summerson, Samantha Rose; Aazhang, Behnaam; Kemere, Caleb T.; Baraniuk, Richard G.; Cox, Steven J.; Robinson, Jacob T.
    This thesis analyzes deep brain stimulation (DBS) as a treatment for the motor symptoms of Parkinson's disease (PD) at multiple levels. Although this treatment is currently used on human patients, little is understood about the mechanism of action which allows patients to experience therapeutic benefits. The work here investigates efficacy of DBS in computational and experimental manners in order to enhance the understanding of the effects on neural activity and behavior. First, I examine computational models of the nuclei within the motor circuit of the brain and used these models to test novel electrical stimulation signal designs. I show that irregular spacing of stimulation pulses allows for increased variability in neuronal firing rate responses within the basal ganglia. Also, I develop a model of the stimulation-frequency-dependent nature of antidromic spiking induced in the motor cortex as a result of DBS. Second, I use the hemi-Parkinsonian rat model to demonstrate motor and cognitive behavioral effects of DBS in the globus pallidus internus (GPi). The work validates this animal model for translational research on DBS of the GPi and demonstrates results consistent with reports for DBS of the subthalamic nucleus (STN) in the same model. Additionally I study recorded neural activity in the motor cortex while stimulating the STN in order to characterize the corresponding changes in neural activity. I found that regular and irregular stimulation patterns both decrease Parkinsonian entropic noise in the output layer of the motor cortex, with irregular stimulation having the greatest benefit towards reducing this noise. Third, I consider a new material for its biocompatibility and applicability as a material for stimulating electrodes. In the rat model that I previously validated, I verify that behavioral results using a stimulating electrode made from carbon nanotube fibers (CNTf) match results from previous experiments using standard platinum iridium (PtIr) electrodes. Additionally, it is shown that CNTf electrodes produce lower inflammation, gliosis and damage to the blood brain barrier. Together, all three aspects of the work demonstrate significant contributions to the functionality and engineering of DBS as a neuromodulation therapy for PD.
  • Thumbnail Image
    Item
    GhostiPy: An Efficient Signal Processing and Spectral Analysis Toolbox for Large Data
    (Society for Neuroscience, 2021) Chu, Joshua P.; Kemere, Caleb T.
    Recent technological advances have enabled neural recordings consisting of hundreds to thousands of channels. As the pace of these developments continues to grow rapidly, it is imperative to have fast, flexible tools supporting the analysis of neural data gathered by such large-scale modalities. Here we introduce GhostiPy (general hub of spectral techniques in Python), a Python open source software toolbox implementing various signal processing and spectral analyses including optimal digital filters and time–frequency transforms. GhostiPy prioritizes performance and efficiency by using parallelized, blocked algorithms. As a result, it is able to outperform commercial software in both time and space complexity for high-channel count data and can handle out-of-core computation in a user-friendly manner. Overall, our software suite reduces frequently encountered bottlenecks in the experimental pipeline, and we believe this toolset will enhance both the portability and scalability of neural data analysis.
  • Thumbnail Image
    Item
    Realtime decoding with state space models of neural activity
    (2023-10-06) Chu, Joshua P.; Kemere, Caleb T.
    Decoding algorithms provide a powerful tool for understanding the firing patterns that underlie cognitive processes such as motor control, learning, and recall. When implemented in the context of a real-time system, decoders also make it possible to deliver feedback based on the representational content of ongoing neural activity. That in turn allows experimenters to test hypotheses about the role of that content in driving downstream activity patterns and behaviors. While multiple real-time systems have been developed, they are typically implemented in C++ and are locked to a specific data acquisition system, making them difficult to adapt to new experiments. Here we present a Python software system that implements online clusterless decoding, using state space models in a manner independent of data acquisition systems. The parallelized system processes neural data with temporal resolution of 6 ms and median computational latency < 50 ms for medium- to large-scale (32+ tetrodes) rodent hippocampus recordings without the need for spike sorting. It also executes auxiliary functions such as detecting sharp wave ripple and multiunit population burst events. Compared against state-of-the-art solutions which use compiled programming languages, performance is not significantly impacted. Next, we demonstrate its use in a live rat behavior experiment in which the decoder allowed closed loop neurofeedback based on decoding hippocampal activity. Early results indicate the possibility of strengthening neural representations ("replay") of spatial locations experienced in the real world.
  • Thumbnail Image
    Item
    The Neural Computations of Spatial Memory from Single Cells to Networks
    (2012-09-05) Hedrick, Kathryn; Cox, Steven J.; Knierim, James; Sorensen, Danny C.; Embree, Mark; Kemere, Caleb T.
    Studies of spatial memory provide valuable insight into more general mnemonic functions, for by observing the activity of cells such as place cells, one can follow a subject’s dynamic representation of a changing environment. I investigate how place cells resolve conflicting neuronal input signals by developing computational models that integrate synaptic inputs on two scales. First, I construct reduced models of morphologically accurate neurons that preserve neuronal structure and the spatial specificity of inputs. Second, I use a parallel implementation to examine the dynamics among a network of interconnected place cells. Both models elucidate possible roles for the inputs and mechanisms involved in spatial memory.
  • Thumbnail Image
    Item
    Universal Microbial Diagnostics using Random DNA Probes
    (2014-04-23) Aghazadeh Mohandesi, Amir Ali; Baraniuk, Richard G.; Kemere, Caleb T.; Drezek, Rebekah A.
    The accurate and efficient identification of microbial organism such as viruses and bacteria has mounting importance in the fields of health care, environmental monitoring, and defense. As an example, sepsis from bacterial infection is currently the 11th leading cause of death in the United States. However, current microbial detection strategies are cost-prohibitive, time-consuming and inevitably use unique sensors that are specific to each species to be detected. In this thesis we present a novel microbial sensing platform capable of both detecting the presence and estimating the concentration of microbial organisms in an infectious sample using a small number of random DNA probes. Our Universal Microbial Diagnostics (UMD) platform leverages the theory of sparse signal recovery (compressive sensing) to stably identify the composition of a sample containing several bacteria from a potentially large library of target bacteria. We experimentally validate UMD in vitro using a set of random sloppy molecular beacons to recover pathogenic bacteria without DNA amplification. We also evaluate the average performance of UMD in silico for genus and species level identification of 38 common human pathogens. A particularly promising property of UMD for health care, environmental monitoring, and defense applications is that a fixed set of random measurement probes are universal in the sense that they can characterize novel organisms not present in the target library.