Gesture tracking and neural activity segmentation in head-fixed behaving mice by deep learning methods

The typical approach used by neuroscientists is to study the response of laboratory animals to a stimulus while recording their neural activity at the same time. With the advent of calcium imaging technology, researchers can now study neural activity at sub-cellular resolutions in vivo. Similarly, r...

Descripción completa

Detalles Bibliográficos
Autor: Abbas, Waseem
Tipo de recurso: tesis doctoral
Estado:Versión publicada
Fecha de publicación:2020
País:España
Institución:CBUC, CESCA
Repositorio:TDR. Tesis Doctorales en Red
OAI Identifier:oai:www.tdx.cat:10803/674476
Acceso en línea:http://hdl.handle.net/10803/674476
Access Level:acceso abierto
Palabra clave:neurociència
neurociencia
neuroscience
activitat neuronal
actividad neuronal
neural activity
dades de comportament
datos de comportamiento
behavioral data
xarxa neuronal convolucional tridimensional (3D-CNN)
red neuronal convolucional tridimensional (3D-CNN)
3-dimensional convolutional neural network (3D-CNN)
xarxa de memòria a llarg i curt termini (LSTM)
red de memoria a largo y corto plazo (LSTM)
long-term and short-term memory network (LSTM)
Neurociència
616.8
Descripción
Sumario:The typical approach used by neuroscientists is to study the response of laboratory animals to a stimulus while recording their neural activity at the same time. With the advent of calcium imaging technology, researchers can now study neural activity at sub-cellular resolutions in vivo. Similarly, recording the behaviour of laboratory animals is also becoming more affordable. Although it is now easier to record behavioural and neural data, this data comes with its own set of challenges. The biggest challenge, given the sheer volume of the data, is annotation. A traditional approach is to annotate the data manually, frame by frame. With behavioural data, manual annotation is done by looking at each frame and tracing the animals; with neural data, this is carried out by a trained neuroscientist. In this research, we propose automated tools based on deep learning that can aid in the processing of behavioural and neural data. These tools will help neuroscientists annotate and analyse the data they acquire in an automated and reliable way.