Repository logo

Infoscience

  • English
  • French
Log In
Logo EPFL, École polytechnique fédérale de Lausanne

Infoscience

  • English
  • French
Log In
  1. Home
  2. Academic and Research Output
  3. EPFL thesis
  4. Behaviorally Relevant Information Content in Interneuron Activity
 
doctoral thesis

Behaviorally Relevant Information Content in Interneuron Activity

Barzegarkeshteli, Mahsa  
2025

The nematode Caenorhabditis elegans offers a uniquely tractable model for studying the relationships between neuronal activity, connectivity, and behavior. Despite its simplicity, key questions remain about how information flows through its nervous system and how neural activity is transformed to generate behavior. Sensory neurons relay information through firstand second-layer interneurons to command interneurons and motor neurons, forming a largely feedforward network. Along this pathway, neuronal activity transitions from closely mirroring environmental stimuli at the sensory level to correlating strongly with behavior at the motor level. Interneurons, particularly those in the second layer such as RIA, RIB, and RIM, exhibit more complex dynamics, integrating both external sensory inputs and internal signals. The functional role of these interneurons in encoding and transmitting behaviorally relevant information remains incompletely understood. To address this, we employed simultaneous neuronal and behavioral recordings, combined with advanced deep learning tools, to gain deeper insights into how second-layer interneurons encode concurrent and future behaviors during chemotaxis A key technological contribution of this work is the development of Targettrack, a novel analysis pipeline for segmenting and tracking neuronal activity in freely moving animals. Traditional convolutional neural networks (CNNs) for neuron tracking require extensive manual annotations of diverse brain postures, which becomes infeasible for highly deformable brains, such as in C. elegans. Targettrack introduces "targeted augmentation," a method that learns internal brain deformations fromlimited ground-truth annotations and uses this information to generate synthetic annotations for diverse postures. This drastically reduces the need for manual annotation and proofreading while maintaining high accuracy. Implemented as an end-to-end pipeline with a graphical interface, Targettrack enables the segmentation and tracking of neurons as 3D volumes or points acrossmultiple animals and experimental conditions. Using Targettrack, we investigated interneuronal dynamics in freelymoving C. elegans exposed to periodic chemosensory stimuli. We observed increasing coordination among interneurons, with stronger correlations at larger temporal scales, suggesting synchronization mechanisms that may facilitate encoding of sustained behavioral states. RIA and RIB interneurons exhibited particularly strong correlations, with RIA axonal segments reflecting RIB activity, suggesting the potential for functional relationship between these two interneurons. Furthermore, regres sion analyses revealed that RIM and RIB activities collectively predict the worm's current and future locomotion speed, with RIB alone emerging as a stronger predictor across individuals. These findings suggest complementary roles for RIA, RIB, and RIMin encoding sensory and behavioral dynamics.

  • Files
  • Details
  • Metrics
Loading...
Thumbnail Image
Name

EPFL_TH11120.pdf

Type

Main Document

Version

Not Applicable (or Unknown)

Access type

openaccess

License Condition

N/A

Size

49 MB

Format

Adobe PDF

Checksum (MD5)

1312d23b4c51380c0a654a4b1c744aab

Logo EPFL, École polytechnique fédérale de Lausanne
  • Contact
  • infoscience@epfl.ch

  • Follow us on Facebook
  • Follow us on Instagram
  • Follow us on LinkedIn
  • Follow us on X
  • Follow us on Youtube
AccessibilityLegal noticePrivacy policyCookie settingsEnd User AgreementGet helpFeedback

Infoscience is a service managed and provided by the Library and IT Services of EPFL. © EPFL, tous droits réservés