Nelson D. Medina
Research Publications Contact

Nelson D. Medina

Neuroscientist · Postdoctoral Scientist

I am a neuroscientist working at the intersection of function, structure, and computation. My research asks how the brain stores and produces learned behavior—and recently, specifically, whether a memory can be read from the wiring diagram of a brain.

I have training in patch-clamp electrophysiology, Hodgkin–Huxley neural modeling, calcium imaging, and extracellular recordings from my PhD work. Now based at the MRC Laboratory of Molecular Biology in Cambridge, UK, in the lab of Dr. Joergen Kornfeld, I use serial-section electron microscopy to reconstruct neural circuits at synapse resolution.

Connectomes alone may be insufficient to explain behavior—we also need to understand the biophysical rules that bring circuits to life. The zebra finch song system, a compact circuit driving a single learned behavior, is an ideal model to bridge this gap.

Research

Found that rebound excitation of premotor neurons in zebra finch HVC correlates with temporal features of learned song. Built Hodgkin-Huxley network models linking intrinsic neuronal properties to sequence detection and song timing.

Electrophysiology Modeling Behavior
eGFP fluorescence patch clamp recording eGFP fluorescence patch clamp channel Zebra finch in recording cage

Analyzed dendritic synaptic innervation patterns in the MICrONS dataset (1 mm³ of mouse V1). Developed a simplified wafer-based method for direct collection of thousands of ultrathin brain sections onto silicon wafers.

Volume EM Connectomics Methods
Brain sections on silicon wafer Brain sections on silicon wafer Section collection process

No study has linked an individual memory to a synaptic-level wiring diagram. My work in the Kornfeld Lab aims to change that by reconstructing the zebra finch premotor–motor pathway (HVC→RA) at synapse resolution, paired with in vivo calcium imaging from the same animals during singing. The central question: can a learned memory—a bird’s unique song—be read directly from a connectome?

Using a hybrid-resolution strategy on a multibeam SEM, the premotor nucleus HVC and motor nucleus RA are imaged at 10 nm isotropic resolution, with connecting axon tracts at 15 nm. Functional recordings anchor the structural data: two-photon calcium imaging captures HVC activity during natural singing, and co-registration maps those neurons into the EM volume. Automated segmentation, synapse detection, and biophysical modelling then test whether connectivity alone can reproduce the temporal and spectral structure of vocal output.

Multibeam SEM Calcium Imaging Connectomics Song System
Connectome project workflow: sample prep, staining, embedding, sectioning, imaging, reconstruction, analysis

Tools

MOSS

Microscopy Oriented Segmentation with Supervision

Interactive PyQt6 tool for microscopy image segmentation with real-time U-Net model training.

Python PyTorch U-Net
View on GitHub

Background segmentations each produced with just minutes of simultaneous annotation and training:

Mitochondria Myelinated Axons Nuclei

A.N.E

Author Network Explorer

Chrome extension for visualizing scientific co-author networks, finding collaboration paths, and tracking new publications via OpenAlex.

JavaScript Chrome Extension OpenAlex
View on GitHub

Publications

Contact

MRC Laboratory of Molecular Biology

Francis Crick Avenue, Cambridge CB2 0QH, UK

Kornfeld Lab · Neurobiology

nelsmedina010 [at] gmail [dot] com Google Scholar GitHub Bluesky