Bryan Chen
Notebook

NotebookEntry 02

Walled off: the foreign-body response

Scrub through twelve weeks after an electrode goes into the brain. Microglia swarm, astrocytes build a scar, and nearby neurons die back. Then swap the stiff silicon shank for a soft, porous gel.

  • Interactive figure
  • Illustrative model
  • 3 sources
  • Made with Claude
  • 26 September 2026
4 weeks

The kill zone: roughly 40% fewer neurons within 100 µm of a stiff silicon shank.

100 µm from the probe 50 µm
  • Microglia, the brain's immune sentinels
  • Astrocytes, the scar builders
  • Neurons
  • Blood and plasma leak
  • The probe, in cross-section
  • 100 µm from the probe

Scar sheath, GFAP-dense

85 µm

Neurons within 100 µm

60%

Microglia

chronic
Fig. 1An illustrative model, not measured data. The trends follow the papers listed below; the soft probe past four weeks is an extrapolation.

What you are looking at

A slice of cortex about 320 µm across, cut straight through an implanted probe. Every implant sets off the same cascade: the insertion tears capillaries, proteins coat the surface within minutes, microglia and macrophages arrive within days, and over weeks reactive astrocytes wall the device off. Neurons close to the shank die back, which is a big part of why recordings fade.

The soft probe shows the bet from the brief: a soft hydrogel with 40 µm pores that tissue can grow into. A softer implant strains the tissue around it less every time the brain moves.

Sources

  1. Polikov, Tresco and Reichert. Response of brain tissue to chronically implanted neural electrodes. J Neurosci Methods, 2005 (review).
  2. Biran, Martin and Tresco. Neuronal cell loss accompanies the brain tissue response to chronically implanted silicon microelectrode arrays. Exp Neurol, 2005. The source of the roughly 40% loss within 100 µm at 2 to 4 weeks.
  3. Dryg et al. Adv Healthc Mater, 2026. Soft hydrogels with 40 µm pores, 4 weeks in rat cortex. doi.org/10.1002/adhm.71522

Claude wrote the model and the drawing. The numbers are illustrative, chosen to follow the trends in these papers, not measured data.