Computational neuroscience · Brain imaging · AI

Jignesh Mistry

Research Associate · Sudha Gopalakrishnan Brain Centre, IIT Madras
Visiting researcher · ACE Lab, McGill University

I build computational tools that make the human brain easier to see and analyze, from whole-brain MRI down to single cells under the microscope.

GitHub See my research Publications Contact

  • MLMI 2026 · poster, MICCAI 2026 satellite events
  • 14–24 weeks · fetal brains reconstructed from serial sections
BigBrain cut open, showing the folded cortex

BigBrain, a whole human brain imaged at 20 µm, rendered live in your browser.

About

“Mistry, trying to solve the Mystery of the human brain.”

Illustrated portrait of Jignesh Mistry at sunset by the sea

I'm a computational neuroscientist: a Research Associate at the Sudha Gopalakrishnan Brain Centre, IIT Madras, and currently a visiting researcher in Alan C. Evans's ACE Lab at McGill University, one of the homes of BigBrain. I build computational methods to map the human brain from cells to the whole organ: stacking thousands of serial sections into 3D brains, registering them to MRI, and measuring how cells and layers change as the fetal brain develops.

Before this I worked with Keith St. Lawrence's group at Western University, measuring blood flow in the adult brain with near-infrared light, from outside the skull. Now I look at the brain from the inside, one section at a time. Next, I want to trace neuronal pathways at micron resolution through myelin, and understand how the human brain develops and wires itself up before birth.

  • Histology ↔ MRI registration
  • 3D reconstruction
  • Fetal neurodevelopment
  • Deep learning for imaging
  • Myelin pathways
  • Diffuse optics

Recent

  • Now · Visiting researcher, ACE Lab, McGill University, Montreal
  • Poster · BigBrain Workshop 2026
  • Poster · MAIN 2026, Montreal AI and Neuroscience
  • Poster · MLMI 2026, MICCAI workshop

Research

A brain after a stroke

One human brain followed from the clinic to the microscope: diffusion MRI in life, MRI after death, block-face images as it was cut, and stained sections (Nissl, H&E, immunostains), all registered into one space. From the C-Stroke atlas of the Sudha Gopalakrishnan Brain Centre; my part here was fitting the block-face stack into the MRI.

C-Stroke atlas Preprint (PDF)

The stroke
One sagittal plane through the stroke in every modality: ex vivo MRI, diffusion MRI in life, the block-face photograph, Nissl, H&E, two immunostains, and the Nissl again with the lab's outline of neuronal loss

From BigBrain to small brains

I study the human brain at every size: from BigBrain, a whole adult brain cut into 7,404 sections, down to the smallest brains of all, fetal brains from 14 weeks of gestation.

Human Fetal Neuro­development

How the human brain builds itself before birth, studied in high-resolution histology of fetal brains.

These are the five public DHARANI fetal brains, each reconstructed in 3D from serial Nissl sections (the 14-week brain with VALIS, the others with the mcmadras stacking pipeline) and drawn at true relative size.

  1. 14
  2. 17
  3. 21
  4. 22
  5. 24

14 gestational weeks

Project

whereismyslice

Localizes human brain histology sections in 3D MRI space with cross-modal embeddings, trained on synthetic BigBrain + MRI pairs with no manual labels. Retrieves matching MRI slices and flags misordered, flipped or damaged sections.

  • Python
  • Deep learning
  • Histology
  • Deep Learning For Neuroimaging: self-supervised and cross-modal models trained on synthetic data, such as BigBrain reslices

Nissl histology

T1 MRI

0% close

Drag the blade into the brain until your section matches this one.

An illustration, not the model: your browser already knows the target plane. Both volumes are the same 24-week brain: the MRI in its native shape, placed rigidly, so one blade cuts the same anatomy in each.

Research

Histology ↔ MRI Registration

Automated, annotation-free alignment of human brain histology sections to MRI volumes. Learn the Metric, Not the Deformation: A Small Contrastive Encoder for Cross-Modal Histology–MRI Registration. Accepted as a poster at MLMI 2026, MICCAI 2026 Satellite Events (Springer LNCS).

Poster (PDF)

Block face
Registration results: MRI, raw histology, overlay after registration and checkerboard for BigBrain, Allen and AHEAD

Same method, other brains

MRI, histology, overlay after registration and checkerboard. Rows: BigBrain, Allen, AHEAD (zero-shot).

Registration results: MRI, raw histology, overlay after registration and checkerboard for BigBrain, Allen and AHEAD
One Nissl section of the 24-week brain

The same 24-week brain, one Nissl section.

Research

Ganglionic Eminence

Mapping the ganglionic eminence, the transient fetal structure that gives rise to cortical interneurons, in the developing human brain.

The 24-week brain is cut open along a sagittal plane through the ganglionic eminence, the dense band of cells on the wall of the lateral ventricle.

Rotating 3D reconstruction of the ganglionic eminence inside a translucent fetal brain
A 17-week fetal brain (DHARANI) reconstructed from serial Nissl sections, with the eminence segmented on every section: both eminences (orange) inside the translucent brain.
The ganglionic eminence on the Nissl section

Research

What lives in the ganglionic eminence

Two immunostained neighbouring sections of another fetal brain, overlaid in green and red: elongated migrating cells and their processes streaming out of the eminence.

Green and red immunostained cells streaming out of the ganglionic eminence

Research

Cytoarchitecture

Quantifying gradual transitions between cortical areas using learned image embeddings.

BorderLands

Measuring gradual cytoarchitectonic transitions in embedding space.

  • Cortex
  • Embeddings
Nissl-stained fetal cortical wall from the pial surface to the ventricle
The fetal cortical wall at 22 weeks, pia (left) to ventricle (right): cortical plate, subplate, intermediate zone and germinal zones. Nissl, DHARANI.
Three immunostained markers in red, green and blue across the fetal cortical layers
Three immunostained markers, one per colour, mark different layers of the same wall.

Cell bodies send out axons. In the mature brain, myelin wraps them. Stain the myelin instead of the cells, and the pathways appear.

Research

Myelin-Based Connectomics

Tracing the brain's actual fibre pathways from myelin-stained sections, as ground truth beyond MRI tractography.

Streamlines traced through the white matter of one myelin-stained section of a human brain. Colour is the fibre direction in the section plane.

Direction in the section plane: red horizontal, cyan vertical.

Fibre streamlines across a whole coronal section, coloured by direction
Close-up of streamlines around the internal capsule
Close-up: fibres fanning around the internal capsule.

Same section, same brain

0.5 µm per pixel

The same 24-week brain, one Nissl section: from the whole section into the ganglionic eminence and down to its cells.

Cells of the ganglionic eminence at 0.5 micrometres per pixel

Project

Image processing playground

A testbed for classic image-processing algorithms and their effect on pre- and post-processing pipelines.

  • Image processing

Selected publications

  • Mistry J., Ram K., Mulay S., Neelakantan S., Sivaprakasam M. Learn the Metric, Not the Deformation: A Small Contrastive Encoder for Cross-Modal Histology–MRI Registration. Machine Learning in Medical Imaging (MLMI 2026), MICCAI Workshop. Accepted; to appear in Springer LNCS. Poster
  • Shoemaker L.N., Milej D., Mistry J., St. Lawrence K. Using depth-enhanced diffuse correlation spectroscopy and near-infrared spectroscopy to isolate cerebral hemodynamics during transient hypotension. Neurophotonics 10(2), 025013, 2023. DOI
  • Shoemaker L.N., Milej D., Sajid A., Mistry J., St. Lawrence K., Shoemaker J.K. Characterization of cerebral macro- and microvascular hemodynamics during transient hypotension. Journal of Applied Physiology 135(4), 717–725, 2023. DOI
  • Milej D., Shoemaker L.N., Mistry J., St. Lawrence K. Importance of depth-enhanced optical methods for measuring cerebral hemodynamics during transient hypotension. European Conferences on Biomedical Optics (Proc. SPIE 12628), 126280P, 2023. DOI
  • Mistry J., Shoemaker L.N., Milej D., St. Lawrence K. Assessing extracerebral contamination in cerebral blood flow pulsatility measured by diffuse correlation spectroscopy. Optical Tomography and Spectroscopy of Tissue XV (Proc. SPIE 12376), 112–117, 2023. DOI

Full list on Google Scholar.

One pixel of the scan, 0.5 µm across

Contact