Best Place To BuildWorld’s largest fetal-brain mapping dataset is being built here in India! | Dr Richa Verma on S2E10
At a glance
WHAT IT’S REALLY ABOUT
Inside IIT Madras’ Dharani atlas: cell-level fetal brain mapping revolution
- Brain mapping is framed as building a “Google Maps”-like atlas of the brain, moving from region-level views to cell-level resolution across the whole organ to create reference maps for health and disease.
- The Brain Centre’s mission is to map the human brain across the lifespan—from prenatal stages to 100 years—while also studying developmental disorders, stroke/ischemia, neurodegeneration, and aging.
- Their flagship open-access release, the Dharani atlas, is described as the largest and most detailed second-trimester developing human brain map, enabling global researchers to explore cellular anatomy and 2D/3D reconstructions.
- Achieving whole-brain, cellular-resolution imaging requires extensive engineering innovation in freezing, slicing, transferring large delicate sections, staining, scanning at 0.5 micron, and building software to manage and navigate the resulting data.
- The project is expensive (about ₹2 crore per brain in consumables) and relies on philanthropic and institutional funding, plus major compute/visualization collaborations (e.g., NVIDIA) to handle petabyte-scale datasets and tools like Neurovoyager.
IDEAS WORTH REMEMBERING
5 ideasCell-level whole-brain maps act as “reference standards” for disease research.
By mapping normal brains at specific ages, researchers can compare diseased or atypical brains against a baseline to localize what changed—regions, cell types, counts, or cortical/gray-matter thickness—before investigating mechanisms.
The core technical leap is closing the “volume vs. resolution” gap.
Clinical MRI provides whole-brain coverage but typically around ~1 mm resolution, while histology provides cellular detail but usually only for small regions; the Brain Centre’s pipeline targets both whole-organ scale and cellular detail.
Whole-brain histology at scale is primarily an engineering and process-control problem.
Freezing a large water-rich organ without cracks or artifacts, slicing ~10–20 micron sections at large area (e.g., ~6"×8"), and reliably transferring them to slides requires specialized equipment, tuned rates, and repeatable protocols.
Open datasets are only useful if they’re navigable, not just downloadable.
With ~10,000 slices per adult brain and multi-modal stains, the viewer/software layer (zooming, browsing series, 2D-to-3D context, modality switching) is essential to make the atlas practical for remote researchers.
Compute and visualization are now as critical as wet lab methods.
Digitizing 10,000 slices at 0.5 micron can produce data in the petabyte range per brain, creating bottlenecks in storage, access speed, analytics, and model training—driving collaborations like NVIDIA and tools like Neurovoyager.
WORDS WORTH SAVING
5 quotesWe are able to go equivalent of that house level within the brain, which is at cell level, but across the whole brain.
— Dr. Richa Verma
That’s currently the largest, most detailed brain map of the developing human brain in second trimester… made freely open-access datasets.
— Dr. Richa Verma
It sounds simple to say imaging of a human brain… but it’s a huge engineering challenge.
— Amrit (Host)
Each brain would cost you… close to two crores per brain… just talking the consumable.
— Dr. Richa Verma
As we generate these atlases, you can overlay the cellular information on the MRI, and then create different models to predict what you are seeing in MRI, what would happen at cell level.
— Dr. Richa Verma
High quality AI-generated summary created from speaker-labeled transcript.
Get more out of YouTube videos.
High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.
Add to Chrome