🛰️ Daily AI Frontier
‹ back to 2026-08-19

BrainNorm: A Foundation Model that knows Normal via Semantic Atlas Pretraining

Research Medical/Healthcare AI

Ranking

Overall 81
Content 100
Popularity 37

Observed public metrics from 1 member.

Representative image for BrainNorm: A Foundation Model that knows Normal via Semantic Atlas Pretraining

Merged summary

TL;DR - BrainNorm is a foundation model pretrained on roughly 66,000 structural brain MRI scans to model healthy, age-dependent anatomy. Its localized deviation scores generalize across aging and disease tasks and may help identify clinically meaningful neurodegeneration patterns.

  • Contrastive pretraining on healthy cohorts creates a Semantic Atlas Latent space with embeddings for individual brain parcels.
  • The model learns parcel-specific healthy-aging trajectories, enabling age-matched comparisons and localized abnormality scoring.
  • Evaluations span six cohorts and 25 settings, including zero-shot, few-shot, and linear-probe age and disease prediction.
  • Frozen BrainNorm embeddings with linear probing outperformed nine end-to-end-finetuned baselines across the reported classification tasks.

Sources (1)

BrainNorm: A Foundation Model that knows Normal via Semantic Atlas Pretraining

arXiv cs.CV Madhumitha Venkatesh, Shanawaj S Madarkar, Konda Reddy Mopuri 2026-08-18 arXiv:2608.17521
Public signals Semantic Scholar citations 0 · Semantic Scholar influential citations 0
Providers: Hugging Face · N/A OpenAlex · N/A Publisher · N/A Semantic Scholar · Citations 0 · Influential citations 0 X · N/A Fetched 2026-08-19 14:03:40.601811 UTC

TL;DR - BrainNorm is a foundation model pretrained on roughly 66,000 structural brain MRI scans to model healthy, age-dependent anatomy. Its localized deviation scores generalize across aging and disease tasks and may help identify clinically meaningful neurodegeneration patterns.

  • Contrastive pretraining on healthy cohorts creates a Semantic Atlas Latent space with embeddings for individual brain parcels.
  • The model learns parcel-specific healthy-aging trajectories, enabling age-matched comparisons and localized abnormality scoring.
  • Evaluations span six cohorts and 25 settings, including zero-shot, few-shot, and linear-probe age and disease prediction.
  • Frozen BrainNorm embeddings with linear probing outperformed nine end-to-end-finetuned baselines across the reported classification tasks.
item →