Turning 20 lung nodules into 3D shows radiologist outlines vary 1.49x
MaziyarPanahi · x · 2026-11-20
Maziyar Panahi converted four radiologists' outlines of the same 20 lung nodules into 3D and found size depends heavily on whose outline you take: median 1.49x volume difference across readers. One nodule spans 1,453–1,892 mm³ — by volume 3 radiologists say Lung-RADS 4A and 1 says 4B; by diameter all 4 say 4B; an open model lands in 4A. 5 of 20 nodules split like this, highlighting annotation consistency issues in medical AI evaluation.
Related event: Radiologists' Lung Nodule Segmentations Vary Up to 1.49x in Volume(3 posts)→
More from Research
- Riemann proof released as blog post and 8 zips, no listed author — eigenhector · 2026-11-21
- NeurIPS paper: Selective Disk Bispectrum keeps all image info minus rotation — ninamiolane · 2026-11-21
- Snorkel's binary-gated GRPO lifts Qwen3.8-27B from 31.9% to 40.7% on OpenAI's GDPval — ajratner · 2026-11-21
- Artificial Analysis Launches Search API Benchmark Covering 26 Products Across 14 Providers — ArtificialAnlys · 2026-11-21
- The 250,000-page Riemann proof claim: when computational verification outpaces human comprehension — SydSteyerhart · 2026-11-21
- Genesis Molecular AI launches Ruby potency model, up to 1000x faster, open-sources training data — edunov · 2026-11-21