AI Paper Atlas

Research map / Architectures and efficiency / Neural architecture search

T-AutoML: Automated Machine Learning for Lesion Segmentation Using Transformers in 3D Medical Imaging

Dong Yang, Andriy Myronenko, Xiaosong Wang, Ziyue Xu, Holger R. Roth, Daguang Xu

ICCV 2021 · Architectures and efficiency › Neural architecture search · approach: AutoML pipeline search · 28 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace accuracy-regression predictor with relation-based (pairwise) predictor (ablation)T-AutoML predictor · LiTS validation set (25 configs)rank correlation: -0.322 -> positive correlationImproved
use transformer encoder instead of MLP in the relation predictor (ablation)relation predictor · LiTS validation set (25 configs)rank correlation strength: more than 20% improvement over MLPImproved
search on lesion segmentation instead of liver (organ) segmentation (ablation)T-AutoML search · LiTSdiscriminatory capacity of predictor / transferImproved
apply T-AutoML configuration searched on LiTS to MSD lung without extra searchT-AutoML · MSD Task06 lung tumormean Dice vs nnU-Net: 0.7397 -> 0.7533Improved
T-AutoML searched model versus nnU-Net on LiTS lesionsegmentation pipeline · LiTS 2017 test (lesion)Dice per case: 0.7630 -> 0.7650Improved
T-AutoML versus nnU-Net on MSD lung NSDsegmentation pipeline · MSD Task06 lung tumormean NSD: 0.7602 -> 0.7768Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy