AI Paper Atlas

Research map / Label-efficient and robust learning / Test-time adaptation

SPDA: Efficient Online Test-Time Adaptation for Promptable Medical Segmentation

Huixuan Xu, Hu Han, Shiguang Shan, Xilin CHEN

ECCV 2026 · Label-efficient and robust learning › Test-time adaptation · approach: Test-time adaptation of vision-language models · 0 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
add SPDA test-time adaptationSAM2 · 11 modalities, box promptsaverage DSC (%): 67.11 -> 74.18Improved
add SPDA test-time adaptationSAM2 · 11 modalities, 3-click pointsaverage DSC (%): 51.48 -> 61.44Improved
add SPDA test-time adaptationMedSAM2 · 11 modalities, box promptsaverage DSC (%): 83.36 -> 88.05Improved
add SPDA test-time adaptationMedSAM2 · 11 modalities, pointsaverage DSC (%): 78.02 -> 81.11Improved
add SPDASAM · other backbonesaverage DSC (%): 65.48 -> 70.16Improved
add SPDAMedSAM · other backbonesaverage DSC (%): 71.64 -> 72.82Improved
add low-frequency radial-spectrum input perturbation (ablation)SPDA · MedSAM2 box ablation subsetaverage DSC (%): 80.07 -> 83.91Improved
add prompt under-sampling perturbation alone (ablation)SPDA · MedSAM2 box ablation subsetaverage DSC (%): 80.07 -> 81.45Improved
add prompt under-sampling on top of input perturbation (ablation)SPDA · MedSAM2 box ablation subsetaverage DSC (%): 83.91 -> 85.04Improved
replace traditional augmentations with our input perturbation (ablation)SPDA with prompt perturbation · MedSAM2 box ablation subsetaverage DSC (%): 83.78 -> 85.04Improved
use post-encoder adapter instead of LoRA (ablation)SPDA · MedSAM2 box ablation subsetaverage DSC (%) and TFLOPs: 85.11 -> 85.04; 78.90 -> 39.88 TFLOPsNo clear effect
use post-encoder adapter instead of LN optimization (ablation)SPDA · MedSAM2 box ablation subsetaverage DSC (%): 80.85 -> 85.04Improved

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