AI Paper Atlas

Research map / Label-efficient and robust learning / Self-supervised and contrastive

Structure-aware Semantic Discrepancy and Consistency for 3D Medical Image Self-supervised Learning

Tan Pan, Zhaorui Tan, Kaiyu Guo, Dongli Xu, Weidi Xu, Chen Jiang, Xin Guo, Yuan Qi, Yuan Cheng

ICCV 2025 · Label-efficient and robust learning › Self-supervised and contrastive · approach: Self-supervised neuroimaging representations · 9 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 patch-to-patch loss L_p2p to contrastive baseline (ablation)S2DC · BTCVDice: 83.43 -> 83.98Improved
add patch-to-structure loss L_p2s to contrastive baseline (ablation)S2DC · BTCVDice: 83.43 -> 84.02Improved
add L_p2s on top of L_p2p (ablation)S2DC · BTCVDice: 83.98 -> 84.14Improved
add patch-to-patch loss L_p2p to contrastive baseline (ablation)S2DC · AUTOPETDice: 45.72 -> 45.85Improved
add patch-to-structure loss L_p2s to contrastive baseline (ablation)S2DC · AUTOPETDice: 45.72 -> 46.23Improved
add L_p2s on top of L_p2p (ablation)S2DC · AUTOPETDice: 45.85 -> 46.47Improved
S2DC pretraining versus VoCo on BTCVSwin-B · BTCVDice AVG: 83.85 -> 84.14Improved
S2DC pretraining versus training from scratch on small-target AUTOPETSwin-UNETR · AUTOPETDice: 34.49 -> 46.47Improved
pretrain with S2DC for CT-to-CT classification on CC-CCII versus scratchSwin-B · CC-CCIIaccuracy: 88.04 -> 95.34Improved
pretrain with S2DC versus scratch for ADNI classificationSwin-B · ADNI-clsaccuracy: 56.17 -> 56.72No clear effect
pretrain on other SSL methods versus scratch in T1->T2 translationSwin-UNETR · BraTs23 T1->T2NMSE: 0.1348 -> 0.1640Hurt
S2DC pretraining versus scratch in T1->T2 translationSwin-UNETR · BraTs23 T1->T2NMSE: 0.1348 -> 0.1161Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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