English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Test-time adaptation (TTA) has increasingly been an important topic toefficiently tackle the cross-domain distribution shift at test time for medicalimages from different institutions. Previous TTA methods have a commonlimitation of using a fixed learning rate for all the test samples. Such apractice would be sub-optimal for TTA, because test data may arrivesequentially therefore the scale of distribution shift would change frequently.To address this problem, we propose a novel dynamic learning rate adjustmentmethod for test-time adaptation, called DLTTA, which dynamically modulates theamount of weights update for each test image to account for the differences intheir distribution shift. Specifically, our DLTTA is equipped with a memorybank based estimation scheme to effectively measure the discrepancy of a giventest sample. Based on this estimated discrepancy, a dynamic learning rateadjustment strategy is then developed to achieve a suitable degree ofadaptation for each test sample. The effectiveness and general applicability ofour DLTTA is extensively demonstrated on three tasks including retinal opticalcoherence tomography (OCT) segmentation, histopathological imageclassification, and prostate 3D MRI segmentation. Our method achieves effectiveand fast test-time adaptation with consistent performance improvement overcurrent state-of-the-art test-time adaptation methods. Code is available at:https://github.com/med-air/DLTTA.

DLTTA: Dynamic Learning Rate for Test-time Adaptation on Cross-domain Medical Images