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 Plus annual

Presents the access of premium content as premium feature

Premium Content

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 Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

This paper proposes a two-stage segmentation model, variable-input baseduncertainty measures and an uncertainty-guided post-processing method forprostate segmentation on 3D magnetic resonance images (MRI). The two-stagemodel was based on 3D dilated U-Nets with the first stage to localize theprostate and the second stage to obtain an accurate segmentation from croppedimages. For data augmentation, we proposed the variable-input method whichcrops the region of interest with additional random variations. Similar toother deep learning models, the proposed model also faced the challenge ofsuboptimal performance in certain testing cases due to varied training andtesting image characteristics. Therefore, it is valuable to evaluate theconfidence and performance of the network using uncertainty measures, which areoften calculated from the probability maps or their standard deviations withmultiple model outputs for the same testing case. However, few studies havequantitatively compared different methods of uncertainty calculation.Furthermore, unlike the commonly used Bayesian dropout during testing, wedeveloped uncertainty measures based on the variable input images at the secondstage and evaluated its performance by calculating the correlation withground-truth-based performance metrics, such as Dice score. For performanceestimation, we predicted Dice scores and Hausdorff distance with the mostcorrelated uncertainty measure. For post-processing, we performed Gaussianfilter on the underperformed slices to improve segmentation quality. UsingPROMISE-12 data, we demonstrated the robustness of the two-stage model andshowed high correlation of the proposed variable-input based uncertaintymeasures with GT-based performance. The uncertainty-guided post-processingmethod significantly improved label smoothness.

Prostate Segmentation from 3D MRI Using a Two-Stage Model and Variable-Input Based Uncertainty Measure