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

As a basic task in computer vision, semantic segmentation can providefundamental information for object detection and instance segmentation to helpthe artificial intelligence better understand real world. Since the proposal offully convolutional neural network (FCNN), it has been widely used in semanticsegmentation because of its high accuracy of pixel-wise classification as wellas high precision of localization. In this paper, we apply several famous FCNNto brain tumor segmentation, making comparisons and adjusting networkarchitectures to achieve better performance measured by metrics such asprecision, recall, mean of intersection of union (mIoU) and dice scorecoefficient (DSC). The adjustments to the classic FCNN include adding moreconnections between convolutional layers, enlarging decoders after up samplelayers and changing the way shallower layers' information is reused. Besidesthe structure modification, we also propose a new classifier with ahierarchical dice loss. Inspired by the containing relationship betweenclasses, the loss function converts multiple classification to multiple binaryclassification in order to counteract the negative effect caused by imbalancedata set. Massive experiments have been done on the training set and testingset in order to assess our refined fully convolutional neural networks and newtypes of loss function. Competitive figures prove they are more effective thantheir predecessors.

Brain Tumor Segmentation Based on Refined Fully Convolutional Neural Networks with A Hierarchical Dice Loss