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

Deep learning has proven to be an essential tool for medical image analysis.However, the need for accurately labeled input data, often requiring time- andlabor-intensive annotation by experts, is a major limitation to the use of deeplearning. One solution to this challenge is to allow for use of coarse or noisylabels, which could permit more efficient and scalable labeling of images. Inthis work, we develop a lopsided loss function based on entropy regularizationthat assumes the existence of a nontrivial false negative rate in the targetannotations. Starting with a carefully annotated brain metastasis lesiondataset, we simulate data with false negatives by (1) randomly censoring theannotated lesions and (2) systematically censoring the smallest lesions. Thelatter better models true physician error because smaller lesions are harder tonotice than the larger ones. Even with a simulated false negative rate as highas 50%, applying our loss function to randomly censored data preserves maximumsensitivity at 97% of the baseline with uncensored training data, compared tojust 10% for a standard loss function. For the size-based censorship,performance is restored from 17% with the current standard to 88% with ourlopsided bootstrap loss. Our work will enable more efficient scaling of theimage labeling process, in parallel with other approaches on creating moreefficient user interfaces and tools for annotation.

Brain Metastasis Segmentation Network Trained with Robustness to Annotations with Multiple False Negatives