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

Gliomas are the most frequent primary brain tumors in adults. Glioma changedetection aims at finding the relevant parts of the image that change overtime. Although Deep Learning (DL) shows promising performances in similarchange detection tasks, the creation of large annotated datasets represents amajor bottleneck for supervised DL applications in radiology. To overcome this,we propose a combined use of weak labels (imprecise, but fast-to-createannotations) and Transfer Learning (TL). Specifically, we explore inductive TL,where source and target domains are identical, but tasks are different due to alabel shift: our target labels are created manually by three radiologists,whereas our source weak labels are generated automatically from radiologyreports via NLP. We frame knowledge transfer as hyperparameter optimization,thus avoiding heuristic choices that are frequent in related works. Weinvestigate the relationship between model size and TL, comparing alow-capacity VGG with a higher-capacity ResNeXt model. We evaluate our modelson 1693 T2-weighted magnetic resonance imaging difference maps created from 183patients, by classifying them into stable or unstable according to tumorevolution. The weak labels extracted from radiology reports allowed us toincrease dataset size more than 3-fold, and improve VGG classification resultsfrom 75% to 82% AUC. Mixed training from scratch led to higher performance thanfine-tuning or feature extraction. To assess generalizability, we ran inferenceon an open dataset (BraTS-2015: 15 patients, 51 difference maps), reaching upto 76% AUC. Overall, results suggest that medical imaging problems may benefitfrom smaller models and different TL strategies with respect to computer visiondatasets, and that report-generated weak labels are effective in improvingmodel performances. Code, in-house dataset and BraTS labels are released.

Weakly Supervised Learning with Automated Labels from Radiology Reports for Glioma Change Detection