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

Gigapixel medical images provide massive data, both morphological texturesand spatial information, to be mined. Due to the large data scale in histology,deep learning methods play an increasingly significant role as featureextractors. Existing solutions heavily rely on convolutional neural networks(CNNs) for global pixel-level analysis, leaving the underlying local geometricstructure such as the interaction between cells in the tumor microenvironmentunexplored. The topological structure in medical images, as proven to beclosely related to tumor evolution, can be well characterized by graphs. Toobtain a more comprehensive representation for downstream oncology tasks, wepropose a fusion framework for enhancing the global image-level representationcaptured by CNNs with the geometry of cell-level spatial information learned bygraph neural networks (GNN). The fusion layer optimizes an integration betweencollaborative features of global images and cell graphs. Two fusion strategieshave been developed: one with MLP which is simple but turns out efficientthrough fine-tuning, and the other with Transformer gains a champion in fusingmultiple networks. We evaluate our fusion strategies on histology datasetscurated from large patient cohorts of colorectal and gastric cancers for threebiomarker prediction tasks. Both two models outperform plain CNNs or GNNs,reaching a consistent AUC improvement of more than 5% on various networkbackbones. The experimental results yield the necessity for combiningimage-level morphological features with cell spatial relations in medical imageanalysis. Codes are available at https://github.com/yiqings/HEGnnEnhanceCnn.

How GNNs Facilitate CNNs in Mining Geometric Information from Large-Scale Medical Images