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

Detecting malignant pulmonary nodules at an early stage can allow medicalinterventions which may increase the survival rate of lung cancer patients.Using computer vision techniques to detect nodules can improve the sensitivityand the speed of interpreting chest CT for lung cancer screening. Many studieshave used CNNs to detect nodule candidates. Though such approaches have beenshown to outperform the conventional image processing based methods regardingthe detection accuracy, CNNs are also known to be limited to generalize onunder-represented samples in the training set and prone to imperceptible noiseperturbations. Such limitations can not be easily addressed by scaling up thedataset or the models. In this work, we propose to add adversarial syntheticnodules and adversarial attack samples to the training data to improve thegeneralization and the robustness of the lung nodule detection systems. Togenerate hard examples of nodules from a differentiable nodule synthesizer, weuse projected gradient descent (PGD) to search the latent code within a boundedneighbourhood that would generate nodules to decrease the detector response. Tomake the network more robust to unanticipated noise perturbations, we use PGDto search for noise patterns that can trigger the network to giveover-confident mistakes. By evaluating on two different benchmark datasetscontaining consensus annotations from three radiologists, we show that theproposed techniques can improve the detection performance on real CT data. Tounderstand the limitations of both the conventional networks and the proposedaugmented networks, we also perform stress-tests on the false positivereduction networks by feeding different types of artificially produced patches.We show that the augmented networks are more robust to both under-representednodules as well as resistant to noise perturbations.

No Surprises: Training Robust Lung Nodule Detection for Low-Dose CT Scans by Augmenting with Adversarial Attacks