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

The current medical standard for setting an oral cancer (OC) diagnosis ishistological examination of a tissue sample from the oral cavity. This processis time consuming and more invasive than an alternative approach of acquiring abrush sample followed by cytological analysis. Skilled cytotechnologists areable to detect changes due to malignancy, however, to introduce this approachinto clinical routine is associated with challenges such as a lack of expertsand labour-intensive work. To design a trustworthy OC detection system thatwould assist cytotechnologists, we are interested in AI-based methods thatreliably can detect cancer given only per-patient labels (minimizing annotationbias), and also provide information on which cells are most relevant for thediagnosis (enabling supervision and understanding). We, therefore, perform acomparison of a conventional single instance learning (SIL) approach and amodern multiple instance learning (MIL) method suitable for OC detection andinterpretation, utilizing three different neural network architectures. Tofacilitate systematic evaluation of the considered approaches, we introduce asynthetic PAP-QMNIST dataset, that serves as a model of OC data, while offeringaccess to per-instance ground truth. Our study indicates that on PAP-QMNIST,the SIL performs better, on average, than the MIL approach. Performance at thebag level on real-world cytological data is similar for both methods, yet thesingle instance approach performs better on average. Visual examination bycytotechnologist indicates that the methods manage to identify cells whichdeviate from normality, including malignant cells as well as those suspiciousfor dysplasia. We share the code as open source athttps://github.com/MIDA-group/OralCancerMILvsSIL

Oral cancer detection and interpretation: Deep multiple instance learning versus conventional deep single instance learning