Two-Staged Self-Attention Based Neural Model For Lung Cancer Recognition
Aleksei Samarin, Alexander Savelev, Valentin Malykh
Abstract
Our work is devoted to the neoplasms presence recognition problem in the context of lung computer tomography photographs analysis. This problem is urgent due to the high lung cancer mortality rate. We propose a monochrome lungs tomography photographs analysis engine which could be useful for online medical consultation services. Our approach uses two-staged a self-attention based architecture and demonstrates results of 0.99F <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</inf> score. The presented results are obtained on open dataset of 10052 images.
Topics & Concepts
MonochromeComputer scienceContext (archaeology)Lung cancerArtificial intelligenceArtificial neural networkArchitectureComputed tomographyComputer visionInformation retrievalPattern recognition (psychology)MedicineRadiologyPathologyVisual artsPaleontologyArtBiologyRadiomics and Machine Learning in Medical ImagingLung Cancer Diagnosis and TreatmentCOVID-19 diagnosis using AI