LAUSR

Fast and accurate diagnosis is essential for the triage and management of pneumonia, particularly in the current scenario of a COVID-19 epidemic outbreak, where this pathology is a major symptom. With the objective of providing tools for that purpose, this work assesses the potential of three textural image characterisation methods: radiomics, fractal dimension and the recently developed superpixel-based histon, as biomarkers to be used for training Artificial Intelligence (AI) models in order to detect pneumonia in chest X-ray images. Models generated from three different AI algorithms have been studied: K-Nearest Neighbors, Support Vector Machine and Random Forest. Two different open access image datasets are used in this study. On the first one, a dataset composed of paediatric images, the best generated models achieve ana 83.3% accuracy with 89% sensitivity for radiomics, 89.9% accuracy with 93.6% sensitivity for fractal dimension and 91.3% accuracy with 90.5% sensitivity for superpixels based histon. On the second one, a dataset derived from an image repository developed primarily as a tool for the study of COVID-19 has been used. For this dataset, the best generate models results in a 95.3% accuracy with 99.2% sensitivity for radiomics, 99% accuracy with 100% sensitivity for fractal dimension and 99% accuracy with 98.6% sensitivity for superpixels based histon.