Analysis and Classification of Lung Tissue in Ultrasound Images for Pneumonia Detection

摘要

Pneumonia is an infection of the lungs caused by virus, bacteria or fungi. It affects mainly children under fiveand can be life-threatening. Diagnosis of pneumonia is usually performed using imaging techniques such as chestradiography, ultrasound, and CT. Several studies have shown that ultrasound is an effective, safe and cost-efficienttechnique for pneumonia detection. However, due to the low signal-to-noise ratio of the images, this techniqueis highly dependant on the experience of the practitioner. This paper proposes an approach for pneumoniadetection from image texture features. We used empirical mode decomposition for feature extraction, principalcomponent analysis for dimensionality reduction and supervised learning methods for classification. Results showthat features of the first mode present large differences between healthy and pneumonia patients according tothe Cohen's d index. Pneumonia detection was possible with a rotation forest model with a mean accuracy of83.33%.