Paper Title
Transthoracic Ultrasonography Analysis for Lung Cancer Detection and Diagnosis

Abstract
Lung cancer (LC) was not common before the 1930s, but increased dramatically over the following decades as tobacco smoking increased, becoming one of the main causes of mortality due to neoplasia, both for men and women. Due to late detection, only 14% of the diagnosed patients are able to survive for a longer period of time (5 years). Once lung cancer begins to cause symptoms, it is usually visible after radiological (X ray) and computed tomography (CT) imaging analysis, or bronchoscopy, investigations invasive to the human body which are performed only based on a doctor’s recommendation. An accessible alternative (including bedside exams) with lower costs, no radiation exposure is the transthoracic ultrasonography (US). Nowadays, transthoracic US is underused for lung cancer diagnosis because gray-scale ultrasound images show low contrast and may be hard to interpret by the clinician for an accurate diagnosis even if is proving to be a very useful for the peripheral lung cancers and percutaneous biopsy guiding offering good or even better accuracy, lesser complications compared to the CT guided biopsy. However, driven by the fact that there are few reports that greyscale histogram was previously applied for breast tumors, we aimed to develop a software system for lung cancer detection and diagnosis based on transthoracic ultrasonography using image processing techniques and also neural networks. This application could be of real use for physicians to identify patients with LC and also could improve the percentages of early detection of lung cancer and the life expectancy of patients by making transthoracic ultrasonography part of the annual checkup exam. Keywords- Transthoracic ultrasonography, Tumor detection, Lung cancer, Image processing, Neural networks, Software application.