Computerised Application for Lung Cancer Diagnosis Based on Transthoracic Ultrasonography

Lung cancer (LC) is one of the leading causes of cancer-related deaths worldwide, having one of the lowest survival rates. Compared to other prominent methods of medical imaging, ultrasonography (US) is substantially lower in cost, more flexible and does not use harmful ionizing radiation, proving to be a very useful diagnostic method. It offers good accuracy for the peripheral and pleural abutting lung cancers, allowing also percutaneous biopsy guiding with better precision and lesser complications compared to CT guided biopsy. However, transthoracic US is still considered hard to interpret, but there are reports that greyscale histogram was previously used for breast tumour identification. Driven by this, we developed a computerized application based on image processing techniques and neural networks for lung mass classification and cancer detection. This would help physicians to identify patients with LC and choose the next appropriate method of investigation - lung percutaneous biopsy, bronchoscopy, surgical treatment or surveillance. Also, by using this application in an annual transthoracic ultrasonography checkup, the percentages of early detection of lung cancer can be improved.