Computer-aided diagnosis system for lung-related pathology
The advances on computational power over the last few years have enabled the usage of high performing methods, such as deep learning, and thus increased the possibility of the development of reliable tools for medical practice. Since these methods require high amounts of data to develop, the NLST images and metadata may prove essential to achieve this goal.
The project aims at the development of machine-learning approaches that provide a second-opinion to clinicians during the diagnosis of lung-related pathologies, namely lung cancer. The NLST data will be used for the development and partial validation of the system.
Aims
1) Development of a system capable of detecting abnormalities on lung tissue, with special focus on the detection of lung nodules in CT images;
2) Development of a system that, given a set of abnormalities, properly predicts necessary patient follow-up;
All members of the CBER-Bioimaging Lab research group (www.bioimglab.inesctec.pt), namely: António Cunha, PhD; Guilherme Aresta, PhD student