Researchers develop technology to diagnose pulmonary nodules

12th December 2019

Lung cancer is the world’s deadliest cancer. Even though an early diagnosis has significantly increased the therapeutic success, screening is still a complex process due to the amount of images that need to be analysed. In addition to this, there may be failures in the identification and observation of pulmonary nodules due to human factors. Therefore, these constraints motivated researchers to develop these technologies.

In addition to the three-module prototype, an annotated database of Computed Tomography images was also created, which contains a wide variety of nodules with different sizes, positions, shapes and textures, and the observation map of radiologists during the annotation process of exams.

"INESC TEC and the São João University Hospital developed this new detection technology that can be used in future researches in order to develop more robust systems for the diagnosis of pulmonary nodules that act as a second opinion for radiologists", explains António Cunha, a researcher of INESC TEC and professor at UTAD.

There were already some methods to support the diagnosis in terms of detection, segmentation and classification of pulmonary nodules. However, the technology that was recently developed also suggests the follow-up that a patient should follow.

"Pulmonary nodules present a great variety of shapes and textures, which makes identifying and characterising them a very complex task, especially for small nodules. However, by learning from thousands of examples of nodules, we can significantly improve their detection, which has a significant impact on the diagnosis, follow-up and treatment of a patient in real life," mentions the researcher.

The prototype developed in this project, called LNDetector (System for the Automatic Detection, Segmentation and Classification of Pulmonary Nodules in Computed Tomography Images) and which received a funding of EUR 168,461 from the Foundation for Science and Technology (FCT), has already won an award for best demo at the RECPAD 2019 conference, dedicated to the Portuguese scientific community that studies the areas of pattern recognition, image analysis and processing, computing and related areas.

Isabel Ramos, Director of the Radiology service of CHUSJ, concludes: “The development in the artificial intelligence and 'big data' areas is opening new opportunities in the Health Sector, and the advances in technology that are expected promise new challenges for all those working in this area."

For more information:

Joana Coelho

Communication Service

INESC TEC

FEUP’s Campus

Rua Dr Roberto Frias

4200-465 Porto

Portugal

Tel.: +351 22 209 4297

Mob.: +351 924 444 321

joana.d.coelho@inesctec.pt