Using imaging and pattern recognition techniques, Mexican students developed a computational tool to identify patients with lung cancer at an early stage, the National Polytechnic Institute (IPN) reported Sunday.

The system analyses computed tomography images of the thorax, and classifies them to prepare a pre-diagnosis, which can help medical specialists in the preparation of their conclusive diagnoses.

For the digital analysis of images, the young students used pattern recognition techniques (Artificial Neural Networks), with which they classify images according to certain patterns, features and characteristics.

In Mexico, lung cancer is considered the second cause of death in men and fourth in women.

The system was created by Ximena Cortés, Isaac Aguirre and Sergio Martínez Ávila, students of the Higher School of Computing of the IPN.

In their development, the students worked with a bank of images of 1,400 patients, available on the Internet through the site: “The Lung Image Database Consortium Image Collection (LIDC-IDRI)”, from the United States.

With the images they “trained” a series of algorithms capable of recognizing the differences between tomography images of healthy patients and those with lung conditions.

Aguirre explained that they used a part of the images to train the algorithm, another for the evaluation process that allows measuring the performance of the system and another fraction of tomography was to make tests with images different from the training.

Our system achieved 95% accuracy and 95% completeness,” Aguirre said.

Cortés said the interpretation of the image of a chest tomography is not simple because of the amount of information contained, causing wrong diagnoses and hindering making decisions.

Martinez said that this computational tool is not intended in any way to replace clinical analysis and traditional methods, but will prevent healthy patients from undergoing biopsies or unnecessary invasive treatments.

The program is able to identify images that present the slightest indication of cancer, which allows the specialist to order biopsy only to those patients selected by the system,” he said.

According to the National Cancer Institute of the United States, 70% of cancer diagnoses in early clinical conditions have a better response to treatment.


Source: Agencia ID