Portal de Murcia
www.portaldemurcia.com
Researchers of the Department of Engineering and Computer Technology and Pathology of the University of Murcia (UMU) begin a collaboration to apply the advances of artificial intelligence (AI) to the diagnosis of possible cases of prostate cancer, one of the most widespread in the male population.
In the field of information technology, computational intelligence remains constantly booming.
The application of neural networks in computers, inspired by the functioning of the brain, are being applied to be able to recognize and learn certain patterns.
Large companies such as Google integrate them in their services for voice recognition, voice translation, text recognition and, since its inception, have been applied to the identification of images.
Its field of action has continued to expand in recent years, extending to the world of health, where its implementation is studied as a diagnostic imaging tool.
Faced with this fact, the professors of the UMU, José Manuel García Carrasco and Enrique Poblet, have found a space to apply their studies to the public, through the Reina Sofía General University Hospital.
Until this moment, as Dr. Enrique Poblet emphasizes, the current diagnosis in Medicine has been based on the observation of analogical images through a microscope.
Thus, the experts analyze based on microscopic techniques if the patient suffers from cancer.
Until now, the only way to reach an accurate diagnosis was through biopsy, that is, the analysis of a small fragment of damaged tissue.
"The samples are analyzed by standard visual methods, with the parameters we can use to recognize people or distinguish one face from another.When applying the new system we want to gain precision, time and effectiveness through more quantitative methods, because qualitatively we can make mistakes, we can believe that two people are identical, but if we incorporate technical methods, and here comes the computer, recognition will be faster and better, "confirms the pathologist.
To begin this research, approximately 500 biopsies will be analyzed, which, as confirmed by Poblet, "today are the only way to be sure that a person has the disease."
From the digitalization of these analog images the challenge is to process them using this novel technique.
Once digitized, the role of García Carrasco comes into play.
It is time for the computer through an artificial neural network to be able to learn to detect tumors, that is, through deep learning techniques or 'deep learning' is able to recognize the significant characteristics that determine the disease.
Through this technique, related to artificial intelligence, it will not be given to the computer what particularities are distinguished, but "based on introducing a large number of samples", the machine itself achieves a process of abstraction, just as the human, to determine the diagnosis of the disease.
"We let him learn it himself, the computer is not tired, it does not hurt his head and it's much faster, it's true that the learning process is much slower, it can take sometimes a week or fifteen days, but we can get to an accuracy of 90 or 95% "highlights José Manuel García.
Currently in the medical field have achieved results above 80%, and in other types of non-medical images has exceeded the capacity of humans, with a ratio of 95%.
"It is necessary to have a correct and accurate diagnosis first, in this way the treatments are always more effective, that is, to increase the precision in the diagnosis is fundamental." From this innovation it will be possible to even look for personalized treatments, adapted to each person according to The size or type of tumor In our particular case, through this initial collaboration, we are analyzing prostate cancer, with the idea of ​​improving our diagnostic capacity "explains Poblet.
The first objective of this symbiosis is the creation of a research group interested in the diagnosis of the image, united in the effort to continue advancing in this way.
For now, the work, which has started this last year, has begun with the collaboration of the students of both faculties who are going to complete their Final Degree Projects (TFG) through this collaboration.
During the next months you will get the first results that are already polishing and debugging.
Source: Universidad de Murcia
Privacy policy
- Legal notice
- Cookies