05/15/2021 at 12:28 PM CEST
Technological and scientific evolution improves the living conditions of millions of people around the world, but black holes still exist.
One of them is the fact that medicine is unable to determine the origin of two percent of cancers that are diagnosed every year, which causes a really negative future in these patients: the survival of this type of case does not exceed 16 months.
In this type of cancer, to obtain a diagnosis that helps in its treatment, it is possible to carry out complementary tests that help determine the origin by improving the prognosis, but there is a recurring problem that divides the world: money.
That helplessness, logically, extraordinarily affects places and groups with few resources you don’t have access to that extra.
To end this differentiation and in search of diagnoses with a better prognosis, the team of researchers from the Mahmood Laboratory at Brigham and Women’s Hospital in Boston, in Massachusetts (United States), have developed a system based on Artificial Intelligence that uses slides of Routinely acquired histology to accurately find the origin of metastases.
Technological evolution at the service of health. One more example.
It may interest you: What types of cancer will be diagnosed more in 2021
The algorithm that improves the diagnosis of CUPs
It is on cancers of unknown primary origin, known as CUP, that the research led by Faisal Mahmood, PhD, member of the Division of Computational Pathology at Brigham and an assistant professor at Harvard Medical School, has focused on.
Its difficult diagnosis has led to this analysis that has resulted in the alternative of use of an algorithm based on Deep Learning, or deep learning.
Known as TOAD, Tumor Origin Assessment through Deep Learning, is able to identify the primary tumor and the place where it originates.
To get to this point, researchers have trained Artificial Intelligence with a complete imaging model of tumors in more than 22,000 cancer cases.
A routine that has achieved that the TOAD, in just over 6,500 cases with known primary tumors, has been able to correct the diagnosis 83 percent of the time, raising its accuracy to 96% of the cases taking into account its three main predictions.
This was the previous training for the detection of CUP. In that case, the research data show a 100% coincidence between the Artificial Intelligence diagnosis and the one given by the complementary clinical studies that were carried out to find an answer.
A key conclusion since the dissemination of this technology does not entail the same costs as the complementary studies that had to be carried out in these 317 cases analyzed.
“The main predictions of the model can speed up diagnosis and subsequent treatment by reducing the number of ancillary tests that must be ordered, reducing additional tissue sampling and the total time required to diagnose patients, which can be long and stressful.” points out the lead researcher, Faisal Mahmood.
A great first step to improve the chances of diagnosing these cancers of unknown origin. They must continue to train the algorithm and even participate in clinical trials to make learning ever greater.
This is how Deep Learning works. A technique that is based on automatic learning through observation, with images in this case, to carry out tasks in a simple way and gradually improve until perfection is achieved through training.
In other words, in the case of Mahmood’s Artificial Intelligence, the more CUP images it analyzes, the higher its success rate. Of course, there is no number from which an AI will be “perfect & rdquor ;. This situation varies from case to case, but according to the data, they are on the right track.
It may interest you: Do you know the dangers of “third-hand tobacco”?
The value of histology
For the diagnosis of these cancers in which it is very difficult to determine their origin, in the Mahmood system, histology plays a fundamental role.
‘Almost all patients who are diagnosed with cancer have a histology slide. This has been the diagnostic standard for over a hundred years! ”Faisal Mahmood acknowledges.
Histology understood as the branch of biology that studies the composition, structure and characteristics of the organic tissues of living beings and that provides very valuable information.
“Our work provides a way to harness the acquired data and the power of artificial intelligence to improve the diagnosis of these complicated cases that often require extensive diagnostic studies,” insists Mahmood.
The data, the information, the history & mldr; All of this has a fundamental value if it is used correctly. If you know give a context for your analysis.
In the case of cancers of unknown primary origin, it is demonstrated how the correct treatment of the data can help save thousands of lives.