Improving the detection of malignant lesions in early stages and being able to examine a greater number of tests in less time is the double objective of the pilot project to refine the diagnosis of breast cancer, through artificial intelligence (AI), initiated in the health area of A Coruña and Cee by the General Directorate of Public Health of the Galician Health Service (Sergas). “We started just this past week with the project. At the end of July, we had already collected some initial data to adapt it to our program, and we are starting to implement it now,” explains the doctor. Joaquín Mosquera Osés, radiologist, head of the Breast Unit at the University Hospital Complex of A Coruña (Chuac) and coordinator of screening in that service, who points out that AI programs for diagnosing breast cancer “are already recognized and studied elsewhere,” however, “each of them has its own peculiarity.”
“The idea is to see a little how to implement this specific AI that we are testing in our screening program to get the most out of it.“, points out the head of the Chuac Breast Unit, who details that, within the framework of the initiative launched by Saúde Pública in the Coruña health area, “Around 35,000 mammography studies” will be carried out, using AI-based software to support radiologists in reading these 3D tests. During the four months that the pilot project will last, which is expected to conclude “at the end of the year”, Those 35,000 mammograms from users of the Galician Program for the early detection of breast canker will be evaluated with AI, together with the independent double radiological readingThe results will be compared with those obtained for a similar period in a recent previous year in which only the double reading was done.
“He Galician program for early detection of breast cancer There are three centralised reading points, which is very favourable. These are in A Coruña, Lugo and Pontevedra, and are where mammograms are read. The 35,000 radiological studies selected at random for this pilot project will be distributed equally between these three reading units, so that around 15,000 will be analysed in each one. Once we have all the data, we will decide how the subsequent implementation will be.: in which cases will this AI be used; if it will be used in all of them; if this will mean that perhaps one reading will be done in one of the cases, and double reading in the others… In short, determining how we are going to get the most out of it. From previous experience in many other countries, and in other parts of Spain, it is clear that AI is useful as a support. Not as the only diagnostic weapon, of course, right now, but as a support.. And it is profitable, both at the level of cancer detection, by increasing it, and by reducing the referral, that is, the false positives,” he points out.
“AI programs increase detection more than the impact on false positives”
The radiologist and person in charge of the radiology clarifies this point: Chuac Breast Unitwhich AI programs “increase detection more than impact on false positives” for breast cancer “because, normally,” those “systems have, let’s say, a wake-up call” for much more. “And then the radiologist has to determine whether that is true or not. Because Not everything AI says is true. Above all, among other things, because it does not see the patients’ previous mammograms, and these types of issues,” highlights Dr. Mosquera Osés, before emphasizing that “AI cannot act alone.” “By itself, it does not have the capacity to replace radiologists who read mammograms right now, but it does have a great capacity to help them improve,” affects.
Regarding the timeframes of the pilot project that has just been launched in A Coruña, Dr. Mosquera Osés anticipates: “Ideally, by the end of the year, we will have 35,000 radiological studies passed through the AI program. From there, as we have a background From the previous readings, because everything is very well recorded, we will make a comparison with other situations in which AI was not used. This will help us to draw conclusions, and decide in which circumstances it helps us the most, and where we are going to influence with AI. Or, in cases where AI says that there is nothing, and in which, indeed, we are seeing that this is the case, instead of two radiologists looking at the test with double reading, maybe one will do it… We will see a series of items that will help us implement AI in the most effective way possible.”, advances the radiologist and person in charge of the Chuac Breast Unit.
Subscribe to continue reading
