In cancer, time is health. The sooner a tumor is detected, the better: more chances of treatment and cure. For this reason, early detection tests have become an ally to increase survival and stop early tumors as soon as possible. The scientific literature estimates, for example, that population screening by mammography reduces mortality by 20% by breast cancer, and this decline may be even more pronounced if screening tests and outcome analysis are refined. Artificial intelligence (AI) has already crept into this field and, according to a Swedish study published in The Lancet Oncology With 80,000 women, breast cancer screenings that are supported by artificial intelligence systems to read mammograms detect 20% more tumors than those that follow traditional reading methodology with double review by two radiologists. The preliminary results of the research, which is still ongoing, conclude that the use of AI to analyze mammograms is safe and reduces the workload of radiologists by almost half.
In a breast cancer screening, the analysis of the mammogram is usually done, as recommended by the European clinical guidelines, by two independent radiologists. If they do not agree on the reading, they tend to agree on the discrepancies or make the most aggressive decision prevail —between not calling the patient back or referring them for more tests, for example, opting for the second. One study suggests that this double reading technique will detect 0.44 more tumors for every 1,000 people examined than with a single reading. However, the eye of the specialist is not infallible either: the scientific literature estimates that up to 25% of mammographically visible cancers are still not detected at screening and there are already research suggesting that the accuracy of AI may be similar to or even better than that of radiologists.
To see if AI-assisted detection is indeed not inferior to standard methodology, the Swedish study enrolled more than 80,000 healthy women who participated in population-based breast cancer screening between April 2021 and July 2022. The researchers They divided them into two groups: the control group, whose mammogram analysis was to follow the standard double reading procedure; and the intervention group, which would have the initial help of an artificial intelligence system to analyze the medical tests —and classify their degree of risk— before being reviewed and interpreted by one or two radiologists (one, if the risk marked by the AI was low and two if the mammogram was at the high danger threshold).
“There were no false positives among that 20%. They are confirmed cases of cancer.
Kristina Lång, Lund University
Analysis of the medical tests allowed the detection of 244 women with cancer in the AI-supported group and another 203 in the control group. That is: the incorporation of AI systems in the analysis allowed the diagnosis of 20% more tumors. “With AI-assisted detection, we detect 20% more cancers than with standard detection (double read without AI). There were no false positives among that 20%. They are confirmed cases of cancer. AI-supported detection did not lead to an increase in false positives, which is very good,” says Kristina Lång, a researcher at the Division of Diagnostic Radiology at Lund University (Malmö, Sweden) and author of the study. The false positive rate of all the tests carried out was similar in both groups: 1.5%.
Lång explains that these AI tools are used as “a screening aid in highlighting suspicious findings on images” and that “it can help the radiologist detect more cancers that might have been missed.” “AI can also be used to classify screening exams into low-risk and high-risk groups. An exam classified as high risk can alert the radiologist that there may be something suspicious of malignancy. There is a synergistic effect when human and artificial intelligence come together”, agrees the Swedish scientist.
Reduce workloads
These preliminary results of the study open the door to incorporating AI as an aid tool in the interpretation of mammograms, since apart from increasing case detection, it could serve to optimize the role of radiologists and reduce workloads. The researchers recall that, although the European authorities recommend the analysis of mammograms with double reading by two radiologists, this implies “a great workload for specialists and can potentially increase false positives.” And these inconveniences are not trivial, they point out, taking into account the shortage of radiologists who are experts in reading mammograms. “Also, despite the double reading, some cancers can be missed and diagnosed as interval cancers. [son los que se detectan entre dos pruebas de cribado]”, the scientists emphasize in the article.
The Swedish researchers, in fact, are also analyzing in this study —they need 100,000 participants for this and two years of follow-up—, if the use of AI to support the analysis of mammograms also reduces interval cancers, which usually have a more unfavorable prognosis. A high rate of interval cancer is an indicator that the early detection program is not serving its purpose, since it fails to diagnose that tumor early. “We hypothesize that AI may lead to a better screening program with fewer interval cancers. Since our first results show that we detect more cancers, there is a possibility that we could have a better and more efficient detection program”, Lång assesses.
AI should be a tool for the radiologist and not the other way around”
Kristina Lång, Lund University
The workload of radiologists can become very high and this affects their analysis capacity. Marina Álvarez, a breast specialist from the Spanish Society of Medical Radiology (SERAM), admits that in a day of reading mammograms, a radiologist can analyze more than 100. “And most of them are going to be normal and that, together with the fatigue of the radiologist, favors that some lesions may go unnoticed”, explains the specialist, who is also director of the Radiodiagnosis and breast cancer unit at the Reina Sofía Hospital in Córdoba.
Álvarez, who did not participate in the study, calls the research “very good and methodologically impeccable” and highlights the advantages of incorporating AI to support the analysis of mammograms: “AI has the ability to stratify studies based on risk , save time for the radiologist and improve the performance of the specialist”. In fact, the study confirms, indeed, that in the intervention group, the reading of mammograms was reduced by around 44%. “These systems can classify the study according to whether or not it is possible to have cancer because it can detect nodules, calcifications… This risk stratification, which a radiologist cannot do in such a short time, serves to treat each risk group differently. different: the majority are at low risk, only 30% are at intermediate-high risk and only 3% are at high risk; and in that 3% are 70% of cancers. That is why stratification is so important”, explains Álvarez. With this filter in place, the radiologist can go faster in the analysis of low-risk mammograms and focus more attention on the study of high-risk ones.
Josep Munuera, head of the Diagnostic Imaging service at Hospital Sant Pau in Barcelona and an expert in digital technologies applied to health, also maintains that this study, in which he has not participated, has a “very good” design and reinforces some “expected” results. “There are algorithms with high detection rates that, if you combine them with the human, improve the human rates as well.” The doctor also stresses the importance of the human reader —as in this study— being a highly experienced radiologist.
In the intervention group, 75% of the tumors detected were invasive —they are more widespread— and 25% were in situ —small lesions in the breast that may be harmless. In the control group, 81% were invasive and only 19% in situ. The researchers admit that increased detection of tumors in situ “could be concerning in terms of overdiagnosis” because these kinds of lesions may never progress to cancer or even go away on their own, and there is a risk of embarking the patient on further potentially unnecessary cancer tests or therapies. Álvarez, however, qualifies this point: “There are people who may think that 25% of tumors detected in situ is a high percentage because not all of them end up being cancer and they would be treated much more [de lo necesario]but we cannot differentiate which one will advance, we have no way of knowing if it is going to stay there or advance ”.
The radiologist sends
Faced with this dilemma, Munuera adds, the radiologist has the key. “These lesions, even if they are detected, are so subtle that they do not have to end up evolving into cancer. The radiologist will interpret the image and decide, ”he explains. But he adds: “Introducing the AI tools means seeing 100% of these injuries and you will have to make a decision about them. That is the next step, to see what we do with them”. Munuera celebrates, in any case, that the reduction of the workload in readings can favor “a very positive redistribution of work flows to give the radiologist more reading speed and reduce the waiting list for mammogram analysis and also to be able to carry out more interventional treatments and complementary tests” associated with the results of these screenings.
The Swedish researchers strongly emphasize, however, that AI is a tool at the disposal of the radiologist, but the specialist has the last word. “AI should be a tool for the radiologist and not the other way around,” says Lång. Neither are these systems unambiguous: “Based on retrospective studies, we know that AI misses some types of cancer. AI can also be overly sensitive, flagging many findings that we as radiologists can easily determine to be normal. Therefore, the radiologist is vital to make the final decision if a screening examination is normal or if the woman should be called back for an additional study, ”says the Swedish researcher.
Despite the favorable results of this study, there is still a long way to go to see AI replacing human reading. “European guidelines currently do not recommend that AI should replace a human reader,” agrees Lång. But studies like his underpin the evidence to move forward on that path. Always, yes, under the supervision of the radiologist.
You can follow THE COUNTRY Health and Well-being in Facebook, Twitter and instagram.
#Artificial #intelligencesupported #breast #cancer #screening #detects #tumors