A research led by ISABIAL and Dr. Balmis Hospital evaluates the use of Artificial Intelligence to improve colorectal cancer prevention
The study, published in the prestigious international journal Annals of Internal Medicine, involves more than 3,000 patients from 6 healthcare centers and concludes that AI does not improve the diagnosis of this condition
Researchers Rodrigo Jover and Carolina Mangas explain that the aim is for new technologies to help detect a greater number of lesions, so further research is needed
Alicante (29.08.23). A research led by the Alicante Institute for Health and Biomedical Research (ISABIAL) and Dr. Balmis General University Hospital has evaluated the use of artificial intelligence devices applied to colonoscopy in the detection of colon polyps as a measure to prevent colorectal cancer. The results of the multicenter project, led by researchers Rodrigo Jover and Carolina Mangas, both from the Digestive Medicine Service of Dr. Balmis Hospital, have been published in the prestigious international journal Annals of Internal Medicine.
The study concludes that the use of systems based on artificial intelligence does not improve lesion detection and that further research and more clearly defined detection parameters are needed before its use in clinical practice becomes relevant. These results contradict what studies on artificial intelligence are showing in medical applications and indicate that, in the field of digestive endoscopy, “there is significant room for improvement for this promising technology”.
The research led from Alicante involved the participation of research staff from a total of 6 healthcare centers. In addition to ISABIAL researchers at Dr. Balmis Hospital in Alicante, researchers from the University Hospital Complex of Ourense, Álvaro Cunqueiro Hospital in Vigo, Hospital Clínic of Barcelona, Río Hortega Hospital in Valladolid, and the Central University Hospital of Asturias also took part.
When promoting this study, the research staff took into account that in recent years there has been very significant development in artificial intelligence in general, with advances in its application in the medical field. Specifically, in colonoscopy, devices have been developed that allow the identification of colorectal polyps during the procedure, assisting the endoscopist in detecting these premalignant lesions.
Studies published to date have shown that these artificial intelligence–assisted detection devices improve the visualization of small lesions. However, the clinical relevance of these lesions is relatively low. It is in larger lesions (≥1 mm), with more advanced histological features such as dysplasia or a villous component, where there is a higher risk of developing colorectal cancer (CRC), which is why their identification and removal is a key aspect in preventing these tumors. Researchers indicate that “the role played by the use of artificial intelligence in improving the detection of these lesions is a question that had not yet been addressed”.
Participation
A total of 3,399 patients were included in the study, all of whom were participating in a population-based colorectal cancer screening program after a positive fecal occult blood test. Patients were randomized to receive a standard colonoscopy or one assisted by a computer-aided detection device. The researchers point out that “this is the study that has included the largest number of patients of those carried out worldwide so far”
Researcher Carolina Mangas explains that “contrary to our initial hypothesis, we have not been able to demonstrate that the use of artificial intelligence devices improves the detection of this type of lesion.” However, Dr. Carolina Mangas indicates that “the ultimate goal is for artificial intelligence to help endoscopists miss as few lesions as possible and to improve and standardize the quality of colonoscopy”.
For his part, researcher Rodrigo Jover, who is also head of the Digestive Medicine Service at Dr. Balmis Hospital, explains that “this is the first clinical trial that has focused on the most clinically relevant lesions and has been carried out in a very specific and highly important context, such as population-based colorectal cancer screening”.
Nevertheless, Dr. Rodrigo Jover points out, “it is necessary to continue improving this technology by using larger image databases to train computer-aided detection devices in recognizing this type of lesion”.
The preliminary results have been presented at the congress of the European Society of Gastroenterology and at the American Gastroenterology congress. In addition, it has received the award for best clinical research work in endoscopy granted by the Foundation of the Spanish Society of Digestive Endoscopy in collaboration with the company Casen Recordati.
Author: José Antonio Más Cayuelas
