TueScientists at the Medical University of Mainz have developed an artificial intelligence that is designed to more accurately predict the chances of recovery for patients with colorectal cancer. The Multi Stain Deep Learning model analyzes microscopic images of tumor tissue and evaluates the activity of cancer-attacking immune cells. According to researchers close to Dr. Sebastian Fersch, this method allows a better prediction of the success of therapy than, for example, manual counting of white blood cells, which has been common until now.
To train artificial intelligence, scientists used more than 300,000 microscopic images of tissue from about 1,000 people with colon cancer. By analyzing the recordings, the program determined what is known as the AImmunoscore, which provides information about how effectively the body’s defenses are attacking the tumor. Based on this, it is possible to predict the relapse-free survival time. According to the information, the accuracy of the forecast was 80 percent.
74 percent success with combination therapy before surgery
The research team also tested their model on patients who had already received radiation and chemotherapy prior to surgery. The goal was to predict whether the treatment would work. According to a study published in the journal Nature, 86 out of 117 patients were classified correctly, which corresponds to an accuracy of about 74 percent.
Feursch says some of the forecast models currently available are protected by patents and are only available commercially. “We want to make the program free and make it available to all researchers around the world.” The plan is to develop a web-based application where clinicians can upload image data and immediately receive a prognosis for their patients. “This will improve the treatment of colon cancer in the long term.”
Source: Frantfurter Allgemeine
I have been working as a news website author for the past year, and have written about a variety of healthcare-related topics. I am currently focusing on coverage of medical technology and innovation, as well as patient advocacy. I am also an avid cyclist and runner, and enjoy spending time outdoors exploring new trails or hitting the pavement for a run.