The researchers tested the procedure in a scan taken one year before the same 1,179 patients were diagnosed with tumors, and were able to identify 152 suspicious areas that were later diagnosed with cancer.
Madrid (European News). According to research presented at the International Congress of the European Respiratory Association, AI programs can detect signs of lung carcinoma in CT scans up to a year earlier than the current methods.
Lung cancer is the leading cause of death from cancer. Around 1.8 million people are diagnosed with lung cancer each year. This is a deadly form of cancer that can lead to death in a very advanced stage.
Artificial intelligence can be used to assist lung cancer screening. This could simplify the process and help more patients get diagnosed early.
The CT scan or computed tomography can be used to detect lung tumors. After that, a biopsy is done to confirm malignancy. Each scan is scanned by a radiologist who examines about 300 images to look for any signs of small-sized cancer.
Although computed tomography has shown promise in detecting lung cancer-prone individuals, the screening process is complicated by radiologists having to examine each image individually to determine who requires further investigation.
This new research was presented by Benoît Audelan, a researcher on the Epione project team at the Inria Center (French National Institute of Digital Science and Technology) of the University of the Cote d’Azur, who worked with colleagues from French universities. Côte d’Azur. Azul, Therapixel, Nice University Hospital and Nice Software Company specializing in artificial Intelligence of Medical Images.
Researchers used CT scans taken from 888 patients to train their AI software. Radiologists examined these patients for suspicious growth.
The researchers then tested in another 1,179 patients, who were part a three-year lung screening study. After the last scan, 177 of these patients were diagnosed by biopsy with lung cancer.
These scans revealed that the program was able to detect 172 out of 177 cancerous tumors. This means that 97% of patients were diagnosed with lung cancer. The program missed five tumors near the center, which is where it is more difficult to differentiate healthy tissue from malignant tumors.
Researchers also tried the procedure on a scan that was taken one year prior to the diagnosis of the 1,179 patients with tumors. They were able identify 152 suspicious regions which were later confirmed as cancerous.
The researchers said that the procedure can also identify suspicious areas that aren’t cancerous (false positives) and that it must be improved before it is used in the clinic. This will prevent unnecessary living tissue Inspection.
According to Audelan, “Lung cancer screening will involve more scans, and we don’t have enough radiologists to review all of them. We need computer programs that can assist. This program can detect signs of lung cancer, according to our research. “This program could have detected lung cancer up to a year earlier,” he said.
He stated that his research does not aim to replace radiologists. Instead, it aims to provide them with tools that can detect early signs of lung carcinoma.
The researchers are working on a new method to distinguish malignant from non-malignant tissue. This will allow radiologists to decide which patients should be further investigated.
Professor Joanna Chorostowska-Wynimko, secretary-general of the European Respiratory Society and consultant of the Department of Respiratory Medicine at the National Institute of Pulmonology in Warsaw, who did not participate in this study, pointed out that “early diagnosis of lung cancer is essential for improving survival rates, and screening will be the way to achieve this goal. This is an important step. “Studies have shown that CT screening may reduce the deaths from lung cancer,” he said.
He stated that this work was promising as it showed how artificial intelligence could be used to quickly review scans and detect early signs of cancer. Researchers must first improve their ability distinguish between abnormal but benign lung tissue, and lung tissue that could become cancerous before they can use the program.