Lung and pancreatic cancers are very often difficult to treat, especially when the cancer has spread to other organs.
Today, the five-year survival rate for lung cancer that has spread is 5%. For pancreatic cancer, it's 3%.
Now, a group of researchers have developed a system to use computers to increase a patient's odds.
One of the biggest challenges in cancer treatment is catching it at an early stage, before it has spread.
"Something around 30% to 40% of cancer is missed during the early stages of screening," University of Central Florida Center for Research in Computer Vision Ph.D. candidate Naji Khosravan said.
But doctors may soon have new tool to help in the fight against cancer. Researchers at UCF have developed an artificial intelligence system to detect tiny specks of lung cancer in CT scans.
"The radiologists tend to miss them because they look like the normal tissue, and it is very natural to miss those tumors," Dr. Ulas Bagci said.
The researchers train the system to look for patterns in the scan to find tiny tumors. Then, the system analyzes it and determines whether the tumor is cancerous. It's 95% accurate, compared to 65% when done by a human.
"So we are actually helping them to improve their screening strategies and especially for very small nodules," Bagci said.
The success has led them to a partnership with the Mayo Clinic in Florida to develop a similar system to spot premalignant cysts in the pancreas, which can lead to pancreatic cancer.
"With pancreatic cancers, over half of patients are not diagnosed until a very late stage," UCF Ph.D. student Rodney LaLonde said.
"So, our idea is to find the pancreatic cyst before they turn into cancer," Bagci said. "It is going to improve the life span of people, because you are capturing the lung cancer or pancreatic cancer at the early phrase."
And it is improving the survival rates of patients.
The researchers have already filed for a patent for the AI system to detect pancreatic cysts. Now, they are looking to get Food and Drug Administration approval for clinical use for both the lung and pancreatic cancer systems so doctors can start using them in hospital settings.
TOPIC: USING AI TO STOP CANCER IN ITS TRACKS
REPORT: MB #4591
BACKGROUND: There are two major types of lung cancer, non-small cell lung cancer (NSCLC) and small cell lung cancer (SCLC). Staging lung cancer is based on whether the cancer is local or has spread from the lungs to the lymph nodes or other organs. Because the lungs are large, tumors can grow in them for a long time before they are found. Even when symptoms—such as coughing and fatigue—do occur, people think they are due to other causes. For this reason, early-stage lung cancer (stages I and II) is difficult to detect. Symptoms of lung cancer that are in the chest include coughing, pain in the chest, shoulder, or back unrelated to pain from coughing, shortness of breath and recurrent lung problems, such as bronchitis or pneumonia. If the original lung cancer has spread, a person may feel symptoms in other places in the body. Common places for lung cancer to spread include other parts of the lungs, lymph nodes, bones, brain, liver, and adrenal glands. (Source: https://www.lungcancer.org/find_information/publications/163-lung_cancer_101/266-symptoms)
DIAGNOSING: Ulas Bagci, PhD, Professor of Computer Science at the Center for Research in Computer Visual at UCF has been working on an AI system for over ten years now. He said, "We have millimetric resolution. So one, two, three millimeters, we can still capture these tumors. But usually in the lung cancer screening strategies if the tumor is less than three millimeters it is considered a very small tumor. So our technique is actually very sensitive to capture. Even in that range, one to three millimeters. So those tumors are called early phase because they're small. If you detect them then it is highly likely that the patient can be survived in the easier treatment strategies. And we are able to capture those tumors. Usually those are missed very easily." (Source: Ulas Bagci, PhD)
NEW TECHNOLOGY: Bagci says small tumors can be missed 30 to 40 percent of the time, but with the AI system, "Our overall system has more than 90 percent accuracy and for specific tumor size our system is able to detect more than 95 percent." He and colleagues are also developing an artificial intelligence algorithm to tell if the tumor is malignant or benign. (Source: Ulas Bagci, PhD)