It is an invention that could make it easier to diagnose certain types of cancer, particularly cancer of the pancreas and ovaries. Newatlas.com learns that a team of researchers from the University of Pennsylvania in the United States has developed an artificial nose that can detect signs of cancer in plasma samples. Developed by the Perelman School of Medicine, the non-invasive tool uses AI and machine learning to sniff blood samples and distinguish odors from cancer cells.
Volatile organic compounds as an index
To do this, the device analyzes the odors emitted by volatile organic compounds (VOCs). As a reminder, these are chemical compounds that can easily be found in the environment in gaseous form. Their composition varies depending on the source. This is not the first time such an approach has been used to detect cancer. For example, in previous experiments, another research group successfully diagnosed a patient with lung cancer by using dogs to sniff his breath.
Detect hard-to-diagnose cancers with blood plasma samples
In addition, electronic devices that can detect cancers of the stomach, esophagus and throat have shown promising results. Using urine samples, an artificial nose was even able to detect signs of prostate cancer. For this new research, scientists used an e-nose to analyze the mixture of volatile organic compounds emitted by blood cells.
The research aimed to find signs of cancer that are usually difficult to spot, such as pancreatic and ovarian cancer. The device contained algorithms that were previously trained to detect VOCs associated with different types of cancer. It is also designed to be able to assess the stage of the disease and whether or not it is benign.
Accuracy of up to 95%
For the study, the researchers called on 93 people. 20 of them had ovarian cancer, 20 others had benign ovarian tumors, 13 had pancreatic cancer, and 10 had benign pancreatic disease. The last 30 formed a control group.
The tool was able to differentiate ovarian cancer with 95% accuracy and pancreatic cancer with 90% accuracy. In addition, he correctly identified eight patients with early-stage cancer. “It’s an early study, but the results are very promising,” said Charlie Johnson, co-author of the study, in a statement.
“The data shows that we can identify these tumors in the late and early stages, which is exciting. If developed for the clinical setting, this test could be performed as part of a standard blood test as part of an annual medical exam, ”added Johnson, Forbes said.
Sniff blood samples for cancer. Photo credit: Shutterstock / Yurich20