E-Nose Sniffs Out Hard-to-Detect Cancers With Over 90% Accuracy

Date:

A researcher team at the University of Pennsylvania has developed an electronic nose (e-nose) that could sniff out signs of cancer from blood samples. In lab tests, the device successfully detected a range of cancer types with over 90% accuracy.

Volatile organic compounds (VOCs) are chemicals that have a high vapor pressure at room temperature and are responsible for odors, with different sources emitting different mixtures. The human nose is a sensitive instrument that can detect subtle differences in the makeup and ratio of these VOCs and distinguish whether that odor is coffee, flowers, or lemons. Digitizing that perception is the goal of newer work, including a Yale-led AI model that predicts how people sense odor mixtures.

In 1971, Nobel Prize winner Pauling and his team published what is considered the first report revealing (by gas chromatography) the presence of hundreds of VOCs in human urine and breath. Since then, scientists have explored how VOCs given off by cancer could be detected as part of a diagnostic system. For example, sniffer dogs have shown great promise in detecting cancer in patients’ blood with almost 97% accuracy, while non-invasive breath tests can quickly detect exhaled breath profiles associated with neck and head cancers with 80% accuracy.

For the new study, the team used an e-nose to analyze blood plasma samples for signs of hard-to-detect cancers, like ovarian and pancreatic cancers. The e-nose used algorithms that had previously been trained to identify specific VOC combinations with the different cancers, whether they were benign or not, and what stage of progression they were at.

The researchers studied samples from 93 patients—20 with benign ovarian tumors, 10 with benign pancreatic disease, 20 with ovarian cancer, 13 with pancreatic cancer, and 30 age- and sex-matched controls.

Impressively, the e-nose detected pancreatic cancer with 90% accuracy and ovarian cancer with 95% accuracy. Of those, it differentiated all eight patients who had early-stage cancers, suggesting it could be valuable as a diagnostic tool to find the disease before it’s too late.

Charlie Johnson, the study’s co-author, explained, “It’s an early study, but the results are very promising. The data shows we can identify these tumors at both advanced and the earliest stages, which is exciting. If developed appropriately for the clinical setting, this could potentially be a test that’s done on a standard blood draw that may be part of your annual physical.”

However, since this research involved only a small cohort, much work is needed to confirm the results. Johnson presented their findings at the American Society of Clinical Oncology meeting on June 4.

Luana Steffen
Luana Steffen
I am an artist who enjoys sharing interesting information and creative thinking with the world to inspire people.

Share post:

Popular

GPT-6.2 Astra vs Sol: Why the $2 Model Is the Smarter Default

OpenAI launched GPT-6.2 Astra and GPT-6.2 Sol on November...

CO2 Battery Storage Turns Thin Air Into a Day of Solar Power

On a windy afternoon in Sardinia, a giant fabric...

How AI Weather Forecasting Is Reshaping Atmospheric Science

For more than sixty years, weather prediction has meant...

Qwen4-Max Is Here: Alibaba’s 2.4T Agent Powerhouse Tested

Alibaba flipped the switch on the Qwen 4 era...