Dogs Doing Targeted Cancer Detection: Only the Nose Knows

What if the answer to earlier tumor detection is right in front of our faces—or, more accurately, right in front of our dog’s faces?

Frankie, a scent-trained German Shepherd mix that can recognize the smell of cancer in thyroid tissue.

PHOTO COURTESY of Dr. Andrew Mark Hinson of Arkansas College of Medicine

Over the years, researchers have invested a lot of time and money into developing new technologies able to detect cancer before it is suspected. But what if the answer to earlier tumor detection is right in front of our faces—or, more accurately, right in front of our dog’s faces? The power of a dog’s nose has long been appreciated and put to work doing everything from tracking animals and people to sniffing out bombs and drugs. Now dogs are being trained to detect cancers.

Dina Zaphiris is the founder and CEO of InSitu Foundation, a nonprofit organization dedicated to training dogs to sniff out cancer. Zaphiris has spent most of her adult life training dogs. She started in pet training, but her career quickly shifted to scent detection. “I had my own deployable search and rescue dogs, and we did missing persons work, finding alive or decreased subjects in the wilderness. Then, I started working with officers and their dogs, training them to detect narcotics and explosives,” Zaphiris said. In 2003, her father sent her a news story about a California research group that planned to train dogs to detect cancer. When the scientists asked her if she would help train the dogs, Zaphiris—who was strongly affected by her mother’s breast cancer diagnosis in 1990—readily agreed.

The data were published in 2006 and showed that the dogs were able to distinguish between breath samples from lung cancer patients and controls with 99% sensitivity and 99% specificity and between breath samples from breast cancer patients and controls with 88% sensitivity and 98% specificity.1

“The results were so good that studies on dog’s abilities to smell cancer have popped up all over the world, from Japan to Canada to Australia to Italy,” said Zaphiris. Data shows that dogs can detect bladder cancer, melanoma, prostate cancer, and ovarian cancer. Most recently, a team of researchers from the University of Arkansas for Medical Science reported that they had used scent-trained dogs to detect thyroid cancer in urine samples with 90% accuracy.2

A couple of years ago, Jeffrey Marks, PhD from Duke University in North Carolina contacted InSitu about collaborating on a proof-of-concept study to see whether or not dogs could be trained to distinguish between benign breast disease and breast cancer in plasma. Zaphiris said this differs from previous studies, which investigated whether or not the dogs could identify cancerous breath or urine samples obtained from healthy controls, diseased controls, and cancer patients. Zaphiris works full-time training her dogs and recording all the data she collects for Duke, which she says is done rigorously and methodically.

Zaphiris has 3 dogs on her Duke team; 2 are pure-bred German shepherds that she rescued from shelters. “Any breed can learn how to do this, and the dog doesn’t have to be a pure-bred,” she said. “I’ve had poodles, Portuguese water dogs, Labradors, a Weimaraner, and Schnauzers. From soup to nuts, I’ve trained more than 50 dogs over 10 years to do it.” She said temperament is the most important factor and admitted that she prefers herding breeds like German shepherds and Australian shepherds because they were bred to work for people and have the energy and drive needed to perform the same task for a few hours several days a week.

The training process takes about 6 months. The handlers may expose the dogs to the scents when they are as young as 8 weeks old, but the dogs are not mature enough to work until they are around 18 months old. After that, they can work effectively for 10 to 11 years, provided they receive ongoing training. The dogs are rewarded with treats and balls (most seem to prefer balls) and are never punished. “We use all positive reinforcement. If they make a mistake, you take them out of the room and try again later,” Zaphiris said.

One of the most important things Zaphiris has learned in her years of cancer-detection training is that you cannot use the same samples twice. “Dogs have an amazing ability to memorize samples. If you trained him on 1000 samples, and you present him with the same samples, he will recognize them all,” she said. “A lot of people have done that, and there are tremendous flaws in the research,” she added.

The other risk in repeatedly exposing the dog to the same set of samples is that he will not be able to generalize his training to fresh samples. In other words, the dog learns to detect breast cancer in that single set of samples but if he is identifying the cancer by something other than the volatile organic compound common to all breast cancers, he will not hit on breast cancer in a fresh sample. Zaphiris said hundreds to thousands of samples are needed to train a dog correctly and that she never uses the same sample twice or a sample from the same person twice.

Zaphiris said one unofficial theory is cancer has a common odor, and her anecdotal observations support this. “If dogs are trained only on ovarian cancer, and I expose them to a new type of cancer and healthy controls, they pick up on the cancer,” she explained. Zaphiris said more studies are needed to determine whether the dog can tell the difference between different cancers or whether he smells something all cancer have in common. “But that takes time, funding, and samples.”

Securing funding remains a hurdle and is one reason Zaphiris founded the nonprofit InSitu. “All our work is funded by private donors,”she said. Duke recently included InSitu in a grant proposal to continue their research into dogs and cancer detection, but those funds may not be available for 2 years. University of California-Davis has also approached InSitu about the possibility of opening a center with a few dogs as part of a pilot cancer screening project, which Zaphiris is excited about. “It is really important for us to collaborate with very legitimate, great universities,” she said.

Zaphiris believes dogs have a lot of potential as cancer detectors. She said, “My idea is to have a centralized facility where the dogs are located and samples are sent in, like a laboratory today.” She acknowledged we may be years away from that becoming a reality but believes dogs should be used for cancer detection sooner, rather than later. “The dog could provide a low cost, noninvasive, extremely accurate method of early cancer detection. We train them on stage 0 and 1 cancers; there is no technology out there for such early-stage cancer,” she said. Zaphiris said dogs could also be called in to help when imaging results are inconclusive.

Inspired by the success of dogs at sniffing out cancers, some research teams have been working on mechanical noses to do the same thing. “If they can do that, great, but right now, they aren’t ready and that’s why universities are looking into dogs,” said Zaphiris, adding, “There is something huge here. Even if all the dogs ever do is teach us what it is they are smelling, that would be fantastic.”


  1. McCulloch M, Jezierski T, Broffman M. Diagnostic accuracy of canine scent detection in early- and late-stage lung and breast cancers.Integr Cancer Ther.2006;5:30-39.
  2. Researchers use scent-trained dogs to detect thyroid cancer. Published March 9, 2015. Accessed March 13, 2015.