Editor’s Note: We believe that patients are a key part of developing and leading the conversation in disease communities. Patient Worthy sometimes partners with reputable agencies that wish to speak with patients about opportunities related to their diagnosed conditions. These opportunities can include activities such as sharing stories with other patients or health professionals about their diagnosis journey or recording video testimonials. To learn more about how to get involved with an opportunity for prostate cancer patients, click here.

It’s no secret that dogs have insanely strong olfactory capabilities, meaning they can sniff out a wide variety of things not detectable to the human nose. Since dogs have 300 million olfactory receptors compared to our 6 million, this isn’t quite a surprise. Yet dogs have showed promise in sniffing out explosives, invasive species, bed bugs, and health conditions – like low blood sugar, melanoma, and other forms of cancer. Now, shares GEN, researchers believe that an artificial neural network (ANN) could learn from dogs’ olfactory abilities and be able to identify aggressive prostate cancer. Check out the full study findings in PLOS One.

Training Neural Networks

Currently, the prostate specific antigen (PSA) screening test is one of the most common tests designed to determine whether someone has prostate cancer. However, the test is notoriously finicky. In some points, it mistakes prostatitis or other conditions for prostate cancer. It may also falsely state that a cancer is aggressive when it isn’t, or totally miss aggressive cancer. For this reason, researchers believe that new prostate cancer screening techniques are necessary. In particular, these more sensitive tests should be able to distinguish between high-risk and low-risk cancers.

One potential approach that researchers considered was using dogs and their olfactory sense. Olfactory relates to the sense of smell. Prior research has shown that, using urine samples, dogs can be trained to detect prostate cancer. As the authors of the paper explain:

This study included 362 cases and 540 healthy controls, with a striking mean sensitivity (2 dogs) of 99% and mean specificity (2 dogs) of 98%. These findings support the premise that olfactory detection of prostate cancer holds the promise of rapid and non-invasive diagnosis.

Unfortunately, this would not be an efficient or practical solution on a wider scale. This is due to a limited number of dogs trained to perform this task and the number of resources that would go into training larger groups. However, researchers questioned whether they could harness the olfactory capabilities alongside volatile organic compound (VOC) detection and urinary microbiome analysis to train neural networks to do the same type of detection.

“Sniffing Out” Prostate Cancer

Rather than one singular approach to diagnose prostate cancer, researchers believe that a combination of methods would offer a more comprehensive approach. By using the strengths of multiple methods, researchers believed they could learn about prostate cancer pathology, create more effective diagnostic tools, and create a machine-based option for urinary screening.

To begin their study, researchers submitted urine samples for detection. Samples were sourced from patients with proven prostate cancer and also without proven cancer. For detection, they used VOC identification using GC mass spectrometer analysis, dogs, and microbiome analysis. After training the dogs to identify aggressive prostate cancers, these same dogs were able to identify positive cases around 71% of the time, and negative cases with 70-76% specificity.

Next, researchers analyzed microbial species and volatile compounds in urine and were able to identify positive vs. negative cancer samples. Then, researchers decided to blend the analysis with the olfactory capabilities, creating a more immersive diagnostic tool.

Researchers then took their data from the dogs’ olfactory capabilities and compared it to the microbiota and VOCs. From this, they were able to identify which elements appealed to the dogs’ olfactory sense. They used this data to inform an artificial neural network (ANN). Investopedia describes an ANN as:

the piece of a computing system designed to simulate the way the human brain analyzes and processes information. It is the foundation of artificial intelligence (AI) and solves problems that would prove impossible or difficult by human or statistical standards.

In the future, additional research is needed using a larger sample size. This could also give better insight into these methodologies and how they can be used. One thing researchers need to determine is how to further imbue ANNs with dogs’ skills and intelligence. However, researchers are confident that using a multi-varied approach to diagnosing prostate cancer will ultimately end in more accurate diagnoses and better treatments.

Prostate Cancer

Prostate cancer forms in the prostate, a small gland in males which produces seminal fluid. Globally, this cancer is fairly common. Within the United States, for example, prostate cancer is the 2nd leading cause of death. An estimated 1 in every 9 males will have prostate cancer throughout their lifetime. Most of the time, the cancer stays confined to the prostate. However, in other situations, the cancer is aggressive and might spread. Risk factors include age (50+), obesity, a family history of cancer, and race.

Patients often have better outcomes if the cancer is detected early. However, many people will not show symptoms in early stages. When symptoms do appear, they may include:

• Bloody urine or semen
• Difficulty urinating
• Bone pain
• Erectile dysfunction
• Unintended weight loss
• Decreased force in the urine stream
• Pelvic pain
• Appetite loss

Learn more about prostate cancer.