Our friends at DIG labs are back with a few articles about dog health! Below are several ways they’ve analyzed thousands of💩 images so far to create the next generation of pet technology!
Computer Vision and Dog Poop
How to get a glimpse into your dog’s health in 10 seconds
By Tara Zedayko, Co-Founder and CEO, DIG labs
In our last computer vision article, we chatted about the areas where computer vision has made our lives easier or healthier with handy tools. At DIG Labs, we are excited to add another area where computer vision can help – managing our dogs’ health. Below are several ways we’ve analyzed thousands of💩 images so far to create useful tools for pet parents and their veterinarians, as well as what’s up next!
Dog Poop Photos: The Basics
To start, our first goal was to train computers to classify visual information gleaned from stool images, into existing medical or technical standards. Speaking of standards, an approachable way of characterizing stool is to think about the “4 C’s” of dog stool, which we reviewed in detail here!
The scientists at DIG labs took these 4 C’s and created computer vision models, or ways to characterize images without human intervention, by employing what is known as “deep learning” techniques. Deep learning focuses on ways to train a computer to learn on its own based on feeding in highly structured information. Below, we’ll explain how we’ve built models around each of the 4 C’s of dog stool.
Dog Poop: Consistency
Dog poop consistency can be charted on a 7-point scale from 1 (constipation) to 7 (watery diarrhea), where scores of 3 and 4 are ideal. Known as the Bristol Stool Score (BSS), we can make recommendations about a dog’s stool health, including fiber content, hydration, and more – based on the observed Bristol Stool Score of a particular sample. Unfortunately, very few pet parents are familiar with this scale. In order to bridge this knowledge gap, we’ve been able to automate the classification of a sample into its respective rating – no PhD in BSS needed! Over time, we can also provide real-time nutrition tips and dog digestive system tracking based on each dog’s scores.
Dog Poop: Coating
Have you ever seen a slimy or jelly-like poop? Coating, or mucus, is a biproduct of the digestive tract, and is the unsung hero of immunity – keeping pathogens out and letting vitamins through! While scientists are just scratching the surface when it comes to itemizing all the amazing benefits of mucus, we know that sometimes (but not always!), mucus on stool can be a sign of “digestive distress” – that’s why we created a computer vision model that can detect the presence or absence of mucus on dogs’ stool.
Dog Poop: Color
While at times it may feel like “50 shades of brown”, dog stool color can be extremely insightful! While a dog’s stool color is clearly detectable to our untrained eyes, like yellow dog poop or green dog poop, computer vision models can vastly improve the ability to pick out colors and associate them with dietary or other nutritional factors, including protein types, fresh produce, and even speed of digestion! That’s why we focus on color to help provide an especially comprehensive stool assessment.
Did you know that a dog pooping blood is not always a cause for an emergency room visit? Causes of blood in dog stool can vary, from serious health concerns to stress-induced acute colitis, and nuances in color can help inform next steps. In these often-alarming situations, our goal is to help arm you and your vet with the visual and quantifiable information to best manage your dog’s health.
Dog Poop: Content
The last “c”, content, can include a wide array of characteristics – from easily identifiable undigested pieces of vegetables to harder to detect nuances like the presence of parasites or the type of microbes that comprise the intestinal microbiome.
On the parasite front, we are developing two computer vision models with more on the way! One can detect the presence of proglottids – an egg sac containing thousands of parasitic worm eggs. Proglottids can withstand dormancy and are often the culprit for how dogs contract intestinal worms – by eating leaves, grass, or drinking dirty water that contains proglottids. The other can detect the presence of adult worms. While both are indicative of quite advanced parasitic infections, many dogs present with “normal” looking poop otherwise – and therefore, the presence of eggs and worms can be easy to overlook.
Did you know that certain microbes are associated with the type of diet your dog eats? Dogs that consume higher carbohydrate loads in their diet are more likely to have a subset of bacteria, including Prevotella, present. By conducting canine gut microbiome DNA sequencing alongside image analysis, the scientists at DIG labs have found a novel correlation between visual attributes of stool, and the presence of these kinds of microbes, for dogs who were consuming relatively high carbohydrate diets. More to come on this exciting insight! Sign up for our non-invasive research programs to glean even more insights on your dog’s nutrition from leading science and veterinary experts.
Bringing Dog Poop Tech to Life
While all of this tech talk is great, a tool needs to meet the needs of dog parents – be easy to use, fast, and accessible. That’s why we’ve built the DIG Labs Health App, a convenient way to access these insights in less than 10 seconds no matter where you are! Data-based assessments like these arm pet parents with more robust, actionable, and tangible metrics about their pet’s health, assisting with taking the next steps to either monitor and optimize solo or with the support of your trusted veterinarian.
Sign up here to be one of the first to try the DIG Labs Health App!