Using A.I. to help veterinarians with disease diagnosisDeep Learning applied to Veterinary Medicine
challenge
Our client is a veterinary research lab focused on research on canine diseases. One of their specialties is on the treatment of bladder stones in dogs. Effective treatment depends on accurately identifying the mineral composition of the bladder stones. Some stone types require invasive surgery while others require only special diets. The client wanted to leverage advances in A.I. and computer vision to classify bladder stone compositions from a single radiograph image. Data imbalance among 6+ mineral types as well as ambiguity in composite stone types presented technical challenges.
solution
We implemented a deep learning based image classification solution in Pytorch and deployed the solution as an endpoint on AWS SageMaker.
results
We developed a state of the art classification model exceeding the accuracy requirement of our client. Project was delivered on time and within client budget.