AI detection of distal
Create an AI assistant for physicians to help them with diagnosing a Distal Radius fracture using an x-ray image.
Build new expertise in the Artificial Intelligence area, specifically in building the Machine Learning models for Image Recognition using Image Augmentation* process.
* – Image augmentation is used in deep learning and computer vision tasks to increase the quality of trained models. The purpose of image augmentation is to create new training samples from the existing data.
A web service which is able to automatically detect and classify distal radius fractures using an x-ray image and a machine learning model.
Benefits and Results
According to statistics, about 20% of all fractures occur at the junction of the radius and hand. The developed service allows users to upload their own x-ray image of the wrist with a fracture, or select a picture from the prepared collection. After uploading the image to the service’s server, it is processed by 3 neural networks:
- The first neural network checks if the x-ray image is of a wrist x-ray (i.e. distinguishes wrist x-ray image from anything else)
- The second neural network analyzes the probability of a particular type of fracture (Smith, Colles, Barton)
- The third neural network locates the site of fracture
As a result, we get the type of fracture and its coordinates in the picture available for doctors.
Even with a small number of images initially (~30 images per distal radius fracture types), our developers were able to apply the image augmentation and get new training samples from the existing data to increase the quality of trained models.
Python, Flask, PyTorch, React, Docker, Kubernetes, Image augmentation, Deep Learning