ScoliNosis

Project Briefing

Scoliosis is a condition where the spine curves abnormally. The curves can be "S"-shaped or "C"-shaped. In some situations, scoliosis can worsen over time, with severe scolisis affecting movement and breathing. This project involves using image classification deep learning techniques to detect scoliosis in its early stages and provide a treatment plan to improve the shape of the spine over time. The advantage of using deep learning techniques is to solve more complex connections in the data and make accurate inferences.

Project Highlights

I gained a deeper understanding of application-based deep learning through this project. Prior to this project, I had only explored the mathematical aspect of neural networks and implemented MATLAB scripts to classify 2-D points. This project enabled me to investigate building models in Python using Keras, and with the Azure Custom Vision API. I also gained exposure to implementing object recognition software in OpenCV. Overall, I am satisfied with what we accomplished and glad to have been part of a great team.

Demo

Check out our demonstration below to see our project in action!

Skills

Transfer Learning Convolutional Neural Networks
Model Training Dropout Max Pooling
Dense Layers Data Serialization

Tools

Keras Python OpenCV
Azure Custom Vision API

Experience

Implemented deep learning algorithms that classify images of a user's back based on perceived severity.
Transformed the concept into a working prototype with 3 teammates and a 36-hour turnaround through effective organization, action planning, team strategy and collaboration.

Work Samples

Have more questions? Take a look at our project website here !