What type of Artificial Intelligence for Detecting Early Signs of Diseases via Image Analysis?
I am seeking to build an Artificial Intelligence via Python that recognizes patterns in detecting early signs of disease, as the goal to prevent the disease.
I want to use Supervised Learning.
I was wondering what type of Artificial Intelligence fits this project best. Should I use Neural Networks?
The images I input are basically images from for example X-Ray where I have input in the information in the potential early stages of a patient’s disease, about whether it is a disease or not.
And since it’s supervised learning, I have also output images of diseases and non-diseases which the machine trains on.
In the end, I want to have the AI know from analyzing a picture whether it is a disease or not with great accuracy.
At the moment I am doing research. What concepts, subjects should I focus on?
I am trying this in Jupyter Notebook. Are there any more important factors that I need to know in order to make my project a reality? I am a beginner ANY help, tips would be APPRECIATED!
Image recognition in general is done via Convoluted Neural Networks - CNN in short.
Tensorflow and Google Collabs offer a nice codebase for this.
Though keep in mind, “great accuracy” is… a hard word given you are a single person and didn’t jump to CNN on the word “image”.
I’d recommend doing the Tensorflow course of FCC to get a basic understanding for how NN work and how to create them.
And then to give you a perspective: One of the challenges is to write a NN that can tell if an image is a cat or a dog. I utilized a pre-trained model for image-recognition from Google, which actual ML-scientists and worked on and designed. And it achieved 78% accuracy. Meaning about every 5th image with classified wrong.
So don’t expect to much of a “high” accuracy while looking for very subtle differences that would revolutionize modern medicine.