The Cancer Recognizer
An AI system based on a convolutional neural network that mimics the laboratory detection process and recognises potentially dangerous cells.
About The Cancer Recognizer project (Computer Vision)
Today cancer is one of the most widespread diseases on Earth. Fortunately it is possible to fight back against cancer if it is detected in time. Usually that is done by taking a testing sample and sending it to the laboratory, where the sample is placed under the microscope. After it is enlarged 400 or 1,000 times, the sample is inspected by an experienced doctor, who concludes whether it includes cancer cells or not.
Our development team came up with an idea to mimic that detection process and to train an AI-based system to recognise potentially dangerous cells. In order to achieve that goal we created a special system based on a CNN — a convolutional neural network — whose purpose is to analyse images from the dataset and make conclusions from them.
Before the images are provided to the system, we “cut” every cell out of the initial image — no matter whether it carries cancer or not — and so create a new dataset that includes only cells. This procedure uses a few instruments from the OpenCV library, and the technique eases the training process of our system significantly.
The conclusion of the system is a prediction provided to the user about the current sample: whether it includes cancer cells, and whether their amount is enough for diagnosing cancer or not.
So far the system is ready to diagnose only one type of cancer: mast cell. In the future we expect to expand the possibilities of the system to other types of cancer.
