Overview / Usage

As a student in the medical field, I have observed and learned the wide range use of medical imaging to bridge medical diagnosis and computer algorithms. I have learned and understood how important the immune system is to the human health. The main role of the immune system is to protect us against diseases. It is mainly composed of cells called the white blood cells, amongst other cell types all present in the blood. Knowing the amount and quality of the different categories of this particular cell type present in the blood of any person ,depicts the quality of health of the person. In the clinical setting, laboratory professionals use various methods, from manual to automated mechanisms of detecting and counting the white blood cells and its types in a blood sample. These procedures can be very tedious,slow and inefficient. However, this is a problem with an excellent and promising solution with the help of artificial intelligence. What if some few lines of code could allow anyone anywhere to take a picture of a blood sample, and it instantly tells the number of white blood cells , and the level of immunity conferred by this blood. This is the problem I am solving using deep learning and computer vision library, OpenCV. The solution will take as input, white blood cell images, and produce as output ,the type of white blood cell it is.

Methodology / Approach

My approach to this problem is to apply multiple hidden layers to a preprocessed white blood cell input image, to learn the weights and filters via back propagation and give an accurate multiclass output of the type of white blood cell. It will use Open CV for the image preprocessing during which I will locate the nucleus of a cell, and based on the shape; which could either be a mononuclear or a polynuclear lobed cell and other higher dimensional features to make my classification. Also ,I will augment my training and testing sets using shearing ,flipping ,and rotation of my input images.

I am using keras framework with a tensorflow backend in a python development environment with Anaconda navigator jupyter notebook.

Technologies Used

1. Intel® Core™ i5-4300M CPU @ 2.60GHz × 4 processor

2. Keras deep learning library

3. Open cv

4. Anaconda navigator for python programming