This repository demonstrates how we can simply use transfer learning with TF Hub onto the red blood cells images. The Malaria datasets can retrieve from NIH website here.
The image is the segmented cells from the thin blood smear slide images from the Malaria Screener research activity. The initial file is stored in zip file, with each class stored onto different folders; namely Parasitized and Uninfected.
- Malaria is still one of the deadly and most common infectious diseases.
- The most accurate way to diagnose malaria is by taking a drop of blood, smearing it on a slide, and then examining it under a microscope to look for malaria parasites inside the red blood cells.
- The following process requires the avaialability of microscope, electricity, and trained medical staffs to look at the slides. The process is also time-consuming as the staff can only look at one slide at a time.
- With the advancement of Deep Learning (Machine learning), it can help optimizing the process by looking at the slide images instead.
Below is the sample of red blood cells (no infection and with infection):
