README.md

Using convolutional neural network for predicting diseases via leukocyte classification

Dataset

The LISC - Leukocyte Images for Segmentation and Classification has been used for automatic identification and counting of blood cells

Samples were taken from peripheral blood of 8 normal subjects and 400 samples were obtained from 100 microscope slides. The microscope slides were smeared and stained by Gismo-Right technique and images were acquired by a light microscope (Microscope-Axioskope 40) from the stained peripheral blood using an achromatic lens with a magnification of 100. Then, these images were recorded by a digital camera (Sony Model No. SSCDC50AP) and were saved in the BMP format. The images contain 720×576 pixels. All of them are color images and were collected from Hematology-Oncology and BMT Research Center of Imam Khomeini hospital in Tehran, Iran. The images were classified by a hematologist into normal leukocytes: basophil, eosinophil, lymphocyte, monocyte, and neutrophil. Also, the areas related to the nucleus and cytoplasm were manually segmented by an expert.

Requirements

Python:

Tensorflow:

OpenCV:

Weights:

Weights:

Setup

pip install numpy

pip install tensorflow.keras

pip install cv2

pip install matplotlib