A Deep Learning Model for Human Blood Cells Classification

EasyChair Preprint no. 10561

Abstract

Blood cells carry the significant information to represent the overall health status of the person. The rapid identification of the various blood cells is key to reduce the daily health infection risks. Deep learning model for blood cell classification based on the capability of the convolutional neural network with transfer learning is proposed with Mobilenet as base module for rapid identification of different blood cells in a multi-class identification problem. The proposed model achieves performances as 96.22%, 98.88%, 98.51%, 96.24%, 96.33%, and 98.86% in terms of recall (SE), specificity (SP), accuracy (Az), F1-Score, positive predictive value (PPV), negative predictive (NPV), respectively. The model has achieved promising results comparing with the latest deep learning models for multiclass blood cell classification purpose. The rapid and promising recognition rate for blood cells detection is necessary to accelerate the health care facilities.

Keyphrases: blood cell, CNN, deep learning, Transfer Learning

@Booklet{EasyChair:10561,
  author = {M Pramodha and S Ansith and J V Bibal Benifa and Mohammed Al-Sarem and Hanumanthappa Jayappa Davanagere and A A Bini and Emmanuel Ndagijimana and Faisal Saeed and Md Belal Bin Heyat and Abdulrahman Alqaraf and Abdullah Y Muaad and Channabasava Chola},
  title = {A Deep Learning Model for Human Blood Cells Classification},
  howpublished = {EasyChair Preprint no. 10561},

  year = {EasyChair, 2023}}