An integrated and automated testing approach on Inception Restnet-V3 based on convolutional neural network for...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Hematology

Background:

• White blood cell (WBC) classification is crucial for diagnosing various medical conditions, including infections and leukemia.
• Traditional methods often involve manual analysis, which can be time-consuming and prone to error.
• Machine learning (ML) offers automated solutions for WBC classification, improving efficiency and accuracy.

Purpose of the Study:

• To develop and evaluate an automated WBC classification system using deep learning.
• To improve the accuracy and efficiency of diagnosing blood disorders through AI.

Main Methods:

• The study proposes an Inception ResNet-v3 model, integrating Inception architecture and ResNet connections.
• The model was trained on a dataset of 15.7k human peripheral WBC images across five categories.
• Pathologist-verified diagnoses were used for training the classification model.

Main Results:

• The Inception ResNet-v3 model achieved high accuracy in classifying five types of WBCs.
• Performance was superior to existing models like VGG, U-Net, and ResNet.
• Testing on public datasets (Kaagel, Raabin) yielded accuracies of 98.80% and 98.95%, respectively.

Conclusions:

• The proposed Inception ResNet-v3 model demonstrates significant potential for enhancing clinical blood examination diagnostics.
• It offers a promising, accurate, and efficient alternative to traditional ML methods for WBC classification.
• The model achieved excellent performance metrics, including Accuracy, Precision, Recall, Specificity, and F1 Score.