Machine learning-based detection and quantification of red blood cells in Cholistani cattle: A pilot study

Area of Science:

• Veterinary Medicine
• Hematology
• Machine Learning
• Artificial Intelligence

Background:

• Accurate red blood cell (RBC) counting is crucial for animal health diagnostics.
• Manual methods are labor-intensive and prone to human error.
• Exploring automated, AI-driven solutions for veterinary hematology is essential.

Purpose of the Study:

• To develop and evaluate a machine learning model for detecting and counting normal and abnormal RBCs in Cholistani cattle.
• To compare the performance of a Support Vector Machine (SVM) model against manual counting techniques.
• To assess the classification accuracy and precision for normal RBCs, tear-drop cells, and schistocytes.

Main Methods:

• Utilized pre-annotated blood smear images from Cholistani cattle.
• Applied contrast stretching, segmentation, and resizing for image preprocessing.
• Employed Principal Component Analysis (PCA) for feature reduction and an SVM model for classification, with data augmentation and an 80/20 train-test split.

Main Results:

• No statistically significant difference (P ≥ 0.05) was found between machine learning-based and manual RBC counts.
• The SVM model achieved high classification probabilities for normal RBCs (87%), tear-drop cells (84%), and schistocytes (73%).
• Accuracy was highest for tear-drop cells (0.991), followed by normal RBCs (0.965), and precision was also high for normal RBCs (0.932) and tear-drop cells (0.921).

Conclusions:

• Machine learning, specifically SVM models, can accurately and precisely detect and count normal RBCs and tear-drop cells in Cholistani cattle.
• The AI approach demonstrates comparable results to manual counting, offering a viable alternative for veterinary hematological assessments.
• Further research using deep learning methods like convolution neural networks is recommended to enhance RBC detection capabilities.