Authors:
B. Santosh Kumar, Rakesh Chandrashekar, Thirumalraj Karthikeyan, S. Venkatasubramanian, Sheila Agnes Vidot

Addresses:
Department of Computer Science and Engineering, New Horizon College of Engineering, Bengaluru, Karnataka, India. Department of Mechanical Engineering, New Horizon College of Engineering, Bengaluru, Karnataka, India. Department of Artificial Intelligence, Trichy Research Labs, Quest Technologies, Tiruchirappalli, Tamil Nadu, India. Department of Computer Science and Business Systems, Saranathan College of Engineering, Trichy, Tamil Nadu, India. Department of Medical Consultant, IPS Health, Mahe, Victoria, Seychelles.

Abstract:

White blood cells are needed to assess the immune system, but the pathologist's ability influences the blood smear's conclusion.  Most machine learning methods classify white blood cells at a single level.  White blood cells must be assessed to assess the human immune system, and this study provides a new method. This work improves classification accuracy by leveraging deep learning rather than pathologists' expertise. Preprocessing using Edge Image Enhancement reduces noise. Five datasets with four cell groupings are used. The innovative barnacles mating optimiser (BMO) selects features. The method's core is a progressive Wasserstein generative adversarial network with gradient penalisation (PWGAN-GP).  Progressive training makes training the generator model easier. Built sample resolution is improved iteratively. A loss function is added to the discriminator to quantify sample similarity, thereby improving sample reliability and authenticity.  The Aquila Optimiser Algorithm tunes the hyperparameters of the PWGAN-GP classification model. The proposed PWGAN-GP model outperformed others across all datasets, achieving 99.22% in Blood Cell Detection, 99.18% in Complete Blood Count, 98.7% in White Blood Cells, 98.82% in Kaggle Blood Cell Images, and 98.91% in Segmentation, as well as Classification.

Keywords: Feature Selection; Image Enhancement; Noise Reduction; Gradient Penalisation; Aquila Optimiser Algorithm; Hyperparameter Tuning; Barnacles Mating Optimiser; Cell Detection.

Received on: 02/01/2025, Revised on: 13/03/2025, Accepted on: 12/05/2025, Published on: 06/12/2025

DOI: 10.69888/FTSHSL.2025.000510

FMDB Transactions on Sustainable Health Science Letters, 2025 Vol. 3 No. 4, Pages: 202-217

• Views : 208
• Downloads : 24

Download PDF