Explainable artificial intelligence based microscopic peripheral blood cell image classification by exploiting quadradic convex optimization

doi:10.1016/j.engappai.2025.112040

Explainable artificial intelligence based microscopic peripheral blood cell image classification by exploiting quadradic convex optimization

2025-08-18

0

PRE

AI

DOI：10.1016/j.engappai.2025.112040

Original

Original request for help

Save

Abstract

En 中文

Deep learning models have shown remarkable efficacy in classifying microscopic peripheral blood cell images. However, for clinical applications, it is essential to provide transparency in their evaluation. To improve explainability, the Local Interpretable Model-agnostic Explanations (LIME) in Explainable Artificial Intelligence (XAI) methods effectively translate positive outcomes into practical usage. Benchmark data was collected from Kaggle and pre-processed using Min-Max normalization. Feature extraction was performed using a compound scaling method in neural networks, followed by classification using quadratic convex optimization. Our experimental results indicate that the proposed model achieves an impressive test accuracy of 99.678 % for classifying peripheral blood cell images. The correlation between computational learning and expert insights was observed, emphasizing the impact of human knowledge on enhancing computational outcomes. Explainable AI techniques will empower medical practitioners to identify morphological abnormalities in blood cells and prioritize critical findings, aiding in severity assessments and improving patient care.

Keywords:

deep learning

explainable artificial intelligence

peripheral blood cell classification

LIME

quadratic convex optimization

Explainable artificial intelligence based microscopic peripheral blood cell image classification by exploiting quadradic convex optimization

Abstract

Journal

Organization