A Sustainable Hybrid Explainable AI-Based Prediction Model of Cancer Stem Cell Behavior based on Microscopic Images

This paper presents a cancer stem cell prediction behavior system using a sustainable hybrid explainable artificial intelligence XAI-based model applied to cancer stem cell microscopic images. The proposed system integrates the XAI model in preprocessing extracted features with three enhanced transfer learning models AlexNet, GoogleNet, and ResNet50 to achieve better accuracy results of detection and classification stem cells. LIME (Local Interpretable Model-Agnostic Explanations) is used to enhance both classification accuracy and model interpretability. Experiments were conducted on three benchmark microscopic image datasets, which is (Human Against Machine with 9600 training images ). By cancer stem cell dataset, on the first experiment, Enhanced AlexNet, Enhanced GoogleNet, and Enhanced ResNet50 achieved classification accuracies of 85.64%, 88.08%, and 90.50%, respectively. On the second experiment, the models reached 95.15%, 96.06%, and 97.80% accuracy, respectively. Comparative analysis indicates that AlexNet consistently delivers superior performance across among datasets. By combining explainable LIME XAI technique with high-performing deep learning models, the proposed system provides a reliable and transparent solution for reliable and truth of cancer stem cell detection, supporting clinical decision-making and improving diagnostic confidence.