Enhanced Artificial Rabbit Optimization with attention-based deep learning for leukemia cancer classification

Abstract Background Leukemia is a life-threatening cancer disease that affects people of all age groups and is a substantial cause of mortality globally. Leukemia is a white blood cell (WBC) related disease, which comes with an increase in the number of immature lymphocytes and damages the blood or bone marrow. Manual diagnosis of leukemia using microscopic blood sample images is a tedious, less accurate, and time-consuming process as leukemia cells and normal cells look similar. Methods The research work presents a model-Enhanced Artificial Rabbit Optimization with Attention-based Deep Learning for Leukemia cancer classification (EAROADL-LCC) by using microscopic images. In the presented EAROADL-LCC technique, the attention-based EfficientNet model is used to extract the relevant features from the input images. Then the EARO algorithm is designed by using the Levy flight (LF) model for hyperparameter tuning. Finally, an improved temporal convolution network (ITCN) model is applied for classification purposes. Results The experimental validation of the EAROADL-LCC technique takes place on a medical imaging dataset and the comparison results demonstrated its superior performance over recent methods. From the experimental analysis, it is found in our proposed model. Conclusion The proposed method achieved an accuracy of 99.56%. Additionally, the sensitivity (recall) improved to 98.1% and the precision reached 98.63%. The F1-score showed significant improvements, with values of 97.69% indicating the model's robust performance in distinguishing between leukemia and non-leukemia cases.