OHDLL:Optimized Hybrid Deep Learning Model for Classification of Leukaemia images
Binju Saju, V Asha, Neethu Tressa, R Suhas, Shah Yash Jagesh, V. Sandhya · 2023
A serious haematological cancer that causes mortality and morbidity in people of all ages is leukaemia. The haematologists examine bone marrow and blood to determine Leukaemia. Techniques for manual blood testing that have been in use for a long period are frequently slow and produce less reliable diagnoses. The study proposes a hybrid model to classify leukaemia cell images. Initially, images are pre-processed which includes Image resizing, Gray scale conversion and noise removal using a combination of Weiner filter and Median filter. Further, images are enhanced using Gaussian Mixture based histogram equalization. Features of the images are extracted using Dense Convolutional neural network. An optimized ResNet101 model with improved Aquila optimization is used for classification. Performance of the models is compared with other studies in literature. The proposed model has obtained an F1 score of 98.98%, an accuracy of 98.20%, sensitivity of 98.9%, specificity of 98.1%, and precision of 98%.