“Early Diagnosis of Blood Cancer Using Deep Learning: An Evaluation of CNN, SVM, and Random Forest”

Abstract

Blood cancer is one of the most dangerous and fast-growing types of cancer, affecting people of all ages, especially children. Early diagnosis is critical for effective treatment and improving survival rates. Traditional methods of detecting blood cancer—such as bone marrow tests and manual image analysis—are time-consuming, painful, and not always accurate. This research focuses on using artificial intelligence (AI), particularly deep learning and machine learning techniques, to improve early detection of blood cancer through microscopic blood smear images.

We evaluate and compare the performance of three widely used models: Convolutional Neural Networks (CNN), Support Vector Machines (SVM), and Random Forest (RF). Our approach includes image preprocessing, feature extraction, and classification. CNNs were found to be the most effective, achieving up to 99.12% accuracy using ensemble models like DenseNet, Inception, and Xception. SVM and Random Forest also delivered strong results, especially when used with deep feature extraction techniques.

The findings of this study show that AI-powered models can significantly improve the speed and accuracy of blood cancer diagnosis. This research not only supports the global healthcare goal of early cancer detection but also opens the door for developing smart, low-cost diagnostic tools that can be used in real-world clinical settings. Future work will focus on real- time applications, larger datasets, explainable AI, and integration into healthcare systems.