Deep Learning Based Microscopic Blood Cell Classification For Cancer Detection

This study explores a deep learning approach for classifying blood cell images into benign and malignant subtypes to aid in cancer detection. The classification process utilized the MobileNetV2 model, which was applied to a dataset comprising four categories: Benign, Malignant Pre-B, Malignant Pro-B, and Malignant Early Pre-B. The dataset underwent a preprocessing stage that included real-time data augmentation through image generators to enhance model performance and increase training data diversity.A prediction function was integrated to classify blood cell images by utilizing the trained MobileNetV2 model. The function processes preloaded images after applying the necessary preprocessing steps and outputs the predicted class label. The results demonstrate the model’s applicability for real-time, point-of-care cancer diagnosis. This approach highlights the effectiveness of transfer learning with MobileNetV2 for accurate blood cell classification and shows potential for developing automated cancer detection systems in clinical environments.