Kevyn Alifian Hernanda Wibowo, Jumanto Unjung
Acute Lymphoblastic Leukemia (ALL) is a type of blood cancer characterized by an increased number of abnormal lymphoblasts in the bone marrow. Early detection of ALL is very important to increase the patient's chance of recovery. One method that can be used to help the diagnosis process is the classification of white blood cell images using deep learning-based technology. This research aims to optimize the performance of InceptionV3 model in classifying ALL by applying Convolutional Block Attention Module (CBAM) layer and Generative Adversarial Network (GAN) method. The CBAM layer is used to improve the focus of the model on important features in the image. Meanwhile, the GAN method is used to overcome the problem of data imbalance by generating synthetic data that resembles the original data, thus improving the generalization of the model. The model was developed through several model building scenarios, and the best results were obtained in the fifth scenario, which is the InceptionV3 model optimized with the CBAM layer and the GAN method. The results showed that the combination of the InceptionV3 model, CBAM layer, and GAN method achieved an accuracy of 90.09% and significantly improved the ALL classification performance compared to other model scenarios. Thus, the proposed method is expected to contribute to the development of science related to ALL classification based on white blood cell images and assist patients and health workers in streamlining the ALL classification process. © 2025 IEEE.
Universitas Negeri Semarang, Department of Computer Science, Semarang, Indonesia
