VCaps-Net: Fine-Tuned VGG16 with Capsule Network for Acute Lymphoblastic Leukemia Detection on a Diverse Dataset

Acute Lymphoblastic Leukemia (ALL) is a highly malignant condition that specifically impacts White Blood Cells (WBC) derived from lymphoid cells. Given its potentially lethal nature, it is crucial to diagnose Acute Lymphoblastic Leukemia (ALL) promptly. Currently, the field of medical sciences is heavily impacted by automated computer-assisted techniques that rely on Artificial Intelligence (AI) and Deep Learning (DL). These approaches are essential tools for clinicians to identify diseases rapidly and reduce their workload. This paper proposed a DL-based novel VCaps-Net framework constructed combining fined-tuned VGG16 with Capsule Network to detect ALL efficiently. The fine-tuned VGG16 features are incorporated into the capsule network to extract enhanced and highly discriminative features. Further, a diverse dataset is proposed by combining our private dataset containing 1000 smear images with a publicly accessible dataset comprising 108 images. The proposed VCaps-Net framework obtained a 98.64% accuracy rate on the diverse dataset. Furthermore, the comparative studies signify the superiority of the proposed VCaps-Net Framework over comparing methodologies.