Abstract. Blood cell cancer poses a significant threat to the human body. Cancer a prevalent, multifaceted, and perilous blood disorder, underscores the utmost significance of early detection and treatment. The vital role that blood cells play in the human body allows for their utilization in clinical diagnosis. To predict this condition, various semiautomatic systems have been developed using different medical imaging techniques. This publication conducts a extensive research of the literature regarding the use of vision transformer (ViT) based models in the analysis of blood cell cancer images. It discusses the merits and drawbacks of several ViT-based models, including DBN, CNN, MVT (medical vision transformers), SW-Vit (shifted window vision transformers), MR-ViT (multi-Resolution vision transformers), SNN (spiking neural networks) and MLNN (multilayer neural networks). Furthermore, the review highlights that research on blood cell cancer detection has employed diverse deep learning models across various publicly available datasets. Performance evaluations of these models involved metrics such as accuracy, precision, recall, among others.

Keywords: Vit ViT model Vision Transformer· MVT· MR-ViT·SW-ViT· MLNN· CNN· Accuracy· AUC

| DOI: 10.17148/IMRJR.2025.020602