Deep Learning-Based Blood Cell Disease Classification: A CNN-Enhanced Approach for Accurate Hematological Diagnosis and Treatment
Gunjan Sharma, Vatsala Anand, Rahul Chauhan, Navin Garg, Sheifali Gupta
Abstract
White blood cells, also known as leukocytes, hold significant importance within the human body as they constitute a vital component of the human immune system. The blood cells possess crucial diagnostic information for numerous diseases and play a vital role in safeguarding the body. Diseases such as leukemia, autoimmune diseases, immunological deficiencies, and hematologic disorders can be detected by evaluating the quantity of white blood cells. In contemporary times, the blood analysis process conducted by medical professionals has been associated with the potential drawbacks of time inefficiency and human immune system errors. A diagnostic system with assistance capabilities can be developed to mitigate these errors, facilitate expedited diagnosis, and alleviate the burden of effort. This study aimed to classify 12,500 white blood cell images, including four distinct white blood cell types: Eosinophil, Lymphocyte, Monocyte, and Neutrophil. A CNN model has been proposed in this research to classify the blood cells. The model has shown a very good accuracy of 96.71% and training accuracy of 98.64% with a minimal loss of 0.22. The overall average precision for all these four classes is 95%. This research can be utilized in the medical field for early diagnosis of any hematological disorder.