The automated examination of blood cells serves as an essential tool for early cancer discovery that delivers improved diagnostic accuracy while minimizing laboratory work. A deep learning framework has been developed by this research to detect and classify cancerous blood cells through analysis of microscopic images. A convolutional neural network (CNN) operation used optimized preprocessing techniques which involved normalization together with data augmentation alongside Gaussian blurring and thresholding to boost feature extraction capabilities. Through max pooling sequences and convolutional layers and complete connection pathways with batch normalization abilities, the system utilizes ReLU activation functions alongside categorical cross-entropy loss algorithms for improved classification accuracy. The assessment on multiple datasets produced outstanding outcomes where the detection accuracy reached 91.2% for identifying four distinct blood cell types. Solutions explicitly address class imbalance and image noise which enables real-time detection while maintaining high precision. The research investigates a cost-effective automatic cancer detection system which targets healthcare facilities with limited resources. Research outcomes enable both biomedical imaging development and artificial intelligence diagnostic research projects in medicine.