Blood Cancer Detection Leveraging Deep Learning Fusion Techniques: An Optimized Model Based on VGG16 and AlexNet

A severe and life-threatening illness, blood cancer calls for early discovery and accurate diagnosis to maximize patient outcomes. This work presents a hybrid deep learning model integrated via a late fusion technique using the strengths of two well-known convolutional neural networks models VGG16, and AlexNet architecture. Combining the deep feature extraction powers of the VGG16 network with the AlexNet-inspired model’s capacity to capture varying spatial hierarchies, the proposed model produces a thorough and quite accurate classification of many blood cell types. Combining the two models helps to produce a single feature representation, which is then passed via fully connected layers to do multi-class classification. The Adam optimizer helps to maximize the system; training runs over 100 epochs using early stopping to prevent overfitting. With notable gains in F1-scores, recall, and precision across several blood cell categories, The experimental results indicate that the suggested model attains an accuracy of ${9 8 \%}$. This model emphasizes the possibilities of hybrid deep learning models in medical diagnostics, especially for early blood cancer detection where timely intervention is essential for patient survival. This model might be very helpful to medical professionals, particularly in areas with limited access to specialist healthcare services, by providing a strong and automated tool for blood cancer detection.