A DEEP LEARNING FRAMEWORK FOR BLOOD GROUP DETECTION FROM FINGERPRINT PATTERNS

The accurate and timely determination of an individual’s blood group is crucial in medical diagnostics, emergency care, and transfusion medicine. Traditional methods for blood group detection involve invasive procedures such as blood sampling and laboratory testing, which can be time- consuming, resource-intensive, and sometimes impractical in critical or remote scenarios. In recent years, biometric characteristics have gained increasing attention as potential indicators of physiological and genetic traits. Among these, fingerprint patterns have shown correlations with genetic markers, including blood group antigens.This study proposes a novel, non-invasive approach for blood group detection using fingerprint images by leveraging the capabilities of deep learning techniques. The central hypothesis of this research is that distinct blood groups exhibit subtle but consistent variations in fingerprint patterns that can be learned and classified by deep neural networks. A large dataset of fingerprint images labeled with corresponding ABO and Rh blood groups was curated and preprocessed to remove noise and enhance pattern clarity. Multiple convolutional neural network (CNN) architectures, including VGGNet, ResNet, and a custom-designed lightweight CNN, were trained and evaluated on this dataset to classify the fingerprints into one of the eight blood groups (A+, A−, B+, B−, AB+, AB−, O+, O−).