Deep Learning-Based Non-Invasive Blood Group Classification Using Thermal Fingerprint Imaging and Gabor Feature Analysis

Abstract

Blood group identification is an essential medical procedure that plays a critical role in blood transfusions, emergency care, and clinical diagnosis. Conventional blood grouping methods require invasive blood sample collection, which may not always be practical in remote or emergency situations. This study presents a non-invasive blood group prediction framework using thermal fingerprint images combined with deep learning techniques. Thermal fingerprint images are processed using Gabor filters to extract discriminative texture and orientation features that capture unique biometric characteristics. These extracted features are then classified by a Convolutional Neural Network (CNN) to predict the ABO and Rh blood group categories. The proposed system incorporates modules for user authentication, model training, prediction, and performance evaluation through accuracy, precision, recall, and F1-score metrics. Experimental results demonstrate a classification accuracy of 97%, indicating the effectiveness of integrating thermal fingerprint biometrics with deep learning. The proposed framework offers a rapid, contactless, and reliable solution for blood group prediction in healthcare applications.