Deep Learning Based Web Application for Prediction of Bone Age Range and Gender

The detection of age and gender from palm radio-graphs involves analysing the image to estimate skeletal age and determine gender based on the skeletal structures present in the palm radiograph. Here, the task is to deploy a deep learning (DL)-based web application for age and gender identification from radiography images, which are vital for pediatric medicine to diagnose growth problems and determine the appropriate intervention. Four deep learning models, InceptionV3, MobileNetV3, ResNet50, and VGG19 implemented for classifying the bone age into age categories of 0 to 5 years, 6 to 10 years, 11 to 15 years, and older than 16 years. Evaluating with performance parameters like accuracy, precision, recall and F1-score, Inception V3 proved to be the most effective in achieving the desired outcome. The model achieved an accuracy of 98.79% accuracy with an average 0.9 F1-score and AUC of 0.99. Hence, the model was further used to train for gender prediction. Further, the web deployment of the model was done on the Streamlit system for the real-time prediction of bone age and gender of the palm radiograph image, the ultimate goal of the DL implementation.