Automated Pediatric Bone Age Assessment Using Convolutional Neural Networks

Pediatric medicine widely uses bone age determination to assess skeletal maturity and identify developmental disorders early. However, manual assessment methods are subjective and lack consistency. To address this, we suggest using image preprocessing to isolate vital areas in hand X-rays and enhance features. We then enhance the Inception-V4 model to extract features from these images, integrating gender as a crucial reference. Our model, validated on a large dataset, demonstrates superior bone age prediction compared to prior methods. These automated models offer precise and reliable tools for clinical assessments, showing significant potential for practical application.