Skeletal age evaluation using hand X-rays to determine growth problems.

Area of Science:

• Pediatric Radiology
• Medical Imaging Analysis
• Artificial Intelligence in Healthcare

Background:

• Skeletal age estimation from X-rays is vital for identifying infant and newborn bone growth anomalies.
• Bone abnormalities can stem from various conditions, impacting growth plates and potentially causing permanent joint damage.
• Discrepancies between chronological and skeletal age signal potential growth problems, necessitating accurate assessment for early diagnosis.

Purpose of the Study:

• To develop an automated system for accurate skeletal age estimation using pediatric hand X-rays.
• To evaluate the efficacy of a customized convolutional neural network (CNN) for detecting hand bone maturation.
• To compare the performance of the proposed CNN model against the Visual Geometry Group (VGG) model.

Main Methods:

• Utilized the Radiological Society of North America's Pediatric Bone Age Challenge dataset (12,600 images).
• Developed a customized convolutional neural network (CNN) model for bone age assessment.
• Employed data augmentation techniques to enhance dataset size and model training.

Main Results:

• The customized CNN model achieved 97% accuracy in skeletal age estimation.
• The proposed CNN model demonstrated superior performance compared to the VGG model.
• Data augmentation techniques positively impacted the model's training and accuracy.

Conclusions:

• The developed customized CNN model offers a highly accurate and efficient method for skeletal age estimation.
• This AI-driven approach can aid clinicians in diagnosing growth anomalies and endocrine disorders in children.
• The findings highlight the potential of deep learning in advancing pediatric radiological assessments.