A comprehensive deep learning approach to improve enchondroma detection on X-ray images

Area of Science:

• Orthopedic Oncology
• Medical Imaging Analysis
• Artificial Intelligence in Medicine

Background:

• Enchondromas are benign hyaline cartilage neoplasms originating in the medullary cavity.
• Clinical presentation varies based on lesion size, location, and radiographic features.
• Accurate and timely diagnosis is crucial for patient management.

Purpose of the Study:

• To develop and validate an AI-driven object detection system for identifying enchondromas in bone X-rays.
• To provide preliminary data and innovative diagnostic approaches for orthopedic oncology specialists.
• To assess the performance of advanced deep learning algorithms in detecting bone neoplasms.

Main Methods:

• Acquisition and preprocessing of authentic patient X-ray radiographs under ethical approval.
• Annotation of images by orthopedic oncology specialists.
• Training and optimization of diverse deep learning architectures for object detection.
• Rigorous cross-validation and oversight of all processing steps.

Main Results:

• Successful identification of enchondroma formation in bone tissue using object detection algorithms.
• Achieved an average precision of 0.97 and an accuracy rate of 0.98.
• Results were corroborated by medical professionals, indicating high reliability.

Conclusions:

• The developed AI system demonstrates high efficacy in detecting enchondromas from bone radiography.
• This approach offers a novel and efficient method for preliminary insights to specialists.
• A larger comprehensive study with 1055 patient data is planned to further validate these findings.