Predictive Modeling of Osteonecrosis of the Femoral Head Progression Using MobileNetV3_Large and Long Short-Term...

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Medicine
• Orthopedic Surgery

Background:

• Osteonecrosis of the femoral head (ONFH) assessment is challenging using traditional methods.
• Advanced approaches are needed for accurate and efficient ONFH diagnosis and prediction.
• Deep learning offers a promising solution for enhancing ONFH assessment.

Purpose of the Study:

• To analyze ONFH pathological images using deep learning algorithms.
• To evaluate treatment response, vascular reconstruction, and disease progression in ONFH.
• To identify the optimal deep learning algorithm for precise ONFH assessment and prediction.

Main Methods:

• Magnetic resonance imaging (MRI) data from 30 ONFH patients were analyzed.
• 10 deep learning algorithms were tested, with MobileNetV3_Large identified as optimal.
• MobileNetV3_Large was used for quantifying vascular reconstruction and treatment response, and a LSTM model for dynamic prediction.

Main Results:

• MobileNetV3_Large achieved 96.5% accuracy in ONFH diagnosis, outperforming DenseNet201.
• Vascularized bone grafting increased vascular length by 12.4 mm and branch count by 2.7.
• The model predicted lesion progression with an AUC of 0.92, outperforming ResNet50.

Conclusions:

• Deep learning algorithms offer significant advantages for assessing ONFH treatment response, vascular reconstruction, and disease progression.
• This study provides clinicians with a precise tool for ONFH disease assessment.
• Advanced technological solutions are crucial for improving healthcare practices in ONFH management.