Knee Bone and Cartilage Segmentation Based on a 3D Deep Neural Network Using Adversarial Loss for Prior Shape...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Orthopedics

Background:

• Accurate knee bone and cartilage segmentation on MRI is crucial for orthopedic applications.
• Manual segmentation is time-consuming and prone to observer variability.
• Automated methods are needed to improve efficiency and consistency.

Purpose of the Study:

• To develop and evaluate a 3D deep neural network for automated segmentation of knee bone and cartilage in MRI.
• To enhance segmentation accuracy using adversarial loss and a restoration network.
• To compare the proposed method against a baseline U-Net network.

Main Methods:

• A 3D deep neural network with adversarial loss was employed for bone segmentation.
• A restoration network was utilized to refine bone segmentation accuracy.
• A U-Net-like network was used for cartilage segmentation.
• Combined bone and cartilage segmentations were achieved through post-processing.

Main Results:

• The proposed method achieved a total score higher than 76 on the SKI10 validation dataset.
• A fine-tuned U-Net-like network demonstrated state-of-the-art results without post-processing.
• The method proved robust in segmenting bone and cartilage, including in pathological cases.

Conclusions:

• The developed 3D deep neural network offers a robust and accurate solution for knee bone and cartilage segmentation from MRI.
• The approach has significant potential for improving patient-specific diagnosis and surgical planning in orthopedics.
• Automated segmentation significantly reduces time and variability compared to manual methods.