Visual Cascaded-Progressive Convolutional Neural Network (C-PCNN) for Diagnosis of Meniscus Injury

Area of Science:

• Medical Imaging Analysis
• Artificial Intelligence in Healthcare
• Orthopedic Diagnostics

Background:

• Meniscus injuries are common knee ailments requiring accurate and timely diagnosis.
• Magnetic resonance imaging (MRI) is a key modality for visualizing knee structures, including the meniscus.
• Current diagnostic methods can be time-consuming and may benefit from AI-assisted tools.

Purpose of the Study:

• To develop a novel automatic convolutional neural network (CNN) for meniscus injury diagnosis using MRI.
• To enable visualization of meniscus lesion characteristics.
• To enhance diagnostic accuracy and reduce diagnosis times for knee meniscus injuries.

Main Methods:

• A cascaded-progressive convolutional neural network (C-PCNN) was developed for meniscus injury diagnosis from MRI.
• The C-PCNN was trained and tested on 1396 hospital-collected images using 5-fold cross-validation.
• Performance was evaluated using accuracy, sensitivity, specificity, ROC, and compared with physician diagnoses.

Main Results:

• The C-PCNN achieved 85.6% accuracy for anterior horn and 92% for posterior horn meniscus injury diagnosis.
• The overall average accuracy of the C-PCNN was 89.8% with an AUC of 0.86.
• Physician diagnostic accuracy improved with C-PCNN assistance, becoming comparable to chief physicians.

Conclusions:

• The C-PCNN-based MRI technique demonstrates significant clinical utility for diagnosing knee meniscus injuries.
• This AI tool offers high accuracy, potentially increasing diagnostic speed and reducing errors for clinicians.
• The C-PCNN serves as a valuable assistant for clinical physicians in meniscus injury diagnosis.