Clinical Study of Magnetic Resonance Imaging and Deep Learning of Joint Synovial Disease

Using Magnetic Resonance Imaging and DL Methods to Explore the Diagnosis and Clinical Prognosis of Joint Synovitis.

Through the high-throughput feature extraction of magnetic resonance images, the deep learning prediction model of joint synovial lesions is constructed used for the diagnosis, differential diagnosis and curative effect monitoring of joint synovial lesions.

The study applies magnetic resonance and deep learning (DL) to the diagnosis of joint synovial lesions, aims to have a more comprehensive understanding of the pathophysiology of the occurrence and development of joint synovial lesions. As a non-invasive imaging method to assess the condition of the disease, DL methods excavates the deep features contained in the image, quantifies the joint synovial lesions, and then gives more information to the clinician in the diagnosis and differential diagnosis of the joint synovial lesions, provide important information for the planning of individualized treatment plans for patients with joint synovial diseases.

Inclusion Criteria:

1. Patients diagnosed with joint synovial disease through radiological examination, arthroscopy or pathological biopsy of the joint, or whose clinical manifestations meet the diagnostic criteria of the American College of Rheumatology (ACR) for joint synovial disease.
2. Patients received pre-treatment MR.

Exclusion Criteria:

1. Patients who have received surgery, medication or other systemic treatment before standardized MRI scan.
2. Poor image quality.
3. Articular hemorrhage.