Related Experiment Video
Updated: Oct 17, 2025
Four-Dimensional CT Analysis Using Sequential 3D-3D Registration
Published on: November 23, 2019
Synthesizing Quantitative T2 Maps in Right Lateral Knee Femoral Condyles from Multicontrast Anatomic Data with a
Bragi Sveinsson1, Akshay S Chaudhari1, Bo Zhu1
1Athinoula A. Martinos Center for Biomedical Imaging, Department of Radiology, Massachusetts General Hospital, Harvard Medical School, 149 13th St, Suite 2301, Boston, MA 02129 (B.S., B.Z., N.K., M.S.R.); Division of Musculoskeletal Imaging and Intervention, Department of Radiology, Massachusetts General Hospital, Harvard Medical School, Boston, Mass (M.T.); Department of Radiology, Stanford University, Stanford, Calif (A.S.C., G.E.G.); and Department of Physics, Harvard University, Cambridge, Mass (M.S.R.).
This study developed a convolutional neural network (CNN) using a conditional generative adversarial network (cGAN) to create T2 maps for knee cartilage from MRI scans. The synthesized T2 maps showed good agreement with traditional MESE scans, demonstrating feasibility.
Area of Science:
- Biomedical Imaging
- Artificial Intelligence in Medicine
- Quantitative MRI
Background:
- T2 mapping is crucial for assessing articular cartilage health.
- Traditional T2 mapping methods can be time-consuming.
- Developing faster, accurate methods is essential for clinical applications.
Purpose of the Study:
- To create a proof-of-concept convolutional neural network (CNN) to synthesize T2 maps.
- Utilize a conditional generative adversarial network (cGAN) for synthesis.
- Generate T2 maps from standard anatomic MRI sequences for the right lateral femoral condyle.
Main Methods:
- Retrospective analysis of 4621 patients from the Osteoarthritis Initiative (2004-2006).
- Input: Anatomic MR images (turbo spin-echo, double-echo in steady-state).
- Output: Predicted T2 maps using a cGAN-based CNN; validation with linear regression and radiologist assessment.
Main Results:
- CNN T2 values demonstrated correlation with MESE T2 values (slopes 0.55-0.83).
- Radiologists showed moderate accuracy in distinguishing CNN T2 from MESE T2.
- Feasible synthesis of T2 maps with good agreement to MESE scans.
Conclusions:
- A neural network-based cGAN approach can synthesize T2 maps in femoral cartilage.
- Synthesized T2 maps from anatomic MRI show good agreement with MESE scans.
- This method offers a potential advancement for cartilage imaging and quantification.
More Related Videos
06:28Anterior Cruciate Ligament Transection and Synovial Fluid Lavage in a Rodent Model to Study Joint Inflammation and Posttraumatic Osteoarthritis
Published on: September 2, 2025
07:43In Vivo Quantification of Hip Arthrokinematics during Dynamic Weight-bearing Activities using Dual Fluoroscopy
Published on: July 2, 2021