Background: MRI of the temporomandibular joint has now become the gold standard for diagnosing temporomandibular joint-related diseases and the presence of joint effusion. The scanning time of MRI examination of temporomandibular joint is long, especially when scanning in the open-mouth position, the patient cannot maintain the open-mouth state for a long time, which is prone to uncomfortable movements of the head, face and temporomandibular joint, resulting in motion artifacts in the scanned image and failure of the examination. Deep learning reconstruction（DLR）has recently been applied to MRI, which can reduce image noise, improve image quality, and shorten scan time. Most DLR research focuses on large joints such as hip and shoulder joints and spine MRI(1-5). Currently, there are no studies that apply DLR to MRI of the temporomandibular joint, which has a smaller joint structure and is more affected by motion artifacts.
Purpose: To explore the application value of DLR in improving the quality of MRI imaging of the temporomandibular joint and shortening the scanning time.
Materials and Methods: 40 volunteers who agreed to undergo temporomandibular joint scanning from March 10, 2023 to April 30, 2023 were prospectively included, and each volunteer underwent temporomandibular joint MRI PDWI scan and DLR-PDWI scan. Two radiologists conducted qualitative and quantitative evaluations of the image quality of the two groups respectively. Qualitative evaluation: Likert scale (5 points scale) was used to qualitatively score the image anatomical structure clarity and overall image quality. Quantitative evaluation uses signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) to quantitatively evaluate image quality. One-way analysis of variance and Friedman test were performed on normally distributed and non-normally distributed data respectively. The intra-class correlation coefficient (ICC) was used to compare the consistency of the subjective ratings of the two readers.
Results: Compared with the PDWI group, the scanning time of the DLR-PDWI group was shortened by 67%. The agreement between the two radiologists on the subjective ratings of image anatomical structure clarity and overall image quality was good (ICCs were 0.80 and 0.78, respectively). The scoring results of the two doctors showed that there were significant differences in the image anatomical structure clarity and overall image quality scores between the PDWI group and the DLR-PDWI group (P<0.05); The difference in SNR and CNR between the two groups of images is statistically significant (P<0.05); the qualitative and quantitative evaluation results of the DLR-PDWI group were significantly better than those of the PDWI group.
Conclusion: DLR can significantly shorten scanning time and improve image quality.

Acknowledgements

No acknowledgement found.