5.0T MRI Deep Learning Reconstruction for Nasopharyngeal Carcinoma: Comparison of Image Quality and Diagnostic Efficacy

Background: Currently, magnetic resonance imaging (MRI) is the preferred method for diagnosing and staging nasopharyngeal carcinoma (NPC). The advent of 5.0T MRI presents both opportunities and challenges for magnetic resonance imaging. This study aimed to investigate the effects of DeepRecon technology on image quality and diagnostic efficacy of 5.0T MRI in NPC. Methods: A total of 70 nasopharyngeal carcinoma (NPC) patients underwent 5T MRI scans. The scanning protocol included axial T2-weighted imaging (T2WI), axial T1-weighted imaging (T1WI), axial contrast-enhanced T1WI imaging, and coronal contrast-enhanced T1WI imaging. Images from four sequences were reconstructed at different intensity levels (0-5) using DeepRecon technology, totaling 24 sets. Two physicians evaluated visibility of lesions, boundary sharpness, artifact presence, and overall image quality using a 5-point Likert scale. T-stage evaluation of NPC was performed for both conventional (level=0) and DeepRecon images to compare diagnostic accuracy. Signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) for lesions and lateral pterygoid muscles were also assessed. Results: DeepRecon images demonstrated significantly higher SNR in lesions and lateral pterygoid muscles compared to conventional images (P<0.001), with a maximum improvement of 94%. In axial T2WI, axial contrast-enhanced T1WI, and coronal contrast-enhanced T1WI DeepRecon images, the lesion-to-lateral pterygoid muscle CNR was significantly higher (P<0.001), with a maximum improvement of 108%. Qualitative analysis revealed that DeepRecon images (levels 2-5) were superior to conventional images (P<0.001) across all subjective assessment except artifact reduction. Among reconstruction levels in DeepRecon, level 3 yielded the highest overall image quality score. Additionally, the image T-stage results for DeepRecon images (level 3) and conventional images exhibited a comparable degree of consistency with the clinical T-stage results (κ = 0.81 and 0.78, respectively). Conclusion: DeepRecon technology improves the quality of conventional nasopharyngeal MRI images without effects on T-stage assessment, which contributes to the enhancement of the diagnostic values.