Objective:

To evaluate the utility of deep learning-based image enhancement for improving the image quality and diagnostic performance of 3D contrast-enhanced T1-weighted black blood MR imaging for brain metastases, thereby enhancing clinical decision-making.

Key Findings:

• Deep learning enhancement improved image quality metrics (SNR and CNR) in black blood MR images, with specific percentage improvements noted.
• Thin-slice imaging with deep learning enhancement showed better diagnostic performance for detecting small brain metastases.
• Black blood imaging provides superior contrast between vessels and metastatic lesions compared to conventional imaging.

Interpretation:

The application of deep learning in black blood MR imaging enhances the detection of brain metastases, addressing the limitations of thin-slice imaging related to noise and resolution, with significant implications for clinical practice.

Limitations:

• Retrospective study design may introduce selection bias.
• The study was limited to a specific patient population with underlying malignancies.
• Potential variability in image quality due to different MRI scanner manufacturers and confounding factors.

Conclusion:

Deep learning-based image enhancement significantly improves the quality of black blood MR imaging, facilitating better detection and assessment of brain metastases, which may lead to improved patient outcomes.