T2-weighted imaging of rectal cancer using a 3D fast spin echo sequence with and without deep learning...

Area of Science:

• Medical Imaging
• Radiology
• Artificial Intelligence in Medicine

Background:

• Rectal magnetic resonance imaging (MRI) is crucial for staging rectal cancer.
• T2-weighted (T2W) 3-dimensional (3D) fast spin echo (FSE) sequences are standard for rectal MRI.
• Deep learning reconstruction (DLR) is an emerging technique to enhance image quality.

Purpose of the Study:

• To compare the image quality and clinical utility of DLR versus conventional reconstruction for T2W 3D FSE rectal MRI.
• To evaluate the diagnostic performance of DLR in characterizing rectal cancer features.

Main Methods:

• A comparative study involving 50 rectal cancer patients undergoing T2W 3D FSE rectal MRI.
• Consecutive data acquisition between July 2020 and January 2021.
• Three radiologists assessed image quality (SNR, artifacts, overall quality) and clinical preference, evaluating specific features like tumor invasion and margins.

Main Results:

• Both DLR and conventional reconstruction yielded moderate to good image quality.
• DLR was preferred over conventional reconstruction in nearly all assessed categories.
• The only exception where DLR was not preferred was for bowel wall layer definition.

Conclusions:

• Both conventional and DLR methods provide acceptable image quality for T2W 3D FSE rectal MRI.
• DLR demonstrates superior clinical preference over conventional reconstruction for rectal cancer imaging.
• DLR shows significant potential to improve diagnostic accuracy in rectal MRI.