Three-dimensional conditional generative adversarial network-based virtual thin-slice technique for the morphological
Atsushi Nakamoto1, Masatoshi Hori2, Hiromitsu Onishi3
1Department of Diagnostic and Interventional Radiology, Osaka University Graduate School of Medicine, 2-2, Yamadaoka, Suita, Osaka, 565-0871, Japan. a-nakamoto@radiol.med.osaka-u.ac.jp.
Scientific Reports
|July 16, 2022
Summary
Virtual thin-slice (VTS) technique improves intervertebral space visibility and vertebral height measurement accuracy in CT scans. However, VTS is not recommended for diagnosing spinal compression fractures.
Area of Science:
- Radiology
- Medical Imaging
- Artificial Intelligence in Medicine
Background:
- Computed tomography (CT) is crucial for spinal assessment.
- Thicker CT slices can obscure fine details like intervertebral spaces.
- Generative adversarial networks offer potential for image enhancement.
Purpose of the Study:
- To evaluate the performance of Virtual Thin-Slice (VTS) technology for spinal CT imaging.
- To compare VTS-generated images with standard thick-slice images for anatomical detail and diagnostic accuracy.
Main Methods:
- VTS algorithm applied to 4-mm thick spinal CT images from 73 patients.
- Assessed visibility of intervertebral spaces on 4-mm and VTS images.
- Compared vertebral height measurements between VTS, 4-mm, and gold-standard 1-mm slices.
- Evaluated diagnostic performance for compression fracture detection.
Main Results:
- Intervertebral spaces were significantly more visible on VTS images (P < 0.001).
- VTS images showed smaller differences in vertebral height measurements compared to 1-mm slices than 4-mm slices.
- Diagnostic performance for compression fractures was lower with VTS for one reader (P = 0.02).
Conclusions:
- VTS technology enhances visualization of intervertebral spaces and enables accurate vertebral height measurement.
- Despite improvements in anatomical detail, VTS is not suitable for diagnosing compression fractures.
