Combined multi-kernel head computed tomography images optimized for depicting both brain parenchyma and bone

Area of Science:

• Radiology
• Medical Imaging

Background:

• Hybrid convolution kernel technique allows diverse window settings for computed tomography (CT) image sets.
• Artifacts can be a challenge in hybrid convolution kernel CT imaging.

Purpose of the Study:

• To reduce artifacts in the hybrid convolution kernel technique for head CT.
• To evaluate if improved multi-kernel head CT images can replace both brain and bone images for diagnosis.

Main Methods:

• Developed an improved multi-kernel technique for head CT.
• Generated images by combining CT values from bone and brain images based on Hounsfield unit thresholds.
• Included 44 patients with nondisplaced skull fractures.

Main Results:

• Improved multi-kernel images displayed identically to brain images on brain window settings.
• Radiologists confirmed multi-kernel images were sufficient for diagnosing skull fractures on bone window settings in all cases.

Conclusions:

• The improved multi-kernel technique is practical and simple for clinical application.
• Expect simplified head CT examinations and reduced image storage requirements.