Hepatic dual-contrast CT imaging: slow triple kVp switching CT with CNN-based sinogram completion and material...

Area of Science:

• Medical Imaging
• Computed Tomography
• Artificial Intelligence

Background:

• Dual-contrast protocols in multi-energy CT show promise for liver lesion detection.
• Photon-counting detectors (PCDs) are ideal for multi-energy CT but are not yet widely available.
• Conventional CT systems require alternative solutions for advanced imaging applications.

Purpose of the Study:

• To investigate an alternative solution for dual-contrast imaging on conventional CT systems.
• To perform multimaterial spectral decomposition for dual-contrast imaging using widely available CT scanners.
• To develop a method for dual-contrast imaging without requiring photon-counting detectors.

Main Methods:

• Proposed a slow x-ray tube voltage switching scheme (3 kVp levels) during gantry rotation.
• Developed a convolutional neural network (CNN) framework for sparse sinogram completion (SC) and material decomposition (MD).
• Conducted simulation studies using an abdominal phantom with liver lesions to evaluate the scheme.

Main Results:

• The SC network achieved acceptable image quality up to a 9-degree switching angle per kVp.
• The MD network successfully differentiated iodine and gadolinium in the sinogram domain.
• Average relative quantification errors for iodine and gadolinium were below 10%.

Conclusions:

• A slow triple kVp switching acquisition scheme and a CNN-based processing pipeline were successfully developed.
• The developed method shows potential for dual-contrast agent protocols on existing single-energy CT systems.
• This approach offers a feasible alternative for advanced liver lesion characterization using conventional CT technology.