Synthesizing Contrast-enhanced Computed Tomography Images with an Improved Conditional Generative Adversarial...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Radiology

Background:

• Contrast-enhanced computed tomography (CE-CT) is crucial for liver cancer diagnosis.
• CE-CT requires contrast injection, increasing patient burden.
• Non-contrast CT (NC-CT) lacks detailed enhancement information.

Purpose of the Study:

• To develop a method for generating CE-CT images from NC-CT images.
• To reduce the physical burden on patients by avoiding contrast agents.
• To improve the diagnostic accuracy of liver cancer detection using AI.

Main Methods:

• An improved conditional generative adversarial network (improved cGAN) was proposed.
• The generator incorporated pyramid pooling and feature fusion modules.
• The model was trained and evaluated on a dataset of NC-CT and CE-CT images (arterial and portal venous phases).

Main Results:

• The proposed improved cGAN effectively generated CE-CT images from NC-CT images.
• The model demonstrated superior performance compared to existing GAN-based methods.
• Quantitative and qualitative evaluations confirmed the method's effectiveness.

Conclusions:

• The improved cGAN offers a promising approach for synthesizing CE-CT images.
• This AI-driven method can potentially reduce patient discomfort and healthcare costs.
• The technique holds potential for enhancing liver cancer diagnosis and management.