Modeling Patient-Informed Liver Contrast Perfusion in Contrast-enhanced Computed Tomography

Area of Science:

• Radiology
• Medical Imaging
• Computational Imaging

Background:

• Contrast enhancement in liver imaging is crucial for diagnosis.
• Predicting and optimizing contrast enhancement can improve scan quality and consistency.

Purpose of the Study:

• To establish the relationship between patient characteristics and liver parenchyma contrast enhancement.
• To develop a predictive model for contrast enhancement in liver imaging using patient data.

Main Methods:

• Analysis of 418 chest/abdomen/pelvis CT scans with a training-testing split.
• Development of two regression models: Model A (patient attributes + bolus tracking) and Model B (patient attributes only).
• Assessment of predictive accuracy using Pearson correlation coefficient.

Main Results:

• Patient weight and height were significant predictors of contrast enhancement (P < 0.05), with weight being the strongest factor.
• Model A demonstrated higher predictive accuracy (r = 0.75) compared to Model B (r = 0.42).

Conclusions:

• Patient attributes are valuable predictors for modeling liver parenchyma contrast enhancement.
• The developed prediction model can enhance consistency and optimize contrast-enhanced liver imaging.