Iodinated contrast agent is frequently used in computed tomography (CT) imaging to enhance organ contrast enhancement and improve diagnostic sensitivity. Despite this importance, there currently is a lack of standardization in contrast administration protocol across institutions, leading to many safety and clinical diagnostic risks. To solve this, we built three liver contrast enhancement/perfusion models: two using simple linear regression and another by combining a pre-existing pharmacokinetics mathematical model with clinical data with the eventual goal of individualizing contrast administration protocol to optimize contrast-enhanced CT imaging for each patient. These models primarily use patient attributes, such as height, weight, sex, age and contrast administration information, and bolus tracking information to make such predictions. 418 Chest/Abdomen/Pelvis CT scans were used in this study. 75% of cases were used to train these models and the rest were used to test the prediction accuracy. Pearson’s correlation coefficient test was used to find the correlations between the patient attributes and contrast enhancement in liver parenchyma. Weight, height, BMI, and lean body mass were found to be statistically significant predictors for contrast enhancement (P<0.05), with weight as the strongest predictor. Of the predictive models, we found that including bolus tracking information increases predictive accuracy (r2=0.75 v. 0.42) and that in the absence of bolus tracking information, combining clinical data with pre-existing pharmacokinetics model may provide the needed enhancement curve.
Previous studies have shown that many factors including body habitus, sex, and age of the patient, as well as contrast injection protocol contribute to the variability in contrast-enhanced cross-sectional imaging (i.e., CT). We have previously developed a compartmentalized differential-equation physiology-based pharmacokinetics (PBPK) model incorporated into computational human models (XCAT) to estimate contrast concentration and CT number (HU) enhancement of organs over time. While input to the PBPK model requires certain attributes (height, weight, age, and sex), this still results in a generic prediction as it only cohorts patients into 4 groups. In addition, it does not account for scanning parameters which influence the quality of the image. The PBPK model also requires an estimate of patient’s major organ volumes, not readily-available before a scan, which limits its potential application in prospective personalization of contrast-enhanced protocols. To address these limitations, this study used a machine learning approach to prospectively model contrast dynamics for an organ of interest (liver), given the patient attributes, contrast administration, and imaging parameters. To evaluate its accuracy, we compared the proposed model against the PBPK model. A library of 170 clinical images, with their corresponding patient attributes and contrast and imaging protocols, was used to build the network. The developed network used 70% of the cases for training and validation and the rest for testing. The results indicated a more accurate predictive performance (higher R2), as compared to the PBPK model, in estimating hepatic HU values using patient attributes, scanning parameters, and contrast administration.