Dual-phase multiobjective Bayesian optimization method for estimating hepatocellular carcinoma dynamics parameters
Xin Xiong1, Jingchun Huang1, Siming Li1
1Faculty of Information Engineering and Automation, Kunming University of Science and Technology, Yunnan Key Laboratory of Artificial Intelligence, Kunming, China.
Quantitative Imaging in Medicine and Surgery
|August 11, 2025
Summary
Bayesian optimization (BO) methods can assess hepatocellular carcinoma (HCC) pharmacokinetics. The dual-phase and multiobjective strategies into BO (DPMO-BO) method improves diagnostic accuracy and fitting precision for HCC detection.
Area of Science:
- Medical Imaging
- Pharmacokinetics
- Computational Biology
Background:
- Hepatocellular carcinoma (HCC) pharmacokinetics are assessed using optimization algorithms with dynamic PET/CT scans.
- Bayesian optimization (BO) and a dual-phase/multiobjective strategy (DPMO-BO) were evaluated for HCC pharmacokinetic estimation.
Purpose of the Study:
- To assess the role of the BO and DPMO-BO methods in estimating HCC pharmacokinetics from PET/CT scans.
- To compare the diagnostic performance of different pharmacokinetic models in distinguishing HCC.
Main Methods:
- Kinetic parameters were estimated using a double-input three-compartment model from 27 HCC tumors.
- Methods compared include Bayesian method (BM), BO, and DPMO-BO, with root mean square error (RMSE) assessing fitting deviation.
Main Results:
- DPMO-BO significantly differentiated all pharmacokinetic parameters (P<0.05) and showed higher AUC for Ki (0.709) than BO (0.595).
- DPMO-BO achieved lower RMSE values for HCC and normal liver tissues compared to BM and BO.
- BM differentiated HCC using k2, k3, fa, and vb (P<0.05); BO differentiated using fa and vb (P<0.001).
Conclusions:
- The BO method is suitable for assessing HCC pharmacokinetics.
- The DPMO-BO method enhances diagnostic performance and fitting accuracy for HCC detection using PET/CT imaging.
