RPS 1516 - Advancements in hepatocellular carcinoma (HCC) imaging

Author Block: G. A. Risoleo, M. Ferrari, G. Tripodi, A. Teti, G. Casuscelli, D. Ierace, R. Maccarone, M. Lico, F. Loria; Vibo Valentia/IT
Purpose: CEUS is used in the differential diagnosis between intrahepatic cholangiocarcinoma (ICC) and hepatocellular
carcinoma (HCC). ICC typically shows earlier arterial enhancement and faster wash-out than HCC.
Parametric Micro-Flow Imaging (PMFI) provides further information providing a color-coded map of
vascularization and allow to make a differential diagnosis.
The purpose of this study was to evaluate the diagnostic value of PMFI-CEUS in the differential diagnosis between ICC and HCC.
Methods or Background: We retrospectively reviewed CEUS examinations, recorded on digital data, of 52 patients with histologically
proven ICC, after CEUS the lesions were analyzed using PMFI which produces a color-coded map of
vascularization. During post-processing, the PMFI software generated light-blue for the wash-in phase and
red-yellow for the wash-out phase, providing a detailed view of microbubble dynamics within the lesions.
For each lesion we assessed: wash-in and wash-out timing; PMFI color pattern, and time–intensity curve
(TIC) parameters, including arrival time and time-to-peak. These findings were compared with a control
group of 160 patients affected by HCC.
Results or Findings: ICC showed earlier wash-in (11–22 s) compared with control group of HCC (20–30 s,). Wash-out
occurred between 22–35 s in ICC, while in HCC started after 50–60s. On PMFI-CEUS, ICC were
predominantly blue-light color coded, while HCC were more red-color coded. TIC confirmed
shorter arrival time and time-to-peak in ICC. These findings showed sensitivity 100%, specificity 100% and
accuracy 100% to differentiate ICC from HCC.
Conclusion: PMFI-CEUS provides reproducible parameters for differentiating ICC from HCC. Early wash-in with blue-light
color coded map support ICC, while delayed wash-in with more red-coded PMFI indicates HCC, improving
diagnostic accuracy and clinical decision-making in primary liver tumors.
Limitations: The sample is relative small. A multicentric confirmation is needed.
Funding for this study: No Funding.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: ASP Vibo Valentia

Author Block: S-P. P. Li¹, D. Yin¹, X. Song², Z. Feng², Y. Liu¹, W. Wei¹, L. Liu¹; ¹Hefei/CN, ²Shanghai/CN
Purpose: To investigate the feasibility of microstructure analysis using 5.0T ultra-high-field MRI based on time-dependent diffusion imaging(td-dMRI) in differentiating benign and malignant liver lesions and assessing the pathological grading of HCC
Methods or Background: 51 patients with clinically confirmed focal liver lesions were enrolled. Td-dMRI were acquired using a 5.0T scanner with pulsed and oscillating gradient spin-echo MRI sequences at equivalent diffusion times ranging from 4.0 to 44 msec. The imaging microstructural parameters using limited spectrally edited diffusion (IMPULSED)–based parameters derived from td-dMRI including cell diameter(d), intracellular volume fraction(vin), cellularity, and extracellular diffusivity(Dex). Differences in these parameters between malignant and benign lesions were compared.
Results or Findings: 1. 14 patients had benign lesions and 37 had malignant lesions (including 23 HCC, 5 ICC, and 9 metastases). Benign lesions exhibited lower vin and cellularity but larger cell diameter, with statistically significant differences, and the diagnostic efficacy of cellularity is the highest(AUC=0.714). Dex was higher in benign lesions but not statistically significant.
2. Among 23 HCC cases, 18 were pathologically confirmed, including 7 low-grade (I-II) and 9 high-grade (III-IV) HCC. The high-grade HCC group showed higher vin and cellularity but smaller cell diameter and lower Dex. Statistically significant differences were observed in cellularity and vin between the two groups.
3. When patients were divided into a normal liver group and a hepatitis B virus (HBV)-infected or cirrhotic group, the HBV/cirrhotic group exhibited increased hepatic cellularity and vin, with reduced cell diameter and Dex.
Conclusion: Time-dependent diffusion MRI-based microstructural characterization is a feasible and effective method for differentiating benign and malignant liver lesions and assessing HCC pathological grades, with structural correlations to histopathology.
Limitations: Its potential value in evaluating post-hepatitis B cirrhosis warrants further investigation.
Funding for this study: No
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: Approval number：2022KY-267

Author Block: C. B. Fink, F. Ziayee, J. Böven, K. Jannusch, G. Antoch, P. Minko, E. Tietz; Düsseldorf/DE
Purpose: To assess the prognostic utility of early contrast-enhanced CT after drug-eluting bead transarterial chemoembolization (DEB-TACE) in hepatocellular carcinoma (HCC), focusing on validation of CT as a diagnostic tool and development of a simplified prognostic imaging model compared to established response criteria.
Methods or Background: We retrospectively analyzed 80 HCC patients who underwent DEB-TACE between 2014–2021. Multiphasic CT was performed before and within 7 days after treatment. Two radiologists evaluated morphologic features including arterial phase hyperenhancement (APHE), non-peripheral washout, and capsule enhancement. Regression was defined as disappearance of each feature. Tumor shrinkage was quantified. A Cumulative Regression Score (CRS, 0–3) was derived, and a CT-based Early Response Model (CERM) combining APHE regression and tumor shrinkage was developed. The primary endpoint was progression-free survival (PFS). Survival analysis used Kaplan–Meier and Cox regression; model performance was compared with mRECIST and LI-RADS TR.
Results or Findings: APHE regression was observed in 60.6% of patients and independently predicted longer PFS (HR = 0.49, p = 0.01). was also independently prognostic (HR 0.60; p = 0.03). Washout regression was significant only in univariate analysis; capsule regression was not prognostic. Higher CRS values correlated with stepwise PFS improvement (HR per regressed feature 0.74, p = 0.03). CERM achieved the best prognostic performance (C-index = 0.65), surpassing mRECIST and LI-RADS TR. Internal validation confirmed minimal overfitting and stable calibration.
Conclusion: Early CECT after DEB-TACE is feasible and provides reproducible biomarkers for early PFS prediction. The proposed CERM enables individualized risk stratification within one week, without additional laboratory or functional imaging.
Limitations: Retrospective single-center design; external validation is needed.
Funding for this study: None
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: The institutional ethics committee approved the study (ref. 2023-2630) and waived informed consent.

Author Block: D. Romeo¹, T. Denecke², H-J. Meyer²; ¹Palermo/IT, ²Leipzig/DE
Purpose: Complex associations exist between imaging phenotype and underlying histopathology for hepatocellular carcinoma (HCC). SMARS score discriminates proliferative and non-proliferative HCC in a non-invasive way, associated with treatment outcomes. Systematic validation is needed and it is unclear whether associations between this score with histopathology features exist. This study elucidates correlations between the CT and MRI-defined SMARS score with immunohistochemistry of the pathological specimens in a curatively-treated HCC cohort.
Methods or Background: 44 patients (mean age: 59.6±10.7 years) with histologically confirmed HCC after curative surgical resection were included. Contrast-enhanced MRI and CT were performed before surgery and the SMARS score was calculated. Samples were analyzed for programmed death ligand 1 (PD-L1), Glypican-3, CD3-tumour infiltrating lymphocyte, CD68+ cells, CD34+ microvessel density (MVD).
Results or Findings: The median MRI-derived SMARS score was 1.4 (IQR: -0.32; 2.18) and the CT-derived was − 0.32 (IQR: -1.08; 0.56). According to the proposed threshold, 29 tumours were categorized as proliferative (82.9%) and six as nonproliferative HCC (17.1%) accordingly to the MRI-derived SMARS score. According to the CT-derived SMARS score 24 tumours were categorized as proliferative (61.5%) and 15 as nonproliferative HCC (38.5%). A moderate association was shown between the MRI-derived SMARS score with the Glypican-3 expression (r = 0.37, p = 0.03). CT-derived SMARS score showed correlations with two PD-L1 parameters (PD-L1 tumour positive score r=-0.37, p = 0.02; PD-L1 combined positive score r=-0.35, p = 0.03). No other association with the remaining parameters was detected.
Conclusion: SMARS score is a promising imaging score associated with Glypican-3 and PD-L1 expression in curatively-treated HCC patients. Differences between the CT and MRI-defined score need to be investigated in further trials on larger cohorts.
Limitations: Retrospective single-center design, small patient sample, different time periods between imaging and histopathology.
Funding for this study: None
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: This retrospective study was approved by the institutional ethic committee (University of Leipzig, approval number: 159/25-ek ) and informed consent was waived.

Author Block: A. Mähringer-Kunz¹, L. Müller¹, M-T. Rüßmann¹, I. Schmidtmann¹, M. Moos¹, D. Pinto Dos Santos¹, T. Bäuerle¹, A. Weinmann¹, R. Klöckner²; ¹Mainz/DE, ²Lübeck/DE
Purpose: Extrahepatic metastatic disease (EMD) in patients with hepatocellular carcinoma (HCC) is classified as advanced stage by BCLC. Evidence on prognostic impact of EMD is limited. We investigated the prognostic relevance of EMD — with a special focus on anatomical sites — using a longitudinal approach.
Methods or Background: Patients with HCC treated between 01/2007-12/2021 were included. All cross-sectional imaging was re-evaluated by a radiologist to determine the timing and anatomical sites of EMD. Overall survival (OS) was analysed using a multivariable Cox model with site-specific metastases as time-dependent covariates, adjusting for established risk factors.
Results or Findings: Of 1,563 patients included, 429 (27.4%) had EMD: 190 (12.2%) at diagnosis (synchronous) and 239 (15.3%) during follow-up (metachronous). Median OS was 5.6 months with synchronous EMD vs 20.0 months in those without EMD or with metachronous EMD (p<0.001). The distribution of metastatic sites and corresponding hazard ratios (HR, 95% CI) was: lung (n=183; HR=1.62 (CI=1.32-2.00)), regional lymph nodes (n=168; HR=1.55 (CI=1.27-1.90)), bone (n=102; HR=1.37 (CI=0.84-2.25)), peritoneum (n=101; HR=1.78 (CI=1.40-2.27)), adrenal glands (n=79; HR=1.42 (CI=1.02-1.99)), distant lymph nodes (n=53; HR=2.38 (CI=1.74-3.27)), and soft tissue (n=43; HR=1.79 (CI=1.23-2.60)). Metastases at the sites lung, lymph nodes, peritoneum, adrenal glands, and soft tissue were independent predictors of survival. Each additional site increased the HR by 1.64 (95% CI: 1.54-1.74).
Conclusion: Importantly, the presence of any metastasis is associated with adverse outcomes and an unfavorable hazard ratio. However, bone involvement appears comparatively less detrimental than other forms of extrahepatic spread, whereas distant lymph node metastases are particularly unfavorable. Moreover, prognosis deteriorates progressively with the number of metastatic sites, with each additional site conferring a substantial increase in risk. Thus, regular re-staging with detailed imaging evaluation is crucial.
Limitations: Limitations include the retrospective, single-center design.
Funding for this study: No funding was received for this study.
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information: Approval was not required due to the retrospective nature of the study

Author Block: E. Hoffmann, B. Noto, F. Rennebaum, W. Roll, M. Köhler, M. Masthoff, P. Schindler; Münster/DE
Purpose: The tumor microenvironment (TME) plays a crucial role in the progression and treatment response of hepatocellular carcinoma (HCC). This proof-of-concept study examined whether multiparametric magnetic resonance imaging (mpMRI) could noninvasively characterize the TME and detect early, therapy-induced alterations following transarterial chemoembolization (TACE).
Methods or Background: 18 patients with HCC underwent comprehensive mpMRI at 3T, including diffusion-weighted imaging (DWI), T1 and T2* mapping, as well as dynamic contrast-enhanced (DCE) MRI. 8 patients underwent scans before and after TACE. Quantitative imaging biomarkers were derived to assess perfusion, endothelial permeability, cellular microstructure, hemorrhage and edema in tumorous and nontumorous liver tissue.
Results or Findings: Preliminary quantitative analysis revealed significant differences in TME characteristics between HCC lesions and non-tumor liver tissue. HCC lesions showed disrupted microvascular architecture with increased permeability and altered perfusion [e.g., mean Ktrans (10-³/min) HCC: 796 ± 330, non-tumor: 227 ± 165, p = 0.0036]. These vascular alterations were paralleled by restricted diffusivity [mean ADC (10-³ mm²/s) HCC: 1.30 ± 0.29, non-tumor: 2.23 ± 0.20, p < 0.0001] and prolonged T2* relaxation times [mean T2* (ms) HCC: 39 ± 4, non-tumor: 18 ± 5, p = 0.004], indicating intratumoral hemorrhage, edema and microstructural disorganization. Early post-TACE scans demonstrated that mpMRI can detect therapy-induced remodeling of the TME before changes in size or morphology become apparent.
Conclusion: MpMRI allows for noninvasive characterization of the TME of HCC and demonstrates sensitivity to early treatment-related changes. These findings underscore the potential of mpMRI-derived biomarkers to support personalized therapeutic decisions and improve response evaluation in targeted oncologic therapies.
Limitations: These results are limited by the small preliminary sample size. A prospective clinical trial has recently begun to validate these findings and establish robust imaging biomarkers for MRI-based characterization of the TME of HCC.
Funding for this study: None
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: 2025-207-f-S

Author Block: S. C. Singh¹, C. Xie², Z. Song², Y. Liao²; ¹Lalitpur,Kathmandu/NP, ²Guangzhou/CN
Purpose: This study aims to evaluate the performance of dual-layer spectral CT (DLCT) multi-parameter imaging combined with clinical-radiological features in predicting microvascular invasion (MVI) in hepatocellular carcinoma (HCC), and to develop a logistic regression-based predictive model for clinical decision support.
Methods or Background: This study included 67 patients with pathologically confirmed HCC who underwent preoperative multiphase DLCT between Oct 2022 and Oct 2024. Clinical variables, radiological features and DLCT parameters (electron density, iodine density, MONO-E 40/70/100 keV, Zeff) were collected, and normalized iodine concentration was calculated. Univariate tests compared patients with and without MVI. Variables with clinical relevance and low collinearity (VIF < 5, r < 0.7) entered multivariable logistic regression to build four models (DLCT, clinical, radiological and combined). Model performance was assessed using AUC, sensitivity, specificity, accuracy, predictive values, DeLong’s test, decision curve analysis, and a nomogram.
Results or Findings: Among 67 patients, 21 (31.3%) had MVI. Univariate analysis identified PIVKA-II, tumor size, non-smooth margin, and TTPVI as significant factors, while multivariate analysis retained only TTPVI. Selected DLCT parameters were included for modeling despite non-significance. The combined model demonstrated the highest predictive performance (AUC = 0.880) with balanced diagnostic accuracy (accuracy 0.82, sensitivity 0.76, specificity 0.85). It significantly outperformed the individual models (all p < 0.05), and decision curve analysis indicated clinical benefit. The nomogram provided a straightforward tool for individualized risk estimation.
Conclusion: Combining DLCT multi-parameters with clinical-radiomics features significantly improves preoperative prediction of MVI in HCC, supporting individualized treatment decisions.
Limitations: Single-center study with a small cohort; external validation and comparison with advanced machine learning models are needed.
Funding for this study: Not applicable
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information:

Author Block: P. Chang¹, W. Chiou²; ¹Taichung/TW, ²Taithung/TW
Purpose: To illustrate the clinical applications of photon-counting CT (PCCT) in abdominal interventional radiology (IR) through a series of representative cases, with emphasis on hepatocellular carcinoma (HCC), metastases, and rare hepatic neoplasms.
Methods or Background: We collected abdominal PCCT cases performed at a tertiary referral center between April–August 2025. The cases covered three main clinical scenarios: (1) vascular mapping for TACE and Y-90 treatment; (2) detection of recurrence and follow-up after locoregional therapy or immunotherapy; and (3) evaluation of rare hepatic neoplasms. Each case was analyzed for the incremental value of PCCT compared with conventional energy-integrating detector CT. Specific techniques assessed included virtual monoenergetic imaging (VMI), iodine mapping, virtual non-contrast (VNC), and vessel reconstruction.
Results or Findings: Across the case collection, PCCT consistently provided higher lesion conspicuity, especially at 40–50 keV VMI, allowing better visualization of arterial-phase hyperenhancement and washout in HCC and metastases. Vascular mapping was improved, with clearer identification of segmental/subsegmental feeders, anatomical variants, and arterioportal shunts—crucial for TACE and Y-90 planning. Dose- and contrast-saving protocols achieved reliable imaging with contrast volumes as low as 45 ml. Surveillance cases demonstrated PCCT’s ability to differentiate recurrent HCC from pseudoprogression or ablation changes. Rare tumors such as mucinous cystic neoplasm and IPMN were also better characterized.
Conclusion: PCCT is a promising next-generation imaging tool for safe and personalized interventional strategies in liver cancer and abdominal disease.
Limitations: Single-center retrospective review.

Limited case numbers, especially for rare hepatic neoplasms.

Lack of quantitative outcome correlation with treatment efficacy.
Funding for this study: None
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information:

Author Block: j. pan¹, J. Dai², Q. Wang¹, L. Gao¹; ¹shijiazhuang/CN, ²Shanghai/CN
Purpose: To explore the predictive value of habitat radiomics based on intravoxel incoherent motion (IVIM) for the pathological grading of hepatocellular carcinoma (HCC).
Methods or Background: Clinical and imaging data were retrospectively collected from 83 patients with pathologically confirmed and Edmondson-Steiner graded HCC (96 lesions in total) between May 2018 and June 2025. HCC was categorized as low-grade (Ⅰ-Ⅱ) or high-grade (Ⅲ-Ⅳ). Lesions were randomly split into a training set (n=67) and validation set (n=29) at a 7:3 ratio. All patients underwent preoperative IVIM scans on a 3.0 T MRI system. Based on the k-means clustering results of IVIM parametric maps, the tumor region of interest (ROI) was segmented into multiple subregions, from which radiomic features were subsequently extracted. Univariate/multivariate analyses screened clinical features to construct a logistic regression (LR) clinical model. For habitat features, Spearbvtman correlation and LASSO selected variables used to build multiple classifiers (LR, RF, SVM, k-NN, Bayesian), with the optimal model determined by 5-fold cross-validation. Model performance (AUC, calibration, decision curves) and comparisons (Delong test) were evaluated.
Results or Findings: Univariate and multivariate analyses identified age as an independent predictor for HCC pathological grading. The tumor region was partitioned into four subregions using k-means clustering. After radiomic feature screening, 10 features were retained. Among habitat-based models, the RF model performed best (training set AUC: 0.926, 95% CI: 0.851–0.980; validation set AUC: 0.863, 95% CI: 0.712–0.971). The combined model showed the highest diagnostic performance (training set AUC: 0.950, 95% CI: 0.894–0.990; validation set AUC: 0.926, 95% CI: 0.810–1.000).
Conclusion: IVIM-based habitat radiomics has favorable predictive value for HCC pathological grading. The combined model further improves performance, providing a novel methodological reference for preoperative non-invasive HCC grading.
Limitations: Not applicable.
Funding for this study: No funding was provided for this study.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: The ethics committee notification can be found under the number 2021KY121

Author Block: C. Liu¹, W. Liang¹, J. Meng¹, X. Liu², D. Liu¹; ¹Wuzhou/CN, ²Guangzhou/CN
Purpose: To investigate the diagnostic value of spectral CT multi-parameter imaging in detecting residual tumor activity in hepatocellular carcinoma (HCC) lesions after transarterial chemoembolization (TACE).
Methods or Background: Thirty-nine HCC patients with 43 lesions (26 with lipiodol deposition, 17 with residual tumor) who underwent TACE were retrospectively analyzed. Pathology or digital subtraction angiography (DSA) served as the reference standard. All patients received spectral CT plain and four-phase enhanced scans. Quantitative parameters, including conventional CT, virtual mono-energetic images (VMIs at 40/70 keV), iodine concentration (IC), normalized iodine concentration (NIC), effective atomic number (Zeff), spectral slope (λHU), electron density (ED), arterial enhancement fraction (AEF), and extracellular volume fraction (ECV) were measured in the early arterial (EAP), late arterial (LAP), venous (VP), and parenchymal (PP) phases. ROC analysis was performed to assess diagnostic performance.
Results or Findings: Significant differences between lipiodol deposition and residual tumor were observed in Zeff, λHU, IC, and NIC in EAP; Zeff, VMI 40 keV, λHU, IC, and NIC in LAP; and ED in PP (all P < 0.05). AEF(EAP/VP), AEF-(LAP/VP), AEF-(EAP/PP), AEF-(LAP/PP), and ECV showed no significant differences. The best single-parameter performance was obtained with Zeff and λHU in LAP (cut-off values: 8.24 and 3.44; AUC = 0.725 and 0.724). The combined multi-parameter model achieved excellent diagnostic accuracy (AUC = 0.953, 95% CI: 0.896–1.00, P < 0.001), with 100% sensitivity and 76% specificity.
Conclusion: Spectral CT multi-parameter imaging enables comprehensive and quantitative evaluation of HCC lesions after TACE, overcoming the limitations of conventional imaging obscured by lipiodol deposition. This approach may facilitate early detection of tumor recurrence and guide personalized treatment strategies.
Limitations: None
Funding for this study: None
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: None