Computer-Aided Diagnosis for Hepatocellular Carcinoma Microvascular Invasion

Hepatocellular carcinoma (HCC) is a common malignancy in China with a high mortality rate. Early recurrence and long-term prognosis are closely linked to tumor aggressiveness. Microvascular invasion (MVI), defined as the presence of tumor cells within small branches of the portal or hepatic veins, is a critical marker of malignant biological behavior. Clinically, MVI is strongly associated with early postoperative recurrence and serves as an important reference for determining surgical margin extension, adjuvant treatment, and postoperative management strategies. At present, definitive diagnosis of MVI still relies on postoperative pathological examination, and reliable preoperative assessment methods are lacking. Although prior studies have attempted to predict MVI using preoperative imaging, their clinical application remains limited by poor generalizability, weak interpretability, and insufficient cross-center adaptability.

This study aims to develop a high-performance, generalizable, and interpretable computer-aided diagnostic (CAD) system for preoperative prediction of HCC-MVI using multiphase CT imaging, artificial intelligence techniques, and clinical prior knowledge. The system will be evaluated prospectively in an observational, multicenter clinical study to assess its diagnostic value and clinical applicability.

The CAD system integrates three categories of imaging features: (1) high-level representations automatically extracted by deep neural networks; (2) predefined radiomics features such as tumor morphology, texture, and intensity distributions; and (3) structured prior features derived from radiological expertise, including tumor margin blurriness and spatial relationships with adjacent portal veins. Sparse constraints and redundancy suppression mechanisms will be applied to identify stable and efficient MVI-related representations. In addition, the system adopts a spatial domain strategy covering tumor, peritumoral, and distant regions, in order to capture invasion patterns from both local morphology and microenvironmental context, thereby constructing reproducible and clinically interpretable imaging biomarkers.

To overcome the limitations of single-domain models, the system employs a multi-source heterogeneous fusion strategy that integrates morphological-textural features, dynamic enhancement patterns, and spatial graph structures. The model architecture combines convolutional neural networks (CNNs) to capture fine-grained textures, Transformer modules to model long-range dependencies, and graph neural networks (GNNs) to represent tumor-vascular topological relationships. This hybrid approach enables comprehensive understanding of both local details and global structures. Furthermore, the model incorporates uncertainty quantification and attention-like mechanisms to dynamically adjust prediction confidence and generate saliency heatmaps. These outputs are designed to enhance clinicians' interpretability and trust in the system. An interactive visualization interface will also be developed to support risk interpretation and surgical planning.

The study will conduct a prospective observational validation across multiple clinical centers, with unified inclusion/exclusion criteria and standardized data collection protocols. Model predictions will be blindly compared against postoperative pathological results. In addition to conventional metrics (accuracy, sensitivity, specificity, and AUC), the study will observationally evaluate the impact of model-based predictions on preoperative risk stratification and surgical decision-making. By testing the system across diverse patient populations, the study aims to confirm its generalizability, clinical utility, and potential for real-world translation of intelligent diagnostic technologies.

• Computer-Aided Diagnosis System for Preoperative Prediction of MVI in HCC (Diagnostic test): This intervention is an artificial intelligence-based computer-aided diagnosis (CAD) system developed to predict microvascular invasion (MVI) in patients with hepatocellular carcinoma using preoperative multiphase CT imaging. The system integrates deep learning, radiomics, and expert-defined imaging features to provide risk assessment and visualization of MVI prior to surgery. In this study, the CAD system will be evaluated retrospectively and prospectively in an observational manner only. The results will not influence clinical decision-making or patient management, and all treatments will follow standard of care.

Primary outcomes

• Area Under the Receiver Operating Characteristic Curve (AUC) [Within 1 month after surgery]: The AUC will be calculated by comparing CAD system predictions with the reference standard of postoperative pathological diagnosis of microvascular invasion in hepatocellular carcinoma.
• Accuracy [Within 1 month after surgery]: Accuracy will be defined as the proportion of correctly classified cases (both MVI-positive and MVI-negative) by the CAD system compared with postoperative pathology.

Secondary outcomes

• Sensitivity [Within 1 month after surgery]: Sensitivity will be calculated as the proportion of true positive MVI cases correctly identified by the CAD system compared with postoperative pathology.
• Specificity [Within 1 month after surgery]: Specificity will be calculated as the proportion of true negative MVI cases correctly identified by the CAD system compared with postoperative pathology.
• Calibration [Within 1 month after surgery]: Calibration performance will be assessed using calibration curves, Hosmer-Lemeshow goodness-of-fit tests, and Brier scores, to determine agreement between predicted probabilities and observed MVI outcomes.

Other outcomes

• Processing Time [Within 1 month after surgery]: Average computational time required for the CAD system to generate predictions and visualization outputs will be recorded to assess feasibility for integration into clinical workflow.
• Physician Confidence Score [Within 1 month after surgery]: Physician Confidence Score will be measured using a questionnaire that asks physicians to rate their confidence in assessing patient risk after reviewing CAD-generated probability scores and saliency maps. Responses will be collected on a 5-point Likert scale (1 = not confident at all, 2 = slightly confident, 3 = moderately confident, 4 = confident, 5 = very confident). The score will be recorded as the numerical Likert scale value selected by each physician.

Sex: All, ages: 18 Years to 18 Years

Inclusion Criteria:

* Age ≥ 18 years.
* Confirmed diagnosis of hepatocellular carcinoma (HCC) according to the Chinese Clinical Practice Guidelines for Primary Liver Cancer.
* Eligible for surgical intervention (hepatic resection or liver transplantation) according to the Chinese Clinical Practice Guidelines for Cancer, including stages Ia, Ib, and IIa.
* Preoperative imaging examination performed within 1 month before surgery.
* Availability of histopathological evaluation with documented microvascular invasion (MVI) status.

Exclusion Criteria:

* History of prior antitumor treatment, including preoperative surgical intervention, transarterial chemoembolization (TACE), radiofrequency ablation (RFA), systemic therapy, or any other preoperative intervention.
* Presence of major vascular invasion, bile duct invasion/thrombosis, extrahepatic metastasis, or lymph node involvement.
* Diffuse hepatocellular carcinoma or tumor rupture with hemorrhage.
* Lack of key data required for primary analysis.
* Poor image quality that prevents reliable qualitative or radiomics analysis.