Comprehensive multi-phase 3D contrast-enhanced CT imaging for primary liver cancer

Area of Science:

• Medical Imaging
• Oncology
• Artificial Intelligence

Background:

• Primary liver cancer presents a major global health challenge with high mortality rates.
• Accurate diagnosis and subtype classification are essential for effective treatment and improved patient outcomes.
• Contrast-enhanced computed tomography (CECT) offers high accuracy for liver cancer diagnosis, but limited datasets hinder model development.

Purpose of the Study:

• To create a comprehensive 3D CECT dataset for primary liver cancer.
• To support the development and validation of diagnostic and segmentation models for liver cancer.
• To address the limitations of existing CECT scan datasets in terms of subtype coverage and scan phases.

Main Methods:

• Compiled a CECT dataset including 278 primary liver cancer cases (hepatocellular carcinoma, intrahepatic cholangiocarcinoma, combined hepatocellular-cholangiocarcinoma) and 83 non-liver cancer controls.
• Included CECT images from multiple complete scan phases for each subject.
• Annotated liver and lesion areas within the CECT scans, resulting in over 50,000 lesion images.

Main Results:

• Established a large-scale, annotated CECT dataset specifically for primary liver cancer research.
• The dataset encompasses diverse liver cancer subtypes and complete CT scan phases.
• Provides a valuable resource for training and validating AI-driven diagnostic and segmentation tools.

Conclusions:

• The developed CECT dataset is crucial for advancing AI in primary liver cancer diagnosis and classification.
• This resource will facilitate the creation of more robust and accurate diagnostic and segmentation models.
• Facilitates research into improving patient outcomes through better characterization of liver cancer subtypes.