A Deep Learning Workflow for Mass-Forming Intrahepatic Cholangiocarcinoma and Hepatocellular Carcinoma Classification
Yangling Liu1, Bin Wang2, Xiao Mo1
1Faculty of Information Engineering and Automation, Kunming University of Science and Technology, Kunming 650504, China.
Current Oncology (Toronto, Ont.)
|January 20, 2023
Summary
A new deep learning workflow accurately differentiates mass-forming intrahepatic cholangiocarcinoma (MF-ICC) from hepatocellular carcinoma (HCC) using MRI. This approach enhances classification performance, aiding personalized treatment strategies for liver cancer.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Oncology
Background:
- Accurate differentiation between mass-forming intrahepatic cholangiocarcinoma (MF-ICC) and hepatocellular carcinoma (HCC) is critical for effective patient treatment.
- Magnetic resonance imaging (MRI) is a key modality for diagnosing liver lesions, but distinguishing between MF-ICC and HCC can be challenging.
Purpose of the Study:
- To develop and evaluate a novel deep-learning-based workflow for precise classification of MF-ICC and HCC using MRI.
- To improve the performance of deep learning models in classifying these liver cancers, especially on small datasets.
Main Methods:
- A semi-segmented preprocessing (Semi-SP) method was introduced to select relevant regions of interest (ROIs).
- A strided feature fusion residual network (SFFNet) was employed, featuring a multilayer feature fusion module (MFF), stationary residual blocks (SRB), and a convolutional block attention module (CBAM) for enhanced feature extraction and integration.
Main Results:
- The SFFNet model achieved an overall accuracy of 92.26% and an area under the curve (AUC) of 0.9680.
- High sensitivity (86.21%) and specificity (94.70%) were obtained for MF-ICC classification.
Conclusions:
- The proposed Semi-SP method and SFFNet workflow demonstrate strong capabilities in differentiating MF-ICC from HCC.
- This advanced classification approach offers valuable complementary information for developing personalized treatment strategies in liver cancer care.
