Visualization and segmentation of liver tumors using dynamic contrast MRI

Area of Science:

• Medical Imaging
• Oncology
• Machine Learning

Background:

• Hepatocellular carcinoma (HCC) is a prevalent malignancy with high mortality.
• Accurate quantification of liver tumor burden and viability is crucial for treatment assessment.
• Dynamic Contrast Enhanced MRI (DCE-MRI) shows promise but requires improved quantitative methods.

Purpose of the Study:

• To develop a semi-supervised technique for visualizing and measuring liver tumor burden and viability from DCE-MRI.
• To address the challenges in segmenting liver tumors in DCE-MRI data.
• To enhance the quantitative analysis of hepatocellular carcinoma using medical imaging.

Main Methods:

• A semi-supervised approach is proposed for DCE-MRI analysis.
• K-means clustering is utilized in a hybrid intensity-spatial feature space.
• Spatio-temporal characteristics of DCE-MRI data are exploited to improve segmentation.

Main Results:

• The proposed method enables visualization and measurement of liver tumor burden.
• The technique aids in assessing tumor viability from DCE-MRI.
• Improved segmentation accuracy for liver tumors is achieved.

Conclusions:

• The developed semi-supervised technique offers a robust solution for quantifying liver tumor burden and viability.
• This method enhances the clinical utility of DCE-MRI for hepatocellular carcinoma management.
• Further development in quantitative imaging biomarkers for liver cancer is warranted.