Computer algorithm for analysing breast tumor angiogenesis using 3-D power Doppler ultrasound
Ruey-Feng Chang1, Sheng-Fang Huang, Woo Kyung Moon
1Department of Computer Science and Information Engineering, National Chung Cheng University, Chiayi, Taiwan. rfchang@cs.ccu.edu.tw
Ultrasound in Medicine & Biology
|October 19, 2006
Summary
This study introduces a novel method using 3-D power Doppler ultrasound to analyze tumor vascular morphology for improved breast cancer diagnosis. The approach accurately differentiates benign from malignant tumors based on vessel structure, offering a new tool for oncologists.
Area of Science:
- Oncology
- Medical Imaging
- Biomedical Engineering
Background:
- Tumor angiogenesis, driven by factors like VEGF, is crucial for tumor growth and metastasis.
- Tumor vasculature exhibits abnormal morphology (disorganization, tortuosity) compared to normal vessels.
- These morphologic features offer potential for improved diagnostic approaches in oncology.
Purpose of the Study:
- To develop and evaluate a new method for capturing and analyzing tumor vascular morphology from 3-D power Doppler ultrasound (PDUS) images.
- To assess the feasibility of using extracted vascular features to differentiate between benign and malignant breast tumors.
Main Methods:
- A 3-D thinning algorithm was employed to extract vessel skeletons from PDUS images.
- Seven quantitative morphologic features were extracted: vessel-to-volume ratio, number of vascular trees, number of bifurcations, mean radius, and three tortuosity measures.
- A neural network was utilized to classify 221 solid breast tumors (110 benign, 111 malignant) based on these features.
Main Results:
- All seven extracted vascular features were statistically significant (p < 0.05) in differentiating tumor types.
- The method achieved an overall accuracy of 80.09%, with sensitivity and specificity of 80.18% and 80% respectively.
- The area under the receiver operating characteristic curve (A(Z)) was 0.89, indicating good diagnostic performance.
Conclusions:
- The proposed method for analyzing 3-D PDUS-derived vascular morphology is feasible for breast tumor characterization.
- Extracted morphologic features show significant potential for improving the accuracy of distinguishing benign from malignant breast lesions.
- This technique offers a promising, non-invasive tool that complements existing diagnostic methods in breast oncology.