Related Experiment Videos
Computer-aided diagnosis for 3-dimensional breast ultrasonography
Dar-Ren Chen1, Ruey-Feng Chang, Wei-Ming Chen
1Department of General Surgery, China Medical College and Hospital, Taichung, Taiwan, Republic of China. dlchen88@ms13.hinet.net
Archives of Surgery (Chicago, Ill. : 1960)
|March 4, 2003
Summary
Three-dimensional (3-D) ultrasonography (US) significantly improves computer-aided diagnosis (CAD) for breast tumors compared to 2-D US. This advanced imaging enhances accuracy in distinguishing benign from malignant lesions, offering a valuable tool for reducing misdiagnosis.
Area of Science:
- Medical Imaging
- Diagnostic Technology
- Computational Pathology
Background:
- Conventional 2-dimensional (2-D) ultrasonography (US) is widely used in breast imaging.
- 3-dimensional (3-D) US is an emerging technique offering potentially more detailed information.
- Computer-aided diagnosis (CAD) systems can aid in interpreting breast US images.
Purpose of the Study:
- To develop and evaluate a CAD method utilizing 3-D US breast images.
- To compare the diagnostic performance of CAD with 3-D US versus 2-D US.
- To assess the potential of 3-D US in improving the classification of breast tumors.
Main Methods:
- A CAD system was developed using 3-D US images of 107 benign and 54 malignant breast tumors.
- Texture features were extracted from 3-D US images using novel 3-D autocorrelation coefficients.
- A neural network classified tumors as benign or malignant based on extracted texture features.
Main Results:
- The 3-D US CAD system achieved an area under the receiver operating characteristic curve (Az) of 0.97.
- The 2-D US CAD system achieved an Az value of 0.85.
- Accuracy, sensitivity, specificity, and predictive values were significantly higher with 3-D US compared to 2-D US.
Conclusions:
- The proposed CAD system using 3-D US is a promising tool for classifying breast tumors.
- 3-D US offers a significant advantage over 2-D US for computer-aided diagnosis of breast lesions.
- This technology can serve as a valuable second reading to minimize diagnostic errors.