Related Experiment Video
Updated: Mar 15, 2026
13:44
Detection of Architectural Distortion in Prior Mammograms via Analysis of Oriented Patterns
Published on: August 30, 2013
43.8K
An interval prototype classifier based on a parameterized distance applied to breast thermographic images.
Marcus C Araújo1, Renata M C R Souza2, Rita C F Lima1
1Departamento de Engenharia Mecânica, Universidade Federal de Pernambuco, Av. Prof. Moraes Rego, 1235, Cidade Universitária, Recife, PE, 50670901, Brazil.
Medical & Biological Engineering & Computing
|September 16, 2016
Summary
This study introduces a new method using interval temperature data from breast thermography to classify breast abnormalities. The approach effectively detects breast cancer, outperforming existing methods in accuracy and sensitivity.
Area of Science:
- Medical Imaging
- Machine Learning
- Biomedical Engineering
Background:
- Breast cancer remains a leading cause of mortality in women.
- Thermographic imaging offers a non-invasive approach for breast cancer diagnosis.
- Accurate classification of breast abnormalities (malignant, benign, cyst) is crucial.
Purpose of the Study:
- To propose an innovative approach for breast cancer detection using interval temperature data from thermography.
- To classify breast abnormalities by considering internal interval variations.
- To enhance the separation and classification of breast abnormality types.
Main Methods:
- Employing interval temperature data for breast abnormality classification.
- Mapping interval data into a separable feature space.
- Building class prototypes and using parameterized Mahalanobis distance for interval-valued data.
- Applying the classifier to a breast thermography dataset from Brazil.
Main Results:
- Achieved 16% misclassification rate and 93% sensitivity to the malignant class in one scenario.
- Achieved 100% sensitivity to the malignant class with 20% overall misclassification rate in another scenario.
- Demonstrated superior performance compared to existing interval data classification methods for breast thermography.
Conclusions:
- The proposed interval data classification method shows significant promise for breast cancer detection.
- The approach effectively classifies breast abnormalities with high sensitivity for malignant cases.
- This method offers an improved alternative to current techniques in breast thermography analysis.
