Related Experiment Videos
Multiclass cancer classification by using fuzzy support vector machine and binary decision tree with gene selection
Yong Mao1, Xiaobo Zhou, Daoying Pi
1National Laboratory of Industrial Control Technology, Institute of Modern Control Engineering and College of Information Science and Engineering, Zhejiang University, Hangzhou 310027, China.
Journal of Biomedicine & Biotechnology
|July 28, 2005
Summary
This study introduces two novel multiclass cancer classifiers using gene selection for improved accuracy. Fuzzy Support Vector Machine (FSVM) with recursive feature elimination demonstrated superior performance in identifying key cancer-related genes.
Area of Science:
- Bioinformatics
- Computational Biology
- Genomics
Background:
- Multiclass cancer classification from gene expression data presents significant challenges.
- Accurate gene selection is crucial for effective cancer classification models.
Purpose of the Study:
- To develop and evaluate novel multiclass cancer classification techniques incorporating gene selection.
- To compare the performance of different gene selection methods and classifier architectures.
Main Methods:
- Proposed two multiclass classifiers: Fuzzy Support Vector Machine (FSVM) with gene selection and a binary classification tree based on SVM with gene selection.
- Employed F-test and recursive feature elimination (RFE) based on SVM for gene selection.
- Tested techniques on breast cancer, small round blue-cell tumors, and acute leukemia datasets.
- Utilized pre-selection of strong genes to enhance computational efficiency.
Main Results:
- FSVM with RFE achieved high recognition accuracy in multiclass cancer classification.
- The proposed FSVM-based method identified important genes affecting specific cancer types effectively.
- Demonstrated superior performance compared to existing methods and other tested binary classification trees.
Conclusions:
- FSVM combined with RFE is a powerful approach for multiclass cancer classification using gene expression data.
- This method excels at identifying critical genes associated with various cancers.
- Offers a promising avenue for advancing cancer diagnostics and personalized medicine.