Early detection of breast malignancy using wavelet features and optimized classifier

Breast cancer considered to be a significant health issue among women. Early detection will ensure the treatment is easier and more successful. Recently, numerous methodologies have developed using medical imaging to investigate breast cancer. This research seeks to build a computer-aided diagnostic (CAD) system to interpret mammograms. The first stage of CAD includes preprocessing, Fuzzy c means based segmentation applied to a localized area. In the second stage of the CAD method, the extraction of the feature is carried out using three distinct wavelet families with decomposition level at 4 and 6. The ANN, SVM, and ELM classifiers are used in the final stage to enable accurate classification. This article proposes ELM with the Grasshopper Optimization Algorithm (ELM-GOA) to adjust the weight between the input and hidden layer to obtain maximum performance at the middle layer. This method adopts mammogram enhancement, optimum image segmentation, wavelet-based feature extraction, and grasshopper optimization algorithm based ELM to ameliorating the accuracy and reducing the computational cost. The result shows that ELM-GOA has precision and sensitivity of 100% and 98% respectively. The CAD system can identify tumors with 99.33 % accuracy.

Automated summary

This study establishes a computer-aided diagnostic system to interpret breast mammograms, utilizing different wavelet families for feature extraction, employing ANN, SVM, and ELM classifiers for accurate classification, and enhancing middle layer performance through the ELM-GOA algorithm. The results show that the ELM-GOA system can accurately identify breast tumors with a high level of precision.