A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Histopathology Breast Cancer Detection and Classification using Optimized Superpixel Clustering Algorithm and Support Vector Machine
In recent decades, nuclei cell classification in histology has played a significant role in early diagnosis of breast cancer. However, it is a challenging task, due to the presence of small variant sizes of cells and heavy noise in histology images. To overcome these problems, an optimization-based superpixel-clustering algorithm is introduced to segment non-nuclei and nuclei cell separately. Initially, the histopathology breast images are collected from BreaKHis database. Then, image pre-processing is accomplished by applying normalization technique that increases the quality of image. After normalization, segmentation is performed using superpixel with Particle Swarm Optimizer (PSO) and Grey Wolf Optimizer (GWO) to segment nuclei, and non-nuclei cells. The proposed segmentation algorithm has the advantages of parallel computing and global optimization search, which helps to find a superior result of the superpixel-clustering algorithm. Later, feature extraction is carried out using Local Direction Ternary Pattern (LDTP), perimeter, solidity, circularity, Grey Level Co-Occurrence Matrix (GLCM), eccentricity, and colour autocorrelogram. Finally, Support Vector Machine (SVM) classifier is utilized to grade the histology breast images as malignant or benign images. In the experimental section, the proposed model obtained better performance in breast cancer detection compared to the existing research works and showed 7–8% improvement in classification accuracy.
Histopathology Breast Cancer Detection and Classification using Optimized Superpixel Clustering Algorithm and Support Vector Machine
In recent decades, nuclei cell classification in histology has played a significant role in early diagnosis of breast cancer. However, it is a challenging task, due to the presence of small variant sizes of cells and heavy noise in histology images. To overcome these problems, an optimization-based superpixel-clustering algorithm is introduced to segment non-nuclei and nuclei cell separately. Initially, the histopathology breast images are collected from BreaKHis database. Then, image pre-processing is accomplished by applying normalization technique that increases the quality of image. After normalization, segmentation is performed using superpixel with Particle Swarm Optimizer (PSO) and Grey Wolf Optimizer (GWO) to segment nuclei, and non-nuclei cells. The proposed segmentation algorithm has the advantages of parallel computing and global optimization search, which helps to find a superior result of the superpixel-clustering algorithm. Later, feature extraction is carried out using Local Direction Ternary Pattern (LDTP), perimeter, solidity, circularity, Grey Level Co-Occurrence Matrix (GLCM), eccentricity, and colour autocorrelogram. Finally, Support Vector Machine (SVM) classifier is utilized to grade the histology breast images as malignant or benign images. In the experimental section, the proposed model obtained better performance in breast cancer detection compared to the existing research works and showed 7–8% improvement in classification accuracy.
Histopathology Breast Cancer Detection and Classification using Optimized Superpixel Clustering Algorithm and Support Vector Machine
J. Inst. Eng. India Ser. B
Saturi, Rajesh (author) / Chand Parvataneni, Prem (author)
Journal of The Institution of Engineers (India): Series B ; 103 ; 1589-1603
2022-10-01
15 pages
Article (Journal)
Electronic Resource
English
Superpixel Endmember Detection
| Online Contents | 2010
Predicting tunnel squeezing using support vector machine optimized by whale optimization algorithm
| Springer Verlag | 2022