Enhanced Histopathological Image Classification through the fusion of Thepade Sorted Block Truncation Code and Otsu Binarization features

Abstract

Histopathology is the branch of pathology that investigates the structure of cells and tissues of organisms at a microscopic level. Histopathological images are crucial in the decision-making process for effective therapies, determining the health of a particular biological structure and identifying diseases like cancer. With machine learning models, it may be feasible to increase the accuracy of medical data, decrease patient rate variations, and cut costs associated with medical care. Most medical scientists are drawn to such new technologies of predictive models in chronic disease forecasting. A novel approach for more accurate classification of histopathological images is proposed in this paper. The technique involves fusing the features extracted from two methods, namely Otsu's binarization and Thepade Sorted Block Truncation Code, to achieve improved results. The KIMIA Path960 dataset comprising 960 images is utilized for experimental validation with performance indicators like accuracy, specificity, and sensitivity. Ensembles of Simple Logistics, Multilayer Perceptron, Logistics Model Tree, as well as Simple Logistics, Random Forest, and Logistic Model Tree classifiers, demonstrated superior performance for the fusion of Thepade Sorted Block Truncation Code 7-ary and Otsu features, achieving an accuracy of 97.39 percent in a 10-fold cross-validation scenario.

Keywords

• Histopathology
• Thepade SBTC
• Feature Fusion
• Ensembles

References

1. [1] MDinesh Kumar,MBabaie, S Zhu, S Kalra, and HR Tizhoosh. A comparative study of cnn, bovw and lbp for classification of histopathological images. In IEEE Symposium Series on Computational Intelligence (SSCI), pages 1–7, 2017.
2. [2] TJ Alhindi, S Kalra, KH Ng, A Afrin, and HR Tizhoosh. Comparing lbp, hog and deep features for classification of histopathology images. In International Joint Conference on Neural Networks (IJCNN), pages 1–7, 2018.
3. [3] A Ganguly, R Das, and SK Setua. Histopathological and lymphoma image classification using customized deep learning models and optimization algorithms. In 11th International Conference on Computing, Communication and Networking Technologies (ICCCNT), pages 1–7, 2020.
4. [4] A Anurag, R Das, GK Jha, SD Thepade, N D’Souza, and C Singh. Feature blending approach for efficient categorization of histopathological images for cancer detection. In IEEE Pune Section International Conference (PuneCon), pages 1–6, 2021.
5. [5] RR Kadhim and MY Kamil. Evaluation of machine learning models for breast cancer diagnosis via histogram of oriented gradients method and histopathology images. International Journal on Recent and Innovation Trends in Computing and Communication, (10), 2022.
6. [6] I Hirra, M Ahmad, A Hussain, MU Ashraf, IA Saeed, SF Qadri, AM Alghamdi, and AS Alfakeeh. Breast cancer classification from histopathological images using patch-based deep learning modelling. IEEE Access, 9:24273–24287, 2021.
7. [7] R Rashmi, K Prasad, and CBK Udupa. Bchisto-net: Breast histopathological image classification by global and local feature aggregation. Artificial Intelligence in Medicine, 121:102191, 2021.
8. [8] S Khairnar, SD Thepade, and S Gite. Effect of image binarization thresholds on breast cancer identification in mammography images using otsu, niblack, burnsen, thepade’s sbtc. Intelligent Systems with Applications, 2021.

9. [9] SD Thepade and Y Bafna. Improving the performance of machine learning classifiers for image category identification using feature level fusion of otsu segmentation augmented with thepade’s n-ary sorted block truncation coding. In 2018 Fourth International Conference on Computing Communication Control and Automation (ICCUBEA), pages 1–6, 2018.
10. [10] S. D. Thepade and P. R. Chaudhari. Land usage identification with fusion of thepade sbtc and sauvola thresholding features of aerial images using ensemble of machine learning algorithms. Applied Artificial Intelligence, 35:154–170, 2020.
11. [11] N Otsu. A threshold selection method from gray-level histograms. IEEE Transactions on Systems, Man, and Cybernetics, 9(1):62–66, 1979.
12. [12] J Yousefi. Image binarization using otsu thresholding algorithm. 2015.
13. [13] P Rahim, N Mustafa, H Yazid, TX Jian, S Daud, and K Rahman. Segmentation of tumour region on breast histopathology images to assess glandular formation in breast cancer grading. Journal of Physics: Conference Series, 2071:012051, 2021.
14. [14] MK Slifka and JL Whitton. Clinical implications of dysregulated cytokine production. J. Mol. Med., 78:74–80, 2000. ECJSE Volume