Histopathological images play an important role in selecting effective therapies; they are necessary for determining the health of a particular biological structure and diagnosing disorders like cancer. Machine learning for medical diagnosis reduces the likelihood of a misdiagnosis. Efficiency gains are the main benefit of applying machine learning (ML) to medical diagnosis. ML gives the medical community more time to concentrate on their profession as Artificial Intelligence improves. Investigating the classification performance of several ML Algorithms might be fascinating. On datasets, these algorithms are first trained. Machine learning algorithms are trained using many custom features. To improve cancer detection through the classification of histopathological images, the work the paper demonstrates the Thepade SBTC (Sorted Block Truncation Coding) global features and their fusion with Bernsen Thresholding extracted local features. Utilizing the 960 images from the KIMIA Path960 Dataset [1], experimental validation is carried out with performance indicators alias specificity, sensitivity, and accuracy. Here the feature fusion of Thepade SBTC and Bernsen binarization has shown improved classification of histopathological images over consideration of individual features. The ensemble of ML algorithms with majority voting logic has improved the classification of histopathological images over individual ML algorithms. The Ensemble of ‘LMT, Simple Logistic (SL), and Multilayer Perceptron (MP)’ for TSBTC 9-ary and fusion with Bernsen binarization feature give the same accuracy of 97.3958% in a ten-fold cross-validation scenario as the Ensemble of ‘Simple Logistic, Multilayer Perceptron, and Random Forest (RF)’ for TSBTC 9-ary and fusion with Bernsen binarization feature.