Breast Histopathology Images Multi-Classification using Ensemble of Deep Convolutional Neural Networks
Sadura Priscilla Akinrinwa, Opeoluwa Oluwatoyin Olabode, O.C. Agbonifo, Kolawole G. Akintola · International Journal of Scientific Research in Computer Sciences and Engineering · 2022
Breast cancers have constituted a major health challenge as a leading cause of mortality in women. This has led to several interventions in the diagnosis and treatment of the disease. The digital classification and analysis of breast histopathology images provides a means for computerized-clinical diagnosis of breast cancers. In this study, three separate models based on the ensembles of Convolutional Neural Networks (CNNs) for the analysis and classification of histopathological images of breast tissue are presented. The ensembles make use of majority voting, averaging, and stacking techniques. The ensemble models seek to extend the performance of existing CNNs by combining AlexNet, VGGNet and ResNet using majority voting, averaging, and stacking ensemble rules. Furthermore, a model for classification of breast histopathology images was developed called the SaduNet model. This model was developed with few numbers of convolutional neural network layers to reduce computational cost in terms of memory and improve computation time. All the models were trained with histopathology images dataset collected at the Federal Teaching Hospital, Ido Ekiti, Ekiti state, Nigeria, with ethical clearance to ensure that the research is applicable locally. Different learning parameters were used in the different convolutional neural networks developed to ensure that they obtain optimal performances of the models on the histopathology images classification task. The comparative analysis performed showed that the developed models performed as well as those found in literature judging by the accuracies achieved. The ensemble methods also performed better in the terms of the sensitivity and predictability than the individual base models. This is shown in the high prediction and recall values obtained by the ensemble models. During testing, the base models generated the following accuracies: AlexNet: 92.91%, VGG16: 96.28% and ResNet: 99.25%. When tested with the FTH breast histopathology data, the averaging ensemble has accuracy of 99.47% while the majority voting ensemble has accuracy of 99.30% and stacking ensemble model has accuracy of 97.86%. The SaduNet model also achieved an accuracy of 75.89%.