A HIERARCHICAL CLASSIFIER FOR MULTICLASS PROSTATE HISTOPATHOLOGY IMAGE GLEASON GRADING

Shahnorbanun Sahran,Afzan Adam,Mohammed Alweshah,Azizi Abdullah,Dheeb Albashish

doi:10.32890/jict2018.17.2.8256

Abstract

Automated classification of prostate histopathology images includes the identification of multiple classes, such as benign and cancerous (grades 3 & 4). To address the multiclass classification problem in prostate histopathology images, breakdown approaches are utilized, such as one-versus-one (OVO) and one- versus-all (Ovall). In these approaches, the multiclass problem is decomposed into numerous binary subtasks, which are separately addressed. However, OVALL introduces an artificial class imbalance, which degrades the classification performance, while in the case of OVO, the correlation between different classes not regarded as a multiclass problem is broken into multiple independent binary problems. This paper proposes a new multiclass approach called multi-level (hierarchical) learning architecture (MLA). It addresses the binary classification tasks within the framework of a hierarchical strategy. It does so by accounting for the interaction between several classes and the domain knowledge. The proposed approach relies on the ‘divide-and-conquer’ principle by allocating each binary task into two separate subtasks; strong and weak, based on the power of the samples in each binary task. Conversely, the strong samples include more information about the considered task, which motivates the production of the final prediction. Experimental results on prostate histopathological images illustrated that the MLA significantly outperforms the Ovall and OVO approaches when applied to the ensemble framework. The results also confirmed the high efficiency of the ensemble framework with the MLA scheme in dealing with the multiclass classification problem.

Highlights

Prostate cancer is an affliction that is becoming increasingly common in elder males
The performance of the ensemble_OvO, ensemble_o3v3al7l, and ensemble_MLA/MV on two prostate histopathology image datasets for solving the three-class problem in PCa grading is measured in terms of accuracy, sensitivity, and specificity as tabulated in Figure 8, and Tables 2 and 3
When the ensemble_Ovall is tested, four multiclass Ovall approaches are used based on the number of tissue components, where each multiclass Ovall reports three support vector machines (SVM)

Summary

INTRODUCTION

Prostate cancer is an affliction that is becoming increasingly common in elder males. This paper focuses on solving the three-class classificatio2n problem in prostate cancer grading, i.e. Grade 3, Grade 4, and Benign It starts by proposing a new hierarchical multiclass approach called multi-level learning architecture (MLA), strategy. The final prediction for the test sample x is assigned to the class with the maximum decision value as illustrated in Equation (5) In CADs of PCa grading, some studies employed the OVA approach to classify the multiple Gleason grades or different groups in the prostate histopathology images. A new supervised multi-level (hierarchical) learning architecture (MLA) for solving the three-class classification problem in prostate based on the histopathology images is introduced. To demonstrate the concept of the MLA, the researchers gave an example of how it can address the three-class classification problem in PCa grading The global decision for the ensemble_MLA for a test image is obtained via majority voting strategy, which was employed in the previous ensemble_OVO and ensemble_OVall

EXPERIMENTAL RESULTS AND DISCUSSION

Experiment Results and Discussion

CONCLUSION