Search

Background and Objectives Prostate cancer is the most common type of malignant cancer among men and is known as one of the leading causes of cancer mortality in men. The difficulty of diagnostic procedures such as tumor biopsy has made new diagnostic methods, such as magnetic resonance imaging (MRI), to be one of the research priorities for prostate cancer in recent years. The aim of this study is to develop an automated system capable of accurately diagnosing prostate cancer from MRI images. Subjects and Methods In this applied descriptive study, a four-step method was used for diagnosing prostate cancer with MRI technique. In the first step, the effect of noise was reduced by using the discrete two-dimensional wavelet transform and histogram equalization. In the second step, the blackhole optimization algorithm was used for the segmentation of the input image based on the multilevel threshold technique. In this way, the tumor suspicious areas can be identified on the image. In the third step, the features of each target area were extracted. In the final step, a combination of three machine learning algorithms, including artificial neural network, decision tree, and support vector machine, was used to diagnose prostate cancer. The effectiveness of the proposed method was evaluated from various aspects and its performance was compared with other machine learning models. Results The proposed method had an accuracy of 99%, sensitivity of 0.98, precision and specificity of 1 in diagnosing prostate cancer with MRI method. Conclusion The proposed ensemble method using a combination of image processing, optimization, and machine learning techniques, has higher accuracy compared to other machine learning models. [ABSTRACT FROM AUTHOR]