Mustafa Ghaderzadeh; Farahnaz Sadoughi; Arvin Ketabat
Volume 9, Issue 4 , September and October 2012, , Pages 457-464
Abstract
Introduction: In recent years, the concepts of artificial neural networks (ANN) have extensivelyundergone remarkable development in early detection and classification of diseases such asbenign prostatic hyperplasia (BPH). The usage of ANN has become widely accepted in medicalapplications owing to its ...
Read More
Introduction: In recent years, the concepts of artificial neural networks (ANN) have extensivelyundergone remarkable development in early detection and classification of diseases such asbenign prostatic hyperplasia (BPH). The usage of ANN has become widely accepted in medicalapplications owing to its potential capabilities for detecting the complex interactions amongvariables, diagnosis and diseases’ modeling. The present study aimed to design and implement adecision support system (DSS) based on ANN for early detection of prostate cancer.Methods: This survey design was conducted through data collection among 360 males withprostate abnormalities in Urology Department of Imam Khomeini Hospital, Tehran, Iran, fromJanuary 2008 to March 2011. In order to assess the performance and accuracy of the designedsystem, sensitivity, specificity and receiver-operating characteristics (ROC) curve were used asthe indicators of distinguishing prostate cancers from BPH. In order to implement DSS in thisstudy, scaled conjugate gradient (SCG) algorithm was used as the main algorithm for earlydetection of prostate cancer from benign prostate.Results: The proposed intelligent ANN-based system can be used as a strong diagnostic tool with97.0% specificity and 92.1% sensitivity for detecting the prostate cancer and to differentiate itfrom BPH. The results indicated a high potential of artificial neural network as a strong tool inclassification of prostatic neoplasia diseases.Conclusion: A medical decision support system was used aiming to help medical experts in theirclassification and early detection of prostatic neoplasia disorders in the present study. Suchartificial intelligent-based medical intelligent systems, particularly for neural networks, can helpphysicians in accurate decision-making concerning prostate cancer and BPH. Using suchsystems, specialists would be able to eliminate or minimize unnecessary biopsy and reducediagnostic costs. In addition, such systems can accelerate the diagnostic detection time.
