Diagnosis of bladder outlet obstruction using objective parameters and neural networks classifiers

Abstract

This paper presents a neural network system to classifi patients of lower urinary tract symptoms (LUTS) and obtain their degree of bladder outlet obstruction (BOO) according to linear passive urethral resistance relation (PURR) nomogram or schafer grade (0 or I ) for nonobstructed flow, 2 for equivocal and (3,4,5 or 6) for obstructed patient. LUTS patients received routine investigation, consisting of transrectal ultrasonography of the prostate, serum PSA measurement, assessment of symptoms and quality of lye by the International Prostate Symptom Score (IPSS)), urinary flowmetry with determination of maximum flow rate, voided volume and post-void residual urine and full pressure flow studies (PFS) which are the best available method to distinguish BOO, But PFS are too invasive and time-consuming and expensive to be routinely utilized. Thus an Artificial Neural Network (ANN) was constructed to estimate the degree of obstruction (schafer grade). The input to the ANN consisted of four readings (average flow rate A-F-R, maximum flow rate M-F-R, prostate size as measured by transrectal ultrasound TRUS and residual urine Res-Urin) which are most sign$cant and less invasive. The performance of the ANN classij?er was compared with that of a minimum distance and a voting k nearest neighbor classifiers. The ANN revealed better results than both two classifers.