An artificial intelligent system to detect lower urinary tract problems using voiding sound

Abstract

On average in ageing males the prostate enlarges (Benign Prostatic Enlargement or BPE) and may cause Bladder Outlet Obstruction (BOO). The internationally standardized method for diagnosing BOO is based on measurements of urinary flow rate and bladder pressure, using a catheter inserted into the bladder via the urethra. This method is invasive, time-consuming and uncomfortable for the patient. We are developing a novel diagnostic method based on perineal recording of noise during urinary flow. Although it is known that (some aspects of) the recorded noise are (among others) related to the degree of obstruction, an exact and unique relation allowing derivation of the degree of obstruction from the noise recording is not known. In a biophysical model of the urethra it is found that the weighted average frequency, the standard deviation and the skewness of the power spectrum are monotonic related to the degree of obstruction. The standard deviation was found to be the best significant predictor of BOO (89% correct). Based on this model study, we are confident that a simple non-invasive acoustic method for diagnosing BOO caused by BPE can be developed, which would lower the threshold for urodynamic testing of patients with lower urinary tract symptoms.