Abstract:
|
Prostate weight may be an important diagnostic factor in prostate cancer and transrectal ultrasound (TRUS) could be a favorable alternative to the lengthy magnetic resonance imaging (MRI) procedure and to invasive surgery. One issue with TRUS measurements, however, involves underestimation of the prostate weight. In this work, we present three models, unadjusted (1), adjusted for clinical factors as Gleason score, and pathological stage (2), and adjusted for demographic factors as age, BMI, and race, as well as clinical factors (3), that correct for this underestimation. The regression coefficients for the TRUS variable are similar among these models with values and standard errors of 9.309 (0.380), 9.227 (0.381), and 9.090 (0.398), respectively, and the predicted values from these models compare fairly well to the actual prostate weights. R-squared values approximating 0.58 also indicate that each of these models fit the data well. Therefore, we recommend application of these models to TRUS measurements, pending on what other clinical information may be available for the patient.
|