Faculty Advisor

Professor Fred J. Looft

Faculty Advisor

Professor Ross D. Shonat

Faculty Advisor

Professor Christopher Sotak

Faculty Advisor

Professor Stevan Kun

Faculty Advisor

Professor Len Polizzotto

Faculty Advisor

Professor Robert A. Peura

Faculty Advisor

Hannu Harjunmaa, PhD




The addition of urea clearance monitoring to the care regimen of renal failure patients provides a dramatic decrease in complications due to improper or inadequate dialysis. Present methods of monitoring urea clearance are computationally complex and expensive to perform, resulting in poor rates of clinical acceptance of this measurement. Dialysate-side urea levels have been shown to relate to traditional measures of dialysis adequacy without the need for complex calculations. The requirements for photometric reagents or electrodes make determination of the urea level expensive and time consuming. This research is focused on the development of an optical measurement system to determine the sample urea level without the need for reagents. An algorithm was developed to predict the urea concentration of the sample from a set of optical transmission parameters recorded from the sample using a specially developed instrument. This instrument records the difference in sample transmission at two different wavelengths. Energy at the first wavelength is absorbed by urea, and the second wavelength is selected such that the matrix of the sample absorbs energy at both wavelengths equally. This effectively nulls out the absorbance of the background matrix, significantly improving the urea detection sensitivity. The algorithm was developed from an analysis of the instrument data and factors causing variations in the data. Calibration, bench study, and clinical protocols were designed, and performed using these protocols. Using a partial least squares approach, the algorithm was fit to a set of training data. The resulting algorithm was used to predict the urea content of patient hemodialysis samples. Compared to a reference standard (Beckman CX7, standard error /dl), the standard error of prediction for this algorithm was 0.47 mg/dl (N = 34 patients). The algorithm was able to predict dialysate urea at clinically relevant levels in samples collected from hemodialysis patients. Qualitative relationships were developed between the sample urea level and the data recorded from the sample. This system has the potential to provide a method that clinicians can use to efficiently and effectively monitor urea removal over the course of a dialysis session.


Worcester Polytechnic Institute

Degree Name



Biomedical Engineering

Project Type


Date Accepted





hemodialysate, urea, optical, Dialysis, Urea, Measurement