Blood volume measurement at the bedside using ICG pulse spectrophotometry

Background: In the treatment of critically ill patients, blood volume (BV) measurement requires injection of some tracer substance and subsequent blood sampling to analyze the tracer concentration. To obviate both the sampling and laboratory analysis, techniques of pulse oximetry have been adapted to the noninvasive optical measurement in the patient's nose or finger of the arterial concentration of an injectable dye. Methods: The authors report the clinical accuracy of a new noninvasive bedside BV measurement test that uses pulse spectrophotometry (the pulse method). The device detects pulsatile changes of tissue optical density of a nostril or a finger spanned by a probe emitting two infrared wavelengths (805 and 890 nn). After a peripheral or central intravenous injection of indocyanine green, the arterial dye concentration is continuously computed by reference to the previously measured blood hemoglobin concentration. Three types of tests of its accuracy are described here. Results: In 10 healthy volunteers, the authors compared BV determined by the pulse method with an I-131-labeled human serum albumin method. Three subject data sets were excluded because of motion artifact, a low signal:noise ratio, or both. For the other seven volunteers, the bias +/- SD of pulse spectrophotometric RV values were 0.20 +/- 0.24 1(or 4.2 +/- 4.9%) for the nose probe and 0.34 +/- 0.31 I (or 7.3 +/- 6.9%) for the finger probe, with a mean BV of 5 1, In 30 patients who underwent cardiac surgery, the pulse method was compared with a standard indocyanine green method using intcfinittent blood sampling. In three patients, the BV could not be determined by the pulse method because of motion artifact, low signal:noise ratio, or both. In 27 patients, the bias +/- SD of the BV by the pulse method was -0.23 +/- 0.37 1(-5.3 +/- 8.7%) for the nose and -0.25 +/- 0.51(-4.2 +/- 8.4%) for the finger, Patient BV ranged from 2.51 to 7.13 1 (mean, 4.48 1). In 10 additional patients before cardiac surgery, BV was measured by the pulse method before and shortly after removal of 400 ml blood. The pulse method recorded a decrease of By of 480 +/- 114 ml. Three days after venesection, the mean RV was 117 +/- 159 ml less than the predonation control. Conclusions: In most patients, the pulse method provides bedside measurement of BV without blood sampling (except for hemoglobin determination), with an estimated error less than 10%. In 10-30% of tests the method failed because of motion distortion of the record during the 10-min data collection period or because of insufficient pulse amplitude in the test tissue.