Noninvasive Biosensor Algorithms for Continuous Metabolic Rate Determination

Our collaborators in the JSC Cardiovascular lab implemented a technique to determine stroke volume during exercise using ultrasound imaging. Data collection using this technique has been completed for 31 subjects.

Spectral data collection has been completed on 6 subjects, both pre- and post-bed rest. Additionally, spectral data collection has been completed on 5 subjects in the hypovolemia study. Data review and analysis is underway.

Initial data review indicated that we needed to gain a better understanding of the impact of fluid shifts on our NIRS measurements. We conducted a stand test to determine the effect of postural changes on our NIRS measurements, including an independent assessment of blood volume using regional bioimpedance measurements. We learned in this study that it takes 15 min after a transition from standing to supine for the blood volume to normalize across all anatomical sites. We took advantage of this information in a study at UMass (University of Massachusetts) to investigate the sensor's ability to measure SmO2 on different muscles through various fat thicknesses. After 15 min of supine rest to allow blood to normalize across the body we demonstrated on 6 subjects that SmO2 measurements were equal, despite the choice of muscle used for the measurement (shoulder, calf, upper thigh, or lower thigh). This study included fat thickness that ranged from 5-19 mm.

We have received synergistic funding from non-NASA sources to develop a solid state, low profile sensor that can be worn in a space suit for determining metabolic rate. During this year we completed development of the prototype and demonstrated that it was equivalent to our fiber optic model in the measurement of muscle oxygen saturation and pH. With additional National Space Biomedical Research Institute (NSBRI) support we also developed the technology to run the sensor off a battery pack and augmented our system automation to include on-the-fly error checking and correction. The requirements for the additional automation features were the result of feedback from our NASA collaborators, based upon their previous experience using our original fiber optic system. In exercise studies we found that the optical signal can degrade during certain exercise periods. The new sensor is able to detect these problems and correct them within a few seconds, so no data is lost.

SUPPLEMENTAL INFORMATION FOR October 2009-September 2010 (submitted November 2010):

- Over the last twelve months spectral data collection has been completed on 15 subjects, both pre- and post-bed rest. The spectral data collected at JSC is uploaded to a shared server. All spectra are examined for spectral integrity. Spectral parameters SvO2, Hb, and SaO2 are then averaged over 30 sec intervals and matched to the VO2 data determined with the metabolic cart. When the stroke volume algorithm is completed, VO2 will be calculated from the spectra and compared to the gold standard measurement.

- Data collection for the hypovolemia study was completed. A total of 12 subjects (8 male and 4 female) completed the study. NIRS data was processed the same way for the hypovolemia study as is described above for the bed rest studies. Additionally, the hydrogen ion threshold was calculated from the spectra for both the normovolemic and hypovolemic exercise sessions. There were no differences observed between normo and hypovolemic exercises in either H+ threshold or the values of SmO2 through the exercise stages. This is consistent with the observation of similar blood lactate at peak exercise.

- Data collected from 10 healthy subjects were used to determine an initial model for stroke volume. This model was used to estimate VO2 for the same subjects. Estimated accuracy in calculating VO2 was not good enough, with larger errors occurring for exercise levels which are greater than 60% of maximum workload. The source of the error was traced to poor estimation of stroke volume at higher workloads. We continue to work on improving the estimation of stroke volume.

- We made significant improvements in the accuracy and reproducibility of our algorithms to calculate muscle oxygen saturation and hemoglobin concentration. A paper describing the improvements in the SO2 model and its validation on simulated spectra and high fidelity phantom materials is in press at a biomedical optics journal.

- Two solid state sensors were delivered to Johnson Space Center, along with significantly improved user interface software. The new user software incorporates many of the comments from JSC users of our earlier system. The new system reduces the calibration time, automatically resets the sensor during periods of low and high blood flow, and allows incorporation of user comments. Users in the JSC Cardiovascular and Exercise Laboratories were trained on the system. The new system and software were well received. Initial studies with the device have begun in the Exercise Lab.

EDITOR's NOTE (11/9/2010): Principal Investigator (PI) changed affiliation around October 2010 from University of Massachusetts Medical School to Reflectance Medical Inc. and a new grant number was issued. End date thus changed to 9/30/2010, per NSBRI, for Noninvasive Biosensor Algorithms for Continuous Metabolic Rate Determination, grant NCC 9-58-SMS 01301. See PI Soller and project Noninvasive Biosensor Algorithms for Continuous Metabolic Rate Determination--SMS01302 for subsequent reporting