Abstract
To the Editor:
Newton et al. (1997) state in their abstract and conclusion that the current method of measuring CBF by near infrared spectrophotometry (CBFNIRS) in the reflectance mode is inaccurate. It is misleading to emphasize the difference between reflectance and transmission NIRS. The scattering of light at the head is very strong. Therefore, after a short distance, we measure only scattered light whether the detector is fixed on the same side of the head as the emitter or on the opposite. Contrary to what Newton et al. (1997) state, the study by Skov et al. (1991) was actually performed in reflectance with an interoptode distance of 4 to 5 cm. Thus measuring cerebral blood flow in neonates has been validated for reflectance as well as transmission (Bucher et al. 1993) measurements. Even the finding that NIRS overestimates cerebral blood flow compared to the xenon clearance method in Bucher's study and underestimates it in Skov's study has been explained by the different methods of evaluating the data (Wolf et al. 1996). This proves that in neonates both reflection and transmission NIRS are valid for measuring cerebral blood flow and doubts about this are difficult to justify by a geometrically considerably different animal model.
Furthermore, it previously has been shown in adults that as the layers surrounding the brain are thicker and hence the proportion of the path the light spends in these higher, the cerebral blood flow values measured by NIRS can no longer be attributed solely to the brain (Owen-Reece et al. 1996). However, reasonable values were obtained, when the optodes were put directly on the dura measuring reflectance.
Hence the crucial point is: What are the optical properties of these layers? What is the proportion of the total path of light spent in these layers? Consequently, we would have expected more geometrical information about dogs, such as thickness of the cranium, CSF layer, and brain geometry.
It has previously been shown (Wolf et al. 1996) that the test-retest variability increases considerably, if the sample rate is lowered. Thus the high test-retest variability of Newton et al. (1997) can be attributed to the sample rate, which was too low compared to the short transit time in dogs. Only five samples were used for the calculations of the oxygenation increase. Using an appropriate sample rate as well as equipment with low noise, it is possible to reach test-retest variabilities below 20% (Skov et al. 1991; Bucher et al. 1993; Wolf et al. 1996).
In our experience the proportion of the measurements, which have to be rejected, depends on the skill of the person performing the measurements, especially the 31.3% of the measurements, which were not acceptable because of baseline problems with the arterial oxygen saturation. Even though we used conventional pulse oximetry to determine arterial oxygen saturation, which certainly is one of the weak spots of this method, we only had to reject approximately 50% of the measurements (Wolf et al. 1996).
In conclusion, NIRS is a valid and feasible method to measure cerebral blood flow, if it is used in the appropriate way.