454
H. M. Heise
penetration depths, which requires a scaling of the aqueous glucose absorptivity
spectrum (“response spectrum”).
The problems with the repeatability of in vivo spectroscopic measurements are
described by various aspects as follows. Glucose in different compartments, e.g.,
interstitial and vascular space, blood flow and blood volume, skin temperature
gradients, comparability of reference venous or capillary blood glucose concentration with tissue probed concentration and others must be taken into account and
experiments have to be carefully planned for reliable results. For the complexity of
in vivo measurements, several extensive investigations have been carried out with
results published earlier, but the challenge of achieving reliable non-invasive skin
measurement repeatability is still in the focus of researchers; for particular challenges
observed for the developments, see for example [31, 32].
Some information on the variability of tissue spectra can be gained from
Fig. 20.11, which provides spectra as measured at different temperatures. For a
tissue without blood perfusion, the temperature dependency of water as the main
constituent is readily observable; for in vivo measurements of skin at different
temperatures, also other effects come into play due to the response of the vasculature. Thus, increasing temperature causes changing perfusion, blood volume and
also diminishing arterial-venous concentration differences due to higher blood flow.
Another more advanced approach arises from the opportunities, which exist in
photoplethysmography (PPG) [33]. This measurement technique enables the sensing
of information from the arterial vascular compartment. It can detect the periodic
blood volume changes due to the systolic and diastolic blood pressures from the
beating heart throughout the cardiac cycle based on NIRspectra, which has been
routinely implemented in pulse oximetry. A difference to the latter is that broad
spectral intervals need to be analyzed for determining the arterial blood glucose
concentration. This strategy was followed by Yamakoshi and coworkers, starting
with spectra between 900 and 1700 nm and coining their measurement technique as
“pulse glucometry.” After several improvements, an advanced set-up was designed
for side-scattered finger photoplethysmography that was presented in 2017 [34]. First,
PPG signals with three wavelengths: 808 nm, 1160 nm and 1600 nm (coinciding with
nearly peak glucose and strong water absorptions) were compared, while the sourcedetector spacing was successively increased circumferentially around the fingertip.
A second experiment was performed with six wavelengths from 1550 to 1749 nm
for accessing glucose absorption bands. The pulsatile signal-to-noise ratios were
claimed to be measured around 15 dB, giving hope for its potential for realizing a
practical measurement of arterial blood constituents including glucose, but awaiting
further developments.
It is interesting to take a look at early experiments performed by Nahm and
Gehring, who studied the pulsatile spectrum from time-resolved near-infrared spectroscopic experiments with an injected bolus of indocyanine green (ICG) during
transillumination of a fingertip [35]. Their results showed that the pulsatile component was by a factor up to 300 smaller than that of the static vasculature component,
underlining the challenges of pulse glucometry versus integral tissue measurements.
H. M. Heise
penetration depths, which requires a scaling of the aqueous glucose absorptivity
spectrum (“response spectrum”).
The problems with the repeatability of in vivo spectroscopic measurements are
described by various aspects as follows. Glucose in different compartments, e.g.,
interstitial and vascular space, blood flow and blood volume, skin temperature
gradients, comparability of reference venous or capillary blood glucose concentration with tissue probed concentration and others must be taken into account and
experiments have to be carefully planned for reliable results. For the complexity of
in vivo measurements, several extensive investigations have been carried out with
results published earlier, but the challenge of achieving reliable non-invasive skin
measurement repeatability is still in the focus of researchers; for particular challenges
observed for the developments, see for example [31, 32].
Some information on the variability of tissue spectra can be gained from
Fig. 20.11, which provides spectra as measured at different temperatures. For a
tissue without blood perfusion, the temperature dependency of water as the main
constituent is readily observable; for in vivo measurements of skin at different
temperatures, also other effects come into play due to the response of the vasculature. Thus, increasing temperature causes changing perfusion, blood volume and
also diminishing arterial-venous concentration differences due to higher blood flow.
Another more advanced approach arises from the opportunities, which exist in
photoplethysmography (PPG) [33]. This measurement technique enables the sensing
of information from the arterial vascular compartment. It can detect the periodic
blood volume changes due to the systolic and diastolic blood pressures from the
beating heart throughout the cardiac cycle based on NIRspectra, which has been
routinely implemented in pulse oximetry. A difference to the latter is that broad
spectral intervals need to be analyzed for determining the arterial blood glucose
concentration. This strategy was followed by Yamakoshi and coworkers, starting
with spectra between 900 and 1700 nm and coining their measurement technique as
“pulse glucometry.” After several improvements, an advanced set-up was designed
for side-scattered finger photoplethysmography that was presented in 2017 [34]. First,
PPG signals with three wavelengths: 808 nm, 1160 nm and 1600 nm (coinciding with
nearly peak glucose and strong water absorptions) were compared, while the sourcedetector spacing was successively increased circumferentially around the fingertip.
A second experiment was performed with six wavelengths from 1550 to 1749 nm
for accessing glucose absorption bands. The pulsatile signal-to-noise ratios were
claimed to be measured around 15 dB, giving hope for its potential for realizing a
practical measurement of arterial blood constituents including glucose, but awaiting
further developments.
It is interesting to take a look at early experiments performed by Nahm and
Gehring, who studied the pulsatile spectrum from time-resolved near-infrared spectroscopic experiments with an injected bolus of indocyanine green (ICG) during
transillumination of a fingertip [35]. Their results showed that the pulsatile component was by a factor up to 300 smaller than that of the static vasculature component,
underlining the challenges of pulse glucometry versus integral tissue measurements.