Supplementary material
Sensitivity of ODR to change in saturation
The methods section explains that for every unit change in optical density ratio (ODR), the calculated
saturation decreases by 2.47 percentage points. This is about 1.93 times greater rate of change than
for the fundus camera based Oxymap T1 oximeter, according to default manufacturer calibration. A
smaller change in ODR for each unit change in saturation suggests that the SLO oximeter is less
sensitive to change in saturation.
The difference in sensitivity may be due to the different wavelengths used in the two systems. The
fundus camera oximeter uses 570nm (isosbestic) and 600nm (sensitive to oxygen saturation) while
the SLO uses 532nm (close to isosbestic) and 633nm (sensitive).
The light extinction coefficient for a solution of a mixture of Hb and HbO2 can be written as follows:
εsolution = εHbO2*SatO2+ εHb(1-SatO2)1
The optical density of a solution can be written as OD = εsolution*C*d, where C is the total
concentration of hemoglobin and d is the distance, which the light has to travel through the solution.
Now let us assume that we have two solutions of hemoglobin, with saturations SatO2-1 and SatO2-2.
The difference in the OD at the sensitive wavelength is OD = ε*C*d (since C and d do not change in
this example).
Using the equation for εsolution above, we find (after some algebra) that
ε= εsolution-1 - εsolution-2 = (εHbO2-εHb)*(SatO2-1 – SatO2-2)
This means that for one percentage point change in saturation OD1% = (εHbO2-εHb)*C*d
Now let us calculate the change in ODR, with one percentage point change in saturation.
ODR1% = OD1% / ODisosbestic (because ODisosbestic does not change with saturation).
This means that
ODR1% = (εHbO2-εHb)*C*d / εiososbestic C*d = (εHbO2-εHb) / εiososbestic
In short, the ODR change with 1 percentage point change in saturation is (εHbO2-εHb) / εiososbestic
The following table compares the SLO wavelengths and the fundus camera wavelengths with data
from Zijlstra et al.2 The table uses data for 632nm instead of 633nm and an average of extinction
coefficients for HbO2 and Hb at (near-)isosbestic wavelengths.
Table S1. Extinction coefficients for oxy- and deoxyhemoglobin (HbO2 and Hb) at wavelengths
used for SLO oximetry and fundus camera oximetry.* The table also shows the theoretical change in
ODR with 1 percentage point change in oxygen saturation.
SLO wavelengths
Fundus camera wavelengths
εHbO2
0.129
0.961
εHb
1.022
3.759
εIsosbestic
9.92
11.6
-0.090
-0.241
ODR1% =(εHbO2-εHb) / εiososbestic
*The Oxymap T1 fundus camera oximeter uses 570nm (isosbestic) and 600nm (sensitive) light.
The SLO oximeter tested here uses 532nm (close to isosbestic) and 633nm (sensitive). The table uses
data for 632nm instead of 633nm, which was unavailable.
Table S1 shows that ODR should change about 2.7 times faster in the case of fundus camera with the
same change in saturation. This is approximately in agreement with the ratio of the calibration
coefficients (1.93).
The above calculations assume that one wavelength in each system (532nm in the SLO and 570nm in
the fundus camera) are completely insensitive to changes in saturation. In fact, this is close to being
accurate for 570nm but there is considerable difference in εHbO2 and εHb at 532nm (an average is
taken to calculate εIsosbestic in the table above). What difference does this make for oxygen sensitivity
of the SLO oximeter?
According to Ziljstra et al.2, the extinction coefficients at 532nm are 11.42L*mmol-1*cm-1 for HbO2
and 8.411L*mmol-1*cm-1 for Hb. Therefore, with increasing saturation, the OD at 532nm will increase
slightly while the OD at 633nm decreases. This will increase the rate of change in ODR for the SLO
with change in saturation and counteract somewhat the small sensitivity. It is not ideal to have only
an approximately isosbestic wavelength. However, this fact may help to explain why the ratio of the
calibration coefficients is 1.93 while the theoretical ratio is slightly higher when isosbesticity is
assumed at 532nm.
Basic data for retinal vessel oxygen saturation calculation for the 11 healthy
subjects of the study and variability in ODs along the measured vessel
segments
Tables S2 and S3 below show the means of each variable along the measured vessel segment. An OD
value is for example a mean of OD values many cross sections on each segment (38-162 cross
sections, usually above 100). Comparison is made between the SLO oximeter and the Oxymap T1
fundus camera based oximeter.
Table S2. Arterioles. Table shows means of each variable, for arterioles, along a measured
vessel segment. The upper part shows mean values from the SLO for each individual (n=11). The
lower part shows, for comparison, an average of corresponding values from 10 (different) healthy
individuals, measured with a fundus camera based oximeter.
Subject
I0-532
I532
I0-633
I633
OD532
OD633
ODR
SatO2
1
2
3
4
5
6
7
8
9
10
11
SLO
Average
1034
1051
840
636
876
1226
956
1021
1543
1345
1295
1075
477
555
460
415
515
647
579
622
772
689
731
587
1120
1311
880
631
1356
1065
1378
935
1311
1306
1527
1165
935
1141
791
598
1234
939
1210
861
1099
1086
1324
1020
0.332
0.274
0.256
0.184
0.228
0.273
0.215
0.206
0.298
0.288
0.248
0.255
0.078
0.060
0.046
0.024
0.041
0.055
0.057
0.032
0.076
0.080
0.062
0.055
0.236
0.219
0.173
0.127
0.179
0.200
0.264
0.134
0.255
0.277
0.249
0.210
85.5
89.8
101.2
112.5
99.5
94.5
78.6
110.8
80.7
75.5
82.2
91.9
Fundus
camera
Fundus
camera
average
I0-570
I570
I0-600
I600
OD570
OD600
ODR
SatO2
932
458
2112
1759
0.315
0.083
0.263
90.3
*Note: The pixels from the oximeter do not correspond to the same retinal dimensions.
Width
(pixels)
11.4
9.4
10.2
11.4
7.7
12.0
12.9
12.0
11.9
11.3
11.3
11.0
Width
(pixels)
15.4
Table S3. Venules. Table shows means of each variable, for venules, along a measured vessel
segment. The upper part shows mean values from the SLO for each individual (n=11). The lower part
shows, for comparison, an average of corresponding values from 10 (different) healthy individuals,
measured with a fundus camera based oximeter.
Subject
I0-532
I532
I0-633
I633
OD532
OD633
ODR
SatO2
1
2
3
4
5
6
7
8
9
10
11
Average
639
999
897
591
903
1228
1070
1104
1538
1322
1334
1057
368
475
478
340
483
543
554
471
671
536
635
505
1224
1314
898
609
1410
1013
1400
990
1321
1386
1582
1195
1014
1009
748
514
1098
785
1050
712
977
1025
1276
928
0.227
0.319
0.272
0.237
0.271
0.350
0.281
0.359
0.358
0.390
0.321
0.308
0.081
0.114
0.079
0.073
0.108
0.110
0.124
0.143
0.130
0.131
0.093
0.108
0.370
0.359
0.288
0.309
0.399
0.315
0.422
0.398
0.363
0.336
0.292
0.352
52.3
55.0
72.7
67.5
45.2
65.9
34.6
45.4
54.1
60.8
71.8
56.8
Fundus
camera
Fundus
camera
average
I0-570
I570
I0-600
I600
OD570
OD600
ODR
SatO2
923
330
2165
1243
0.448
0.241
0.538
55.2
*Note: The pixels from the oximeter do not correspond to the same retinal dimensions.
Width
(pixels)
11.0
11.3
11.3
19.7
12.0
11.8
16.5
19.6
13.7
14.8
11.5
13.9
Width
(pixels)
17.8
Tables S4 and S5 show the variability in ODs along the measured vessel segments (coefficient of
variation = (SD/ mean)*100%) for arterioles and venules seperately.
Table S4. Arterioles. Table shows the variability in ODs for arterioles along measured vessel
segments (coefficient of variation = (SD/ mean)*100%). Upper part of the table shows values for the
11 healthy individuals, measurements with the SLO. The lower part shows an average and median
from 10 (different) healthy individuals, measured with the fundus camera based oximeter.
Subject
1
2
3
4
5
6
7
8
9
10
11
Mean
Median
Fundus
camera
Mean
Median
OD532
Mean
OD532
SD
0.332
0.274
0.256
0.184
0.228
0.273
0.215
0.206
0.298
0.288
0.248
0.046
0.054
0.090
0.032
0.050
0.070
0.066
0.076
0.068
0.061
0.037
0.059
0.061
OD570
SD
OD532
OD633
OD633
Coeff. var.
Mean
SD
(%)
13.8
0.078
19.8
0.060
35.3
0.046
17.4
0.024
22.1
0.041
25.7
0.055
30.6
0.057
37.0
0.032
22.9
0.076
21.3
0.080
14.9
0.062
23.7
22.1
OD570
Coeff. var.
(%)
0.052
17.0
0.050
16.5
OD600
SD
0.011
0.014
0.022
0.005
0.009
0.014
0.017
0.027
0.017
0.020
0.009
0.015
0.014
OD633
Coeff. var.
(%)
13.7
24.0
46.9
21.7
22.2
26.0
30.3
84.4
22.8
24.6
14.8
30.1
24.0
OD600
Coeff. var.
(%)
0.018
22.5
0.017
23.1
Table S5. Venules. Table shows the variability in ODs for venules along measured vessel
segments (coefficient of variation = (SD/ mean)*100%). Upper part of the table shows values for the
11 healthy individuals, measurements with the SLO. The lower part shows an average and median
from 10 (different) healthy individuals, measured with the fundus camera based oximeter.
Subject
OD532
Mean
1
2
3
4
5
6
7
8
9
10
11
Mean
Median
Fundus
camera
Mean
Median
OD532
SD
0.227
0.319
0.272
0.237
0.271
0.350
0.281
0.359
0.358
0.390
0.321
0.060
0.048
0.072
0.056
0.035
0.080
0.069
0.094
0.047
0.048
0.049
0.060
0.056
OD570
SD
OD532
OD633
OD633
Coeff. var.
Mean
SD
(%)
26.3
0.081
14.9
0.114
26.4
0.079
23.5
0.073
12.9
0.108
22.8
0.110
24.5
0.124
26.2
0.143
13.1
0.130
12.3
0.131
15.2
0.093
19.8
22.8
OD570
Coeff. var.
(%)
0.044
10.0
0.041
8.4
OD600
SD
OD633
Coeff. var.
0.014
0.016
0.025
0.016
0.014
0.025
0.030
0.037
0.019
0.017
0.014
0.021
0.017
16.6
14.1
31.6
21.6
12.7
23.1
24.0
26.1
14.9
13.2
14.6
19.3
16.6
OD600
Coeff. var.
(%)
0.023
9.8
0.021
8.7
As can be seen from the tables, both the isosbestic and the sensitive wavelength yield more variable
data for the SLO, compared to the fundus camera (variability along the vessels). However, the 633nm
does not yield clearly more variable data then the 532nm.
References
1.
Harris A, Dinn RB, Kagemann L, Rechtman E. A review of methods for human retinal oximetry.
Ophthalmic Surg Lasers Imaging 2003;34:152-164.
2.
Zijlstra WG, Buursma A, Van Assendelft OW. Visible and near infrared absorption spectra of
human and animal haemoglobin - Determination and application. Utrecht: VSP; 2000:368.