Downloaded from bjo.bmj.com on October 13, 2014 - Published by group.bmj.com
BJO Online First, published on October 8, 2014 as 10.1136/bjophthalmol-2014-305183 Clinical science
Correlations among ocular surface temperature difference value, the tear meniscus height, Schirmer’s test and fluorescein tear film break up time Tai Yuan Su,1 Wei Ting Ho,2 Chien Yi Lu,1 Shu Wen Chang,2 Huihua Kenny Chiang1 1
Institute of Biomedical Engineering, National YangMing University, Taipei, Taiwan 2 Department of Ophthalmology, Far Eastern Memorial Hospital, New Taipei City, Taiwan Correspondence to Professor Huihua Kenny Chiang, Institute of Biomedical Engineering, National YangMing University, No. 155, Sec. 2, Linong Street, Beitou District, Taipei 1121, Taiwan; [email protected] HKC and SWC contributed equally. Received 3 March 2014 Revised 8 July 2014 Accepted 21 September 2014
ABSTRACT Purpose To report the use of a thermographer for measuring ocular surface temperature, and to evaluate the correlation among the obtained temperature difference values (TDVs) and dry eye parameters (tear meniscus height (TMH); Schirmer’s test results; fluorescent tear breakup time (FTBUT)). Methods Forty-three participants (age 40.2 ±14.7 years; range 21–67 years) from Far Eastern Memorial Hospital, Taiwan were recruited for the study. The surface temperature was measured at the centre of the ocular surface for 4 s after blinking. TDV was defined as the change in corneal surface temperature relative to that of the preceding eye opening, where TDV01, TDV02, TDV03, and TDV04 represent the values obtained 1, 2, 3, and 4 s after blinking, respectively. Anterior segment optical coherence tomography (AS-OCT) was employed to measure the lower TMH. Schirmer’s test with topical anaesthetic was conducted to measure the basal tear secretion. The FTBUT was recorded using a digital camera. Results TDV measurement exhibited high reliability (intraclass correlation coefficient=0.91). TDV03 exhibited the highest significance and strongest positive correlation with the TMH (r=0.52, p=0.0003) and Schirmer’s test value (r=0.39, p=0.008), whereas the TDV03–FTBUT correlation was non-significant. Age correlated negatively and significantly with the TDV (r= −0.35, p=0.021), TMH (r= −0.33, p=0.031), and Schirmer’s test value (r= −0.31, p=0.044). TDV03 remained significantly correlated with the TMH and Schirmer’s test value after adjustment for age. Conclusions The thermographer was effective in capturing temperature changes in the ocular surface. The temperature difference 3 s after blinking appears to be correlated with lower TMH and Schirmer test values.
INTRODUCTION
To cite: Su TY, Ho WT, Lu CY, et al. Br J Ophthalmol Published Online First: [ please include Day Month Year] doi:10.1136/bjophthalmol2014-305183
Dry eye disease is among the most common eye diseases treated in daily clinics, and it is a disease with a multifactorial pathophysiology.1 2 Because dry eye disease is complex, the definition of dry eye disease has been expanded to account for the increased understanding of the disease entity.3 Traditionally, dry eye disease is categorised as either aqueous tear-deficient or evaporative dry eye. The change in the tear film, which may cause decreased production or excessive evaporation, leads to clinical symptoms and affects the quality of vision. When diagnosing dry eye, physicians examine the patient’s medical history to identify any clinical symptoms and conduct various diagnostic tests, including fluorescein tear breakup time (FTBUT),
ocular surface staining, and Schirmer’s tests. However, these diagnostic tests have limited sensitivity and specificity, and some are invasive, which can influence the test result and cause discomfort to the patients. Thus, developing a non-invasive diagnostic method is necessary. Anterior segment optical coherence tomography (AS-OCT) is a noninvasive diagnostic tool used for quantitative tear evaluation.4–7 Because the tear meniscus contains 75–90% of the total volume of tears, the tear meniscus height (TMH) is directly correlated with the thickness of the tear film.8 Shen et al9 reported that AS-OCT can be used to non-invasively measure the TMH, and TMH has high sensitivity and specificity in the diagnosis of dry eye resulting from Sjögren’s syndrome. However, evaluating the TMH by using AS-OCT can be subjected to interference under certain circumstances, such as when a patient has conjunctivochalasis.10 Tear evaporation measurement has been used to evaluate Meibomian gland dysfunction.11–15 The tear evaporation rate can be measured on a sequence of thermographic images,14 which detect temperature changes in real time across the entire ocular surface.16–19 Thermography after 10 s of eye opening has also been advocated as a simple, noninvasive screening test for dry eyes.16 17 19 Temperature difference value (TDV), a parameter derived using thermography, represents the temperature difference relative to the temperature after the last eye opening.16 Previous studies have indicated that the corneal surface temperature decreases faster in patients with normal eyes than in patients with tear deficiencies.12 13 18 19 However, no study has examined the correlation between the change in ocular surface temperature and tear volume. In this study, a thermographer was employed to measure the ocular surface temperature to obtain the TDV. Furthermore, we evaluated the relationship between the TDV and other tear volume indicators, including the TMH, Schirmer’s test value, and FTBUT.
METHODS All procedures in this study were conducted in accordance with the tenets of the World Medical Association Declaration of Helsinki, and approval was received from the Institution Review Board at Far Eastern Memorial Hospital, Taiwan. Informed consent was obtained from each participant after explaining the research purpose and procedures. Forty-three participants (30 women, 13 men) with a mean age of 40.2±14.7 years (range 21–67 years) in good health and with no history of contact lens
Su TY, employer) et al. Br J Ophthalmol 2014;0:1–6. doi:10.1136/bjophthalmol-2014-305183 Copyright Article author (or their 2014. Produced by BMJ Publishing Group Ltd under licence.
1
Downloaded from bjo.bmj.com on October 13, 2014 - Published by group.bmj.com
Clinical science wear, current ocular diseases, fluorescein allergy, or previously diagnosed dry eye were recruited. Extensive anterior segment examination was done to exclude ocular surface abnormalities, previous ocular surface surgery, and superficial punctate keratopathy. None of the participants received eye drop instillations within 6 h before conducting the measurements. All examinations were performed between 10:00 and 15:00 in a room with constant temperature (22±1.8°C) and humidity (70±5.0%). First, the TMH was measured using AS-OCT, followed by thermographic imaging. The FTBUT was recorded using a slit-lamp-adapted digital recording system, and finally, the Schirmer’s test with topical anaesthetics was conducted to determine basal tear secretion. A 10-min break was allowed between each procedure to minimise the carry-over effect.
Tear meniscus height The left eye of each participant was scanned by conducting AS-OCT using the cornea–anterior module in the RTVue V.4.0.7.5 (OptoVue, Fremont, California, USA). The participants looked directly ahead, and the lower tear meniscus was captured in a vertical scan that was centred on the inferior cornea and lower eyelid immediately following an entire blinking motion. The tear meniscus was defined as the triangular wedge of tear film between the lower lid margin and cornea, and the TMH was evaluated using a digital calliper (figure 1). Each eye was measured three times, with three measurements obtained each time. The means of the measurements were calculated for further analysis.
Thermography measurement procedure For ocular surface thermography, we used an IT-85 ocular surface thermograph (United Integrated Services Co, Taiwan) to capture the image non-invasively (figure 2). The measurements were performed as follows: (1) the participants blinked normally; (2) the eye was closed for 9 s; (3) the participants blinked normally again; (4) the participants held their eyes open for 4 s. An audio recording was used to deliver prerecorded instructions to the participants. The images were captured at a rate of 30 films/s for 13 s. The eye measurements were conducted four times.
Region of interest selection After recording the thermogram, the temperature was measured at the region of interest (ROI). The ROI was selected by viewing a thermal image and marking four bars for the outer canthus (P1), inner canthus (P3), upper eye led (P2), and lower
eye led (P4). The bars were marked without considering any visual obstruction caused by the eyelashes (figure 3). Subsequently, the ROI was defined as a 5 mm circle, the vertical position of ROI was defined by the centre between P2 and P4, and the horizontal position was defined by the centre between P1 and P3 (area=20 mm2, 1000 pixels). The ROI temperature was calculated as the mean of all pixels in the ROI.
ROI tracking After obtaining the thermogram, we developed the ROI tracking algorithm to adjust the position of the ROI in each image. The algorithm is a standard approach for feature detection.20 The inner canthus was selected as the template at the start of the test. The thermogram images recorded over the following 4 s were compared with the template to obtain the best-matched position. Although the eye position shifted, the algorithm ensured that the ROI was adjusted accordingly (figure 4).
Temperature difference value After selecting and tracking the ROI, the mean temperature was calculated for each thermogram to indicate the temperature at the ROI. Figure 5 shows the change in temperature during the measurement, which is detailed as follows: (1) the eye was closed until the participant opened their eye (ie, the period marked (a)); (2) as the participant opened their eye, the temperature increased (ie, the point marked (b)); (3) the participant blinked (ie, the period marked (c)); and (4) the participant kept their eyes open for 4 s (ie, the period marked (d)). The TDV was defined as the difference in the temperature relative to the temperature after the participant opened their eye (ie, the point marked (e)). Four TDV values were calculated to represent the temperature difference at 1 (TVD01), 2 (TDV02), 3 (TDV03), and 4 (TDV04) seconds after the participant opened their eye, and the changes were subsequently analysed.
Schirmer’s test The Schirmer’s test with anaesthesia ( Jones test) was performed as follows. First, one drop of a 0.5% proparacaine hydrochloride ophthalmic solution (Alcaine, Alcon Laboratories, Inc, Belgium) was applied twice at 5-min intervals. After adequate ocular surface anaesthesia, filter paper strips (5 mm×35 mm Whatman no. 41 filter paper) were placed at the outer lower conjunctival fornix for 5 min with the eye closed. Subsequently, the paper strips were removed, and the wetting of the paper was measured.
Figure 1 Tear meniscus height (TMH) measurements. (A) A scan was performed using anterior segment optical coherence tomography immediately after a full blink. Three TMH values were measured at the centre of lower eyelid (LL) (b), and 1 mm on the right (c) and left sides (a). (B) Cornea (CO), LL, and TMH (282 mm).
2
Su TY, et al. Br J Ophthalmol 2014;0:1–6. doi:10.1136/bjophthalmol-2014-305183
Downloaded from bjo.bmj.com on October 13, 2014 - Published by group.bmj.com
Clinical science
Figure 2 The ocular surface thermograph. Thermography captures the ocular surface thermographic image in a non-invasive manner.
Fluorescent tear film break up time test
A 2 μL volume preservative-free solution consisting of 2% fluorescent dye was applied onto the bulbar conjunctiva using a micropipette. Then a slit lamp was used to record the fluorescent break up image, and the time taken for the first dark spot to appear was recorded. Each eye was measured four times. The means of the measurements were calculated for further analysis.
Statistical analysis Data analysis was performed using SPSS V.20 (IBM, Chicago, Illinois, USA). Data were presented as mean±SD. Pearson’s correlation coefficient was calculated to test the correlations among the TDV, TMH, Schirmer’s test result, and FTBUT. The correlations between the TDV and the TMH or Schirmer’s test results were adjusted for age in the linear regression analysis, and the level of statistical significance was set at p0.7.
respectively. Regarding the thermography examination, the mean eye movement along the vertical plane was 0.32 ±0.35 mm (maximum 1.9 mm), and along the horizontal plane it was 0.25±0.38 mm (maximum 2 mm). Following the ROI selection and ROI tracking, the image deviation was adjusted accordingly. The ICC among the four TDV measurements for all participants was 0.91 (95% CI 0.84 to 0.94). Figure 6 shows the TDVs. The TDVs increased constantly from TDV01 to TDV04. The increase from TDV02 to TDV03 was statistically significant, which represents an abrupt decrease in ocular surface temperature in this interval. Table 1 shows the Pearson’s correlation coefficient for the correlations between the four TDVs and the TMH, Schirmer’s test values, FTBUT, and age. TDV01–TDV03 exhibited a positive correlation with TMH, and the highest level of significance was observed between TDV03 and TMH. Although a positive correlation was observed between TDV04 and TMH, the relationship was non-significant ( p=0.1). TDV03 exhibited the most positive and significant correlation with the Schirmer’s test value. However, the correlations were non-significant between all TDVs and the FTBUT. All TDVs were negatively correlated with age, and were statistically significant at TDV02 and TDV03. Figure 7 shows the correlation map for TDV03, the TMH, Schirmer’s test value, and age. All parameters correlated negatively with age. TDV03 positively and significantly correlated with the TMH and Schirmer’s test value, and the correlation remained statistically significant after adjusting for age in the linear regression model. However, the correlation between the TMH and Schirmer’s test value was non-significant.
DISCUSSION
Figure 3 Region of interest (ROI) selection. The selection method assisted in avoiding any visual obstruction resulting from the eyelashes. Four bars were marked at outer canthus (P1), inner canthus (P3), upper eye led (P2), and lower eye led (P4). The ROI was positioned on the middle of P1 and P3 in horizontal direction and on the middle of P2 and P4 in vertical direction.
In this study, we demonstrated that an ocular surface thermographer can be used to monitor changes in ocular temperature. The thermography measurement procedure was fast and noncontact. No reflex tearing was reported, and the procedure did not cause any discomfort to the participants. We also developed an algorithm for ROI tracking and selection, which enabled us to monitor the temperature changes continually in a selected area to obtain the TDVs. Our results show that the TDV measurements exhibited high ICCs, and TDV03 correlated significantly with the TMH and Schirmer’s test value. TDVs are the change in ocular temperature, and temperature change results from tear film evaporation; thus, TDVs may relate to the change in tear volume and the evaporation of the tear film. Several features of the ocular surface thermographer system were arranged to achieve reliable and consistent TDV measurement results. First, the ROI selection method was used to identify the centre of the ocular surface; the 5 mm ROI diameter identified the centre of the ocular surface area without interference from the eyelashes, and without causing reflex tearing while the eye was open. This approach can be applied to increase the repeatability of the measurement. Second, an ROI tracking method was adopted to minimise the eye movement effect. The degree of movement was up to 1.9 mm during the measurements, which was relatively large compared to the size of the ROI. To improve the measurement accuracy, we developed a tracking method to reduce the effect of eye movement during the measurements. The ROI tracking algorithm assisted in obtaining accurate measurements. Third, we employed a standardised protocol to instruct the participants on when to blink and open their eyes. Previous studies have shown that blinking can interfere with the ocular surface temperature measurement, because higher blinking rates in patients with dry eye can
Su TY, et al. Br J Ophthalmol 2014;0:1–6. doi:10.1136/bjophthalmol-2014-305183
3
RESULTS The mean TMH, basal secretion Schirmer’s test value, and FTBUT were 218.6±94.8 mm (range 55–516 mm), 7.0±5.1 mm (range 2–28 mm) after 5 min, and 3.8±3.3 s (range 1–15 s),
Downloaded from bjo.bmj.com on October 13, 2014 - Published by group.bmj.com
Clinical science
Figure 4 Region of interest (ROI) tracking method. (A) Initial thermogram with the selected template and an ROI. The relative position of the template and the ROI was fixed. (B) Shifted thermogram at the third second. The template-matching method was used to detect the inner canthus of the image, and adjust the ROI accordingly. The arrow (y) indicates the magnitude of displacement.
resulted in an increase in ocular surface temperature.21 In our study, an audio speaker was used to broadcast a sequence of instructions for the participants, and the participants followed the instructions (ie, close your eyes for 9 s, open your eyes, blink once immediately, and then keep your eyes open for 4 s). This protocol potentially reduced any confounding effects caused by patients aberrantly blinking and opening their eyes; also keeping eyes open for relatively shorter time can reduce the reflex tearing. Employing these standardised measurement procedures for thermography systems improves the consistency of results. In this study, the ICC was 0.91, indicating that the measured TDVs were highly reliable. In this study, we examined the relationships among TDVs, the TMH, Schirmer’s values, and age. TDV03 exhibited the most significant and positive correlation with the TMH and Schirmer’s test (figure 7). Both TMH and Schirmer tests are used for tear evaluation in dry eye patients.4–7 However, these examinations are subject to certain limitations. For example, conjunctivochalasis can occupy the tear pool, which can change the TMH. An excessively large conjunctival fold may interfere with the measurement of the TMH using AS-OCT. TDVs are measurements of temperature change at the ocular surface, and provide real-time information. Furthermore, we selected an ROI at the centre of the ocular surface, thereby reducing the interference from the eyelashes or conjunctival fold. Because TDV03
correlated positively with both the TMH and Schirmer’s test value (table 1), TDV03 can be a reliable parameter for tear film evaluation. Moreover, because tear volume can represent the extent of dry eye disease.4–6 22 23 the significant correlation between TDV03 and the TMH and Schirmer’s test result implies that TDV03 is suitable for evaluating tear volume in future studies. There are some limitations in the present study. The TDV and TMH were measured at different times. The two measurements were performed sequentially with a 10-min break to minimise the carry-over effect. However, the tear film condition could have changed between the two measurements. If the TDV03 and TMH were measured simultaneously, the correlation between them could be strengthened. However, because TDV03 and the TMH were both non-invasive measurement results, the adverse confounding effects may have been minimised. In the normal participants the FTBUT, Schirmer and TMH were within a small range, therefore the relation between these tests and temperature differences were not found. In the present study, a group of participants with a broad age range was tested and the results of the FTBUT, TMH, and Schirmer test were found to range widely. Further studies with large sample sizes including subgroups such as dry eye are warranted. In our study, all parameters correlated negatively and significantly with age. Previous studies have shown that the TMH and Schirmer’s test values decrease as patients increase in age.24 25 Because TDV03 correlated with age (r= −0.35, p=0.021), verifying the correlation between TDV03 and other parameters after adjustment for age is crucial. The linear regression model results confirmed that TDV03 remained positively
Figure 5 Change in mean temperature at the region of interest (ROI) versus time, where (A) represents the change in the temperature while the eye was initially closed, (B) represents where the eye was first fully opened, (C) represents the period in which the participant blinked, (D) represents the following 4 s during which the eye was opened, and (E) represents the temperature difference values (TDVs).
Figure 6 Temperature difference values (TDVs) at 1–4 s after the final eye opening (TDV01–04). A significant increase was observed between TDV02 and TDV03.
4
Su TY, et al. Br J Ophthalmol 2014;0:1–6. doi:10.1136/bjophthalmol-2014-305183
Downloaded from bjo.bmj.com on October 13, 2014 - Published by group.bmj.com
Clinical science Table 1 Correlations between the TDVs and TMH, Schirmer’s test value, FTBUT, and age
TDV01 r p TDV02 r p TDV03 r p TDV04 r p
TMH
Schirmer
FTBUT
Age
0.39* 0.010
0.240 0.121
−0.01 0.53
−0.230 0.140
0.453* 0.002
0.363* 0.017
−0.09 0.583
−0.324* 0.034
0.523* 0.0003
0.39* 0.008
−0.02 0.991
−0.350* 0.021
0.207 0.1
0.461* 0.012
−0.20 0.301
−0.310 0.103
FTBUT, fluorescent tear film break up time; r, Pearson’s coefficient correlation; TDV, temperature difference value; TDV01–04, TDV value at 1–4 s after a blinking; TMH, tear meniscus height. *p
BJO Online First, published on October 8, 2014 as 10.1136/bjophthalmol-2014-305183 Clinical science
Correlations among ocular surface temperature difference value, the tear meniscus height, Schirmer’s test and fluorescein tear film break up time Tai Yuan Su,1 Wei Ting Ho,2 Chien Yi Lu,1 Shu Wen Chang,2 Huihua Kenny Chiang1 1
Institute of Biomedical Engineering, National YangMing University, Taipei, Taiwan 2 Department of Ophthalmology, Far Eastern Memorial Hospital, New Taipei City, Taiwan Correspondence to Professor Huihua Kenny Chiang, Institute of Biomedical Engineering, National YangMing University, No. 155, Sec. 2, Linong Street, Beitou District, Taipei 1121, Taiwan; [email protected] HKC and SWC contributed equally. Received 3 March 2014 Revised 8 July 2014 Accepted 21 September 2014
ABSTRACT Purpose To report the use of a thermographer for measuring ocular surface temperature, and to evaluate the correlation among the obtained temperature difference values (TDVs) and dry eye parameters (tear meniscus height (TMH); Schirmer’s test results; fluorescent tear breakup time (FTBUT)). Methods Forty-three participants (age 40.2 ±14.7 years; range 21–67 years) from Far Eastern Memorial Hospital, Taiwan were recruited for the study. The surface temperature was measured at the centre of the ocular surface for 4 s after blinking. TDV was defined as the change in corneal surface temperature relative to that of the preceding eye opening, where TDV01, TDV02, TDV03, and TDV04 represent the values obtained 1, 2, 3, and 4 s after blinking, respectively. Anterior segment optical coherence tomography (AS-OCT) was employed to measure the lower TMH. Schirmer’s test with topical anaesthetic was conducted to measure the basal tear secretion. The FTBUT was recorded using a digital camera. Results TDV measurement exhibited high reliability (intraclass correlation coefficient=0.91). TDV03 exhibited the highest significance and strongest positive correlation with the TMH (r=0.52, p=0.0003) and Schirmer’s test value (r=0.39, p=0.008), whereas the TDV03–FTBUT correlation was non-significant. Age correlated negatively and significantly with the TDV (r= −0.35, p=0.021), TMH (r= −0.33, p=0.031), and Schirmer’s test value (r= −0.31, p=0.044). TDV03 remained significantly correlated with the TMH and Schirmer’s test value after adjustment for age. Conclusions The thermographer was effective in capturing temperature changes in the ocular surface. The temperature difference 3 s after blinking appears to be correlated with lower TMH and Schirmer test values.
INTRODUCTION
To cite: Su TY, Ho WT, Lu CY, et al. Br J Ophthalmol Published Online First: [ please include Day Month Year] doi:10.1136/bjophthalmol2014-305183
Dry eye disease is among the most common eye diseases treated in daily clinics, and it is a disease with a multifactorial pathophysiology.1 2 Because dry eye disease is complex, the definition of dry eye disease has been expanded to account for the increased understanding of the disease entity.3 Traditionally, dry eye disease is categorised as either aqueous tear-deficient or evaporative dry eye. The change in the tear film, which may cause decreased production or excessive evaporation, leads to clinical symptoms and affects the quality of vision. When diagnosing dry eye, physicians examine the patient’s medical history to identify any clinical symptoms and conduct various diagnostic tests, including fluorescein tear breakup time (FTBUT),
ocular surface staining, and Schirmer’s tests. However, these diagnostic tests have limited sensitivity and specificity, and some are invasive, which can influence the test result and cause discomfort to the patients. Thus, developing a non-invasive diagnostic method is necessary. Anterior segment optical coherence tomography (AS-OCT) is a noninvasive diagnostic tool used for quantitative tear evaluation.4–7 Because the tear meniscus contains 75–90% of the total volume of tears, the tear meniscus height (TMH) is directly correlated with the thickness of the tear film.8 Shen et al9 reported that AS-OCT can be used to non-invasively measure the TMH, and TMH has high sensitivity and specificity in the diagnosis of dry eye resulting from Sjögren’s syndrome. However, evaluating the TMH by using AS-OCT can be subjected to interference under certain circumstances, such as when a patient has conjunctivochalasis.10 Tear evaporation measurement has been used to evaluate Meibomian gland dysfunction.11–15 The tear evaporation rate can be measured on a sequence of thermographic images,14 which detect temperature changes in real time across the entire ocular surface.16–19 Thermography after 10 s of eye opening has also been advocated as a simple, noninvasive screening test for dry eyes.16 17 19 Temperature difference value (TDV), a parameter derived using thermography, represents the temperature difference relative to the temperature after the last eye opening.16 Previous studies have indicated that the corneal surface temperature decreases faster in patients with normal eyes than in patients with tear deficiencies.12 13 18 19 However, no study has examined the correlation between the change in ocular surface temperature and tear volume. In this study, a thermographer was employed to measure the ocular surface temperature to obtain the TDV. Furthermore, we evaluated the relationship between the TDV and other tear volume indicators, including the TMH, Schirmer’s test value, and FTBUT.
METHODS All procedures in this study were conducted in accordance with the tenets of the World Medical Association Declaration of Helsinki, and approval was received from the Institution Review Board at Far Eastern Memorial Hospital, Taiwan. Informed consent was obtained from each participant after explaining the research purpose and procedures. Forty-three participants (30 women, 13 men) with a mean age of 40.2±14.7 years (range 21–67 years) in good health and with no history of contact lens
Su TY, employer) et al. Br J Ophthalmol 2014;0:1–6. doi:10.1136/bjophthalmol-2014-305183 Copyright Article author (or their 2014. Produced by BMJ Publishing Group Ltd under licence.
1
Downloaded from bjo.bmj.com on October 13, 2014 - Published by group.bmj.com
Clinical science wear, current ocular diseases, fluorescein allergy, or previously diagnosed dry eye were recruited. Extensive anterior segment examination was done to exclude ocular surface abnormalities, previous ocular surface surgery, and superficial punctate keratopathy. None of the participants received eye drop instillations within 6 h before conducting the measurements. All examinations were performed between 10:00 and 15:00 in a room with constant temperature (22±1.8°C) and humidity (70±5.0%). First, the TMH was measured using AS-OCT, followed by thermographic imaging. The FTBUT was recorded using a slit-lamp-adapted digital recording system, and finally, the Schirmer’s test with topical anaesthetics was conducted to determine basal tear secretion. A 10-min break was allowed between each procedure to minimise the carry-over effect.
Tear meniscus height The left eye of each participant was scanned by conducting AS-OCT using the cornea–anterior module in the RTVue V.4.0.7.5 (OptoVue, Fremont, California, USA). The participants looked directly ahead, and the lower tear meniscus was captured in a vertical scan that was centred on the inferior cornea and lower eyelid immediately following an entire blinking motion. The tear meniscus was defined as the triangular wedge of tear film between the lower lid margin and cornea, and the TMH was evaluated using a digital calliper (figure 1). Each eye was measured three times, with three measurements obtained each time. The means of the measurements were calculated for further analysis.
Thermography measurement procedure For ocular surface thermography, we used an IT-85 ocular surface thermograph (United Integrated Services Co, Taiwan) to capture the image non-invasively (figure 2). The measurements were performed as follows: (1) the participants blinked normally; (2) the eye was closed for 9 s; (3) the participants blinked normally again; (4) the participants held their eyes open for 4 s. An audio recording was used to deliver prerecorded instructions to the participants. The images were captured at a rate of 30 films/s for 13 s. The eye measurements were conducted four times.
Region of interest selection After recording the thermogram, the temperature was measured at the region of interest (ROI). The ROI was selected by viewing a thermal image and marking four bars for the outer canthus (P1), inner canthus (P3), upper eye led (P2), and lower
eye led (P4). The bars were marked without considering any visual obstruction caused by the eyelashes (figure 3). Subsequently, the ROI was defined as a 5 mm circle, the vertical position of ROI was defined by the centre between P2 and P4, and the horizontal position was defined by the centre between P1 and P3 (area=20 mm2, 1000 pixels). The ROI temperature was calculated as the mean of all pixels in the ROI.
ROI tracking After obtaining the thermogram, we developed the ROI tracking algorithm to adjust the position of the ROI in each image. The algorithm is a standard approach for feature detection.20 The inner canthus was selected as the template at the start of the test. The thermogram images recorded over the following 4 s were compared with the template to obtain the best-matched position. Although the eye position shifted, the algorithm ensured that the ROI was adjusted accordingly (figure 4).
Temperature difference value After selecting and tracking the ROI, the mean temperature was calculated for each thermogram to indicate the temperature at the ROI. Figure 5 shows the change in temperature during the measurement, which is detailed as follows: (1) the eye was closed until the participant opened their eye (ie, the period marked (a)); (2) as the participant opened their eye, the temperature increased (ie, the point marked (b)); (3) the participant blinked (ie, the period marked (c)); and (4) the participant kept their eyes open for 4 s (ie, the period marked (d)). The TDV was defined as the difference in the temperature relative to the temperature after the participant opened their eye (ie, the point marked (e)). Four TDV values were calculated to represent the temperature difference at 1 (TVD01), 2 (TDV02), 3 (TDV03), and 4 (TDV04) seconds after the participant opened their eye, and the changes were subsequently analysed.
Schirmer’s test The Schirmer’s test with anaesthesia ( Jones test) was performed as follows. First, one drop of a 0.5% proparacaine hydrochloride ophthalmic solution (Alcaine, Alcon Laboratories, Inc, Belgium) was applied twice at 5-min intervals. After adequate ocular surface anaesthesia, filter paper strips (5 mm×35 mm Whatman no. 41 filter paper) were placed at the outer lower conjunctival fornix for 5 min with the eye closed. Subsequently, the paper strips were removed, and the wetting of the paper was measured.
Figure 1 Tear meniscus height (TMH) measurements. (A) A scan was performed using anterior segment optical coherence tomography immediately after a full blink. Three TMH values were measured at the centre of lower eyelid (LL) (b), and 1 mm on the right (c) and left sides (a). (B) Cornea (CO), LL, and TMH (282 mm).
2
Su TY, et al. Br J Ophthalmol 2014;0:1–6. doi:10.1136/bjophthalmol-2014-305183
Downloaded from bjo.bmj.com on October 13, 2014 - Published by group.bmj.com
Clinical science
Figure 2 The ocular surface thermograph. Thermography captures the ocular surface thermographic image in a non-invasive manner.
Fluorescent tear film break up time test
A 2 μL volume preservative-free solution consisting of 2% fluorescent dye was applied onto the bulbar conjunctiva using a micropipette. Then a slit lamp was used to record the fluorescent break up image, and the time taken for the first dark spot to appear was recorded. Each eye was measured four times. The means of the measurements were calculated for further analysis.
Statistical analysis Data analysis was performed using SPSS V.20 (IBM, Chicago, Illinois, USA). Data were presented as mean±SD. Pearson’s correlation coefficient was calculated to test the correlations among the TDV, TMH, Schirmer’s test result, and FTBUT. The correlations between the TDV and the TMH or Schirmer’s test results were adjusted for age in the linear regression analysis, and the level of statistical significance was set at p0.7.
respectively. Regarding the thermography examination, the mean eye movement along the vertical plane was 0.32 ±0.35 mm (maximum 1.9 mm), and along the horizontal plane it was 0.25±0.38 mm (maximum 2 mm). Following the ROI selection and ROI tracking, the image deviation was adjusted accordingly. The ICC among the four TDV measurements for all participants was 0.91 (95% CI 0.84 to 0.94). Figure 6 shows the TDVs. The TDVs increased constantly from TDV01 to TDV04. The increase from TDV02 to TDV03 was statistically significant, which represents an abrupt decrease in ocular surface temperature in this interval. Table 1 shows the Pearson’s correlation coefficient for the correlations between the four TDVs and the TMH, Schirmer’s test values, FTBUT, and age. TDV01–TDV03 exhibited a positive correlation with TMH, and the highest level of significance was observed between TDV03 and TMH. Although a positive correlation was observed between TDV04 and TMH, the relationship was non-significant ( p=0.1). TDV03 exhibited the most positive and significant correlation with the Schirmer’s test value. However, the correlations were non-significant between all TDVs and the FTBUT. All TDVs were negatively correlated with age, and were statistically significant at TDV02 and TDV03. Figure 7 shows the correlation map for TDV03, the TMH, Schirmer’s test value, and age. All parameters correlated negatively with age. TDV03 positively and significantly correlated with the TMH and Schirmer’s test value, and the correlation remained statistically significant after adjusting for age in the linear regression model. However, the correlation between the TMH and Schirmer’s test value was non-significant.
DISCUSSION
Figure 3 Region of interest (ROI) selection. The selection method assisted in avoiding any visual obstruction resulting from the eyelashes. Four bars were marked at outer canthus (P1), inner canthus (P3), upper eye led (P2), and lower eye led (P4). The ROI was positioned on the middle of P1 and P3 in horizontal direction and on the middle of P2 and P4 in vertical direction.
In this study, we demonstrated that an ocular surface thermographer can be used to monitor changes in ocular temperature. The thermography measurement procedure was fast and noncontact. No reflex tearing was reported, and the procedure did not cause any discomfort to the participants. We also developed an algorithm for ROI tracking and selection, which enabled us to monitor the temperature changes continually in a selected area to obtain the TDVs. Our results show that the TDV measurements exhibited high ICCs, and TDV03 correlated significantly with the TMH and Schirmer’s test value. TDVs are the change in ocular temperature, and temperature change results from tear film evaporation; thus, TDVs may relate to the change in tear volume and the evaporation of the tear film. Several features of the ocular surface thermographer system were arranged to achieve reliable and consistent TDV measurement results. First, the ROI selection method was used to identify the centre of the ocular surface; the 5 mm ROI diameter identified the centre of the ocular surface area without interference from the eyelashes, and without causing reflex tearing while the eye was open. This approach can be applied to increase the repeatability of the measurement. Second, an ROI tracking method was adopted to minimise the eye movement effect. The degree of movement was up to 1.9 mm during the measurements, which was relatively large compared to the size of the ROI. To improve the measurement accuracy, we developed a tracking method to reduce the effect of eye movement during the measurements. The ROI tracking algorithm assisted in obtaining accurate measurements. Third, we employed a standardised protocol to instruct the participants on when to blink and open their eyes. Previous studies have shown that blinking can interfere with the ocular surface temperature measurement, because higher blinking rates in patients with dry eye can
Su TY, et al. Br J Ophthalmol 2014;0:1–6. doi:10.1136/bjophthalmol-2014-305183
3
RESULTS The mean TMH, basal secretion Schirmer’s test value, and FTBUT were 218.6±94.8 mm (range 55–516 mm), 7.0±5.1 mm (range 2–28 mm) after 5 min, and 3.8±3.3 s (range 1–15 s),
Downloaded from bjo.bmj.com on October 13, 2014 - Published by group.bmj.com
Clinical science
Figure 4 Region of interest (ROI) tracking method. (A) Initial thermogram with the selected template and an ROI. The relative position of the template and the ROI was fixed. (B) Shifted thermogram at the third second. The template-matching method was used to detect the inner canthus of the image, and adjust the ROI accordingly. The arrow (y) indicates the magnitude of displacement.
resulted in an increase in ocular surface temperature.21 In our study, an audio speaker was used to broadcast a sequence of instructions for the participants, and the participants followed the instructions (ie, close your eyes for 9 s, open your eyes, blink once immediately, and then keep your eyes open for 4 s). This protocol potentially reduced any confounding effects caused by patients aberrantly blinking and opening their eyes; also keeping eyes open for relatively shorter time can reduce the reflex tearing. Employing these standardised measurement procedures for thermography systems improves the consistency of results. In this study, the ICC was 0.91, indicating that the measured TDVs were highly reliable. In this study, we examined the relationships among TDVs, the TMH, Schirmer’s values, and age. TDV03 exhibited the most significant and positive correlation with the TMH and Schirmer’s test (figure 7). Both TMH and Schirmer tests are used for tear evaluation in dry eye patients.4–7 However, these examinations are subject to certain limitations. For example, conjunctivochalasis can occupy the tear pool, which can change the TMH. An excessively large conjunctival fold may interfere with the measurement of the TMH using AS-OCT. TDVs are measurements of temperature change at the ocular surface, and provide real-time information. Furthermore, we selected an ROI at the centre of the ocular surface, thereby reducing the interference from the eyelashes or conjunctival fold. Because TDV03
correlated positively with both the TMH and Schirmer’s test value (table 1), TDV03 can be a reliable parameter for tear film evaluation. Moreover, because tear volume can represent the extent of dry eye disease.4–6 22 23 the significant correlation between TDV03 and the TMH and Schirmer’s test result implies that TDV03 is suitable for evaluating tear volume in future studies. There are some limitations in the present study. The TDV and TMH were measured at different times. The two measurements were performed sequentially with a 10-min break to minimise the carry-over effect. However, the tear film condition could have changed between the two measurements. If the TDV03 and TMH were measured simultaneously, the correlation between them could be strengthened. However, because TDV03 and the TMH were both non-invasive measurement results, the adverse confounding effects may have been minimised. In the normal participants the FTBUT, Schirmer and TMH were within a small range, therefore the relation between these tests and temperature differences were not found. In the present study, a group of participants with a broad age range was tested and the results of the FTBUT, TMH, and Schirmer test were found to range widely. Further studies with large sample sizes including subgroups such as dry eye are warranted. In our study, all parameters correlated negatively and significantly with age. Previous studies have shown that the TMH and Schirmer’s test values decrease as patients increase in age.24 25 Because TDV03 correlated with age (r= −0.35, p=0.021), verifying the correlation between TDV03 and other parameters after adjustment for age is crucial. The linear regression model results confirmed that TDV03 remained positively
Figure 5 Change in mean temperature at the region of interest (ROI) versus time, where (A) represents the change in the temperature while the eye was initially closed, (B) represents where the eye was first fully opened, (C) represents the period in which the participant blinked, (D) represents the following 4 s during which the eye was opened, and (E) represents the temperature difference values (TDVs).
Figure 6 Temperature difference values (TDVs) at 1–4 s after the final eye opening (TDV01–04). A significant increase was observed between TDV02 and TDV03.
4
Su TY, et al. Br J Ophthalmol 2014;0:1–6. doi:10.1136/bjophthalmol-2014-305183
Downloaded from bjo.bmj.com on October 13, 2014 - Published by group.bmj.com
Clinical science Table 1 Correlations between the TDVs and TMH, Schirmer’s test value, FTBUT, and age
TDV01 r p TDV02 r p TDV03 r p TDV04 r p
TMH
Schirmer
FTBUT
Age
0.39* 0.010
0.240 0.121
−0.01 0.53
−0.230 0.140
0.453* 0.002
0.363* 0.017
−0.09 0.583
−0.324* 0.034
0.523* 0.0003
0.39* 0.008
−0.02 0.991
−0.350* 0.021
0.207 0.1
0.461* 0.012
−0.20 0.301
−0.310 0.103
FTBUT, fluorescent tear film break up time; r, Pearson’s coefficient correlation; TDV, temperature difference value; TDV01–04, TDV value at 1–4 s after a blinking; TMH, tear meniscus height. *p
