Accurate measurement of ocular biometry is critical to providing optimum refractive outcomes postcataract surgery.1 Ultrasound is the traditional technique for measuring anterior chamber depth (ACD) and axial length (AL) but is generally limited to a resolution of about ±0.15 mm.2 3 Partial coherence interferometry has subsequently been developed as an ocular biometry technique.4 5 Since the advent of the first commercial device in 2001 (IOLMaster, Carl Zeiss Jena GmbH), this has become the technique of choice for ocular biometry. Its popularity is due to its non-contact nature, hence avoiding the risk of corneal abrasion and/or contamination, and due to its significantly higher resolution measurements of axial length (about ±0.02 mm; equivalent to 0.05 D).6 It has been shown to be accurate and repeatable in both cataract biometry assessment7 8 and in the study of refractive error.9 10 The IOLMaster thus improved the refractive outcome results of cataract surgery11 12 and by 2002 was used in over a third of hospital eye units in the UK.13 However, the IOLMaster only uses partial coherence interferometry to measure AL; corneal curvature, horizontal iris width (white-to-white) and ACD is assessed with imaging techniques and there is no assessment of corneal, crystalline lens or retinal thickness.10 Each of the IOLMaster’s three assessments also requires realignment of the device with the visual axis of the eye. It fails to measure in up to 20% of eyes with dense opacities and macular disease,8 14 15 although this can be reduced to less than 10% with more advanced analysis of the interference waveform.6 Ultrasound is only prevented from measurement in eyes filled with silicone oil, but partial coherence interferometry is not.14 16

A new ocular biometry device jointly developed by Haag-Streit (LenStar LS900, Haag-Streit Koeniz, Switzerland) and Wavelight (Allegro Biograph, Wavelight, Erlangen, Germany), is now commercially available. It uses optical low coherence reflectometry to measure corneal thickness, ACD, crystalline or intraocular lens thickness as well as AL. The technique was developed in the late 1980s for reflection measurement in telecommunication devices with micrometre resolution and first applied to in vivo biological tissue (the eye) by Fercher and colleagues.17 The LenStar also assesses central corneal curvature, the horizontal iris width (white-to-white), pupil size, and pupil and visual axis decentration by image analysis, without the need for realignment.

The study evaluates the repeatability of LenStar measurements and its validity when compared with the IOLMaster and A-scan applanation ultrasonograph.

METHODS

One-hundred and twelve patients (36 male and 76 female), with a mean age of 76.4 (SD 9.1) years (range from 41 to 96 years, median 77 years), listed for cataract surgery participated in the study. The purpose of the study was explained and informed consent given. All measurements were performed on one eye by a single practitioner for each of the instruments. The study was approved by the National Research Ethics Committee and conformed to the Declaration of Helsinki (2008).

The LensStar, like the IOLMaster, uses the effect of time domain interferometric or coherent superposition of light waves to measure ocular lengths of the eye in a similar technique to one-dimensional optical coherence tomography. The IOLMaster uses a diode laser, whereas the LenStar uses a superluminescent diode with a Gaussian-shaped spectrum which allows a higher axial resolution; hence the terminology optical low coherence reflectometry, rather than partial coherence interferometry, has been coined.

The LenStar was focused and aligned using the image of the eye on the computer monitor while the patient fixated on a flashing red light. The eyes were in focus when the instrument head was approximately 6.8 cm away from the patient’s eyes. Patients were asked to perform a complete blink just before measurements were taken in order to spread an optically smooth tear film over the cornea. The instrument takes 16 consecutive scans per measurement without the need for realignment, and five measurements were taken to test intrasession repeatability (as recommended). The device uses optical low coherence reflectometry to measure corneal thickness, ACD, crystalline or intraocular lens thickness and AL using the 820 μm superluminescent diode. The retinal thickness can also be determined from the scans, but this requires subjective alignment of a cursor and was not assessed in this study. It also uses 950 μm light to assess by image analysis, central corneal topography using two rings of diameter 1.65 mm and 2.30 mm (for an eye of radius 7.8 mm) of 16 light spot each, reflected off the air/tear interface, the horizontal iris width (white-to-white) by fitting the best circle with the lowest error square to the detected edge and pupil size using the same method, and calculates pupil and visual axis decentration with respect to the centre of the cornea as circumscribed by the limbus.

The IOLMaster, running Advanced Technology version 5 software,6 was used to assess the same eyes being focused and aligned using the image of the eye on the computer monitor while the patient viewed the instrument’s internal illuminated targets. The eyes were in focus when the instrument head was approximately 5.5 cm away from the patient. Patients were asked to perform a complete blink just before measurements were taken in order to spread an optically smooth tear film over the cornea. AL was measured by partial coherence interferometry (laser diode infrared light of wavelength 780 μm), ACD through image analysis of the distance between the anterior corneal pole and the anterior surface of the crystalline lens illuminated by an optic section, and corneal curvature by image analysis of the distance between three opposite pairs of light spots, arranged in a 2.3 mm diameter hexagonal pattern, reflected from the air–tear film interface.10 Five separate measurements were averaged for both AL and corneal curvature, whereas a single shot automatically generated and averaged five measurements of ACD.

In a subgroup of 21 patients (five male and 16 female), with a mean age of 78.1 (8.1) years (range 70 to 90 years, median 77.5 years), A-scan applanation ultrasound (OcuScan, Alcon Surgical, Irvine, California) was also performed. The A-scan applanation device calculated ACD, crystalline lens thickness and AL from the time taken for ultrasound waves to reflect back to its receiver from an optical surface.18 One drop of topical anaesthetic, benoxinate HCl 0.4% (Minims, Chauvin Pharmaceuticals, Romford, UK), was instilled in the patient’s eye 2 min before ultrasound measurement. Care was taken in aligning the transducer probe along the optical axis and to exert minimal corneal pressure. Ten measurements were taken for each eye and the mean calculated.

The intersession repeatability of the LenStar was examined by repeating the measurement again in a second session on the same day on 32 of the patients (nine male and 23 female), with a mean age of 73.7 (9.3) years (range from 41 to 87 years, median 74.5 years).

Statistical analysis

The bias between measurements (the mean difference and 95% confidence interval) was calculated and presented graphically.19 The level of agreement between biometry measurements was tested using the Pearson product moment correlation coefficient. Comparisons between measurements were performed using paired two-tailed t tests. Corneal curvatures were analysed in the steepest and flattest meridian in dioptres, using the refractive index 1.332. As the IOLMaster and ultrasonography determine ACD from the front corneal surface, the corneal thickness calculated by the LenStar was added to its anterior chamber measurement from the back surface of the cornea for comparison.

RESULTS

The mean, 95% confidence interval and range of each of the parameters assessed by the LenStar and IOLMaster in this patient population are presented in table 1. Coherence interferometry measurements failed in 10 patients with dense cataract with the LenStar. The IOLMaster could not take partial coherence interferometry measurements in these patients and one additional patient.

A comparison of the difference between the LenStar and IOLMaster or ultrasound measurements for each individual patient compared with the mean was plotted for each biometry component. The white-to-white corneal measurement was similar as assessed by the LenStar compared with the IOLMaster (table 2; fig 1). The LenStar could be expected to read as much as 0.72 mm above to below 0.60 mm the IOLMaster for the white-to-white diameter. Corneal curvature measurements assessed by the LenStar were similar to those determined with the IOLMaster (table 2; fig 2). The LenStar could be expected to read as much as 0.58 D above to 0.68 D below the IOLMaster for corneal curvature. ACD, as measured by the LenStar, was significantly greater than IOLMaster and ultrasound assessment (table 2; fig 3). However, there was no apparent bias with the magnitude of the ACD. The LenStar could be expected to read as much as 0.88 mm above to 0.68 mm below the IOLMaster and 1.53 mm above to 0.89 mm below applanation ultrasound for ACD. Crystalline lens thickness as measured by the LenStar was similar to that determined by ultrasound (table 2; fig 4). However, the variability was high with the LenStar expected to read as much as 1.79 mm above to 1.46 mm below ultrasound measurements for crystalline lens thickness. AL, as measured by the LenStar, was only slightly but statistically greater than IOLMaster. However, the LenStar determined significantly shorter eyes than ultrasound assessment, and there was a bias towards a greater disparity with increasing AL (table 2; fig 5). The LenStar could be expected to read as much as 0.06 mm above to 0.04 mm below the IOLMaster and 0.16 mm above to 0.44 mm below applanation ultrasound for AL.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

White-to-white: difference between LenStar and IOLMaster. The solid line denotes mean and dashed lines 95% confidence intervals. n = 112 eyes.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Corneal curvature: difference between LenStar and IOLMaster in the flattest and steepest meridians. The solid line denotes the mean and dashed lines 95% confidence intervals of the average curvature. n = 112 eyes.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Anterior chamber depth: difference between LenStar and IOLMaster/A-Scan ultrasonography. The solid line denotes the mean and dashed lines 95% confidence intervals. n = 112/21 eyes.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Crystalline lens thickness: difference between LenStar and A-Scan ultrasonography. The solid line denotes the mean and dashed lines 95% confidence intervals. n = 21 eyes.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

Axial length: difference between LenStar and IOLMaster/A-Scan ultrasonography. The solid line denotes the mean and dashed lines 95% confidence intervals. n = 111/21 eyes.

The LenStar intrasession and intersession variability was small, with intersession variability in the average reading being consistently smaller than the intrasession variability between measurements for optical low coherence interferometry and corneal curvature measurements (table 3). This difference remained if only the first two intrasession measurements were assessed compared with the two intersession measurements (pupil size ±0.054; white-to-white ±0.058 mm; flattest corneal curvature ±0.10 D; steepest corneal curvature ±0.13 D; corneal thickness ±0.002 mm; ACD ±0.049 mm; crystalline lens thickness ±0.078 mm; AL ±0.013 mm). The intrasession repeatability could be improved by using the LenStars software functionality—for example, ACD variability halved to ±0.024 mm by excluding the most aberrant value of the five measurements.

DISCUSSION

The study shows the validity, repeatability and clinical utility of optical low coherence reflectometry for assessing ocular biometry compared with instrumentation currently used in clinical practice. Only 10% of patients could not be measured using the LenStar, similar to the proportion found in this and a previous study with the IOLMaster improved waveform algorithm software.6 In general, measurements of length/thickness were larger as measured by the LenStar compared with the IOLMaster. However, the clinical significance of these effects are minor with the 0.01 mm difference in axial length equating to <0.03 D.6 The greater variability when the device was compared with applanation ultrasonography will be in part due to the lower resolution of this technique2 3 and because laser light is reflected from the retinal pigment epithelium, in contrast to ultrasound waves which are reflected from the internal limiting membrane.18 A compensation to more closely reflect ultrasound values can be selected in the LenStar software. The IOLMaster does not use coherent interferometry to measure ACD, instead image analysing the distance between the anterior surface of the cornea and crystalline lens when illuminated by an optical section with a 0.7 mm width slit beam of light through the anterior segment of the eye at an angle of 38° to the visual axis.10 The LenStar detects the anterior and posterior corneal, and anterior crystalline lens peaks in the optical low coherence reflectometry waveform to measure the anterior chamber depth and corneal thickness, which were combined for comparison with the IOLMaster result. The shorter ACD measured by ultrasonography compared with the IOLMaster has previously been reported.20

The LenStar and IOLMaster were found to measure equivalent values for white-to-white and corneal curvature using image analysis. Caution must be taken when using a diopric representation of corneal curvature, as differences in the refractive index attributed to the cornea between the instruments (n = 1.3375 (IOLMaster) and n = 1.332 (LenStar)) would result in a clinically significant difference in average curvature for both medians of 0.76 (0.21) D (p<0.001) in this study population. The LenStar measurements of crystalline lens thickness were not correlated to those recorded by ultrasonography. The larger intrasession variability (±0.33 vs ±0.09 mm) and range of values (2.83 to 5.06 vs 3.72 to 5.38 mm) with ultrasound compared with the LenStar suggest that optical low coherence reflectometry may be the better technique to assess crystalline lens thickness.

Using the recommended intraocular lens power calculation formulae incorporating many of the discussed biometry measurements, the difference between the LenStar and IOLMaster was 0.01 (0.30) D (96% within 0.5 D) for SRK II, 0.16 (0.30) D (87% within 0.5 D) for Hagis (which uses anterior chamber depth, hence the greater difference) and 0.04 (0.24) D (95% within 0.5 D) for Hoffer Q.6 Hence, despite some statistical differences between ocular biometry measurements between the LenStar and current clinical instruments, these were not considered to be clinically significant.

The coefficient of repeatability for intra- and intersession repeatability using the LenStar are impressive (⩽2% of the average value for each biometric measure) and at least comparable with the IOLMaster10 21 and ultrasound.2 3 As expected, using the average of repeated measurements decreases the variability, and this can be further improved by excluding the most divergent of the results as allowed by the functionality of the LenStar software.

Compared with currently used clinical instrumentation, the LenStar provides a comprehensive range of ocular biometry measurements required by newer, more accurate intraocular lens-power calculation formulae.22 In addition, it allows measurements such as corneal thickness (including the functionality of measurement while the patient views internal off-axis illuminated targets at 2 mm and 2.7 mm eccentricity separated by 22.5°), retinal thickness and the decentration between the visual axis and the centre of the cornea. Some of these measurements may improve the accuracy of optimal intraocular lens power prediction or may be useful in assessing the development of refractive error.10 It is therefore envisaged that the LenStar will be well received in both the clinical and research environment due to its high resolution, good validity and repeatability compared with currently used instrumentation, single alignment requirement and non-contact measurement.