Secondary Logo

Journal Logo


Using continuous intraoperative optical coherence tomography to classify swirling lens fragments during cataract surgery and to predict their impact on corneal endothelial cell damage

Amir-Asgari, Sahand MD; Hirnschall, Nino MD, PhD; Findl, Oliver MD, MBA*

Author Information
Journal of Cataract & Refractive Surgery: July 2016 - Volume 42 - Issue 7 - p 1029-1036
doi: 10.1016/j.jcrs.2016.04.029
  • Free


One of the more severe complications during phacoemulsification is damage to the corneal endothelium. This damage can arise as a result of contact with lens fragments after turbulent flow of the irrigating solution, resulting in corneal damage, corneal edema, and inflammation.1–3 Alternatively, factors such as high ultrasound (US) energy levels required for a hard nucleus,3,4 prolonged duration of phacoemulsification,3,4 a localized temperature increase, or air bubbles3 can lead to a loss of endothelial cells.

The aim of this study was to quantify and classify swirling lens fragments intraoperatively using a prototype of an intraoperative continuous optical coherence tomography (OCT) device (Carl Zeiss Meditec AG) and to develop a regression model to predict the impact of different fragment parameters on endothelial cell loss, central corneal thickness (CCT), and anterior chamber flare.

Patients and methods

This prospective study included patients who were scheduled for cataract surgery. The phaco time and energy during phacoemulsification were noted; however, no preoperative assessment of the cataract density was performed. All the research and measurements followed the tenets of the Declaration of Helsinki and were approved by the local ethics committee. Written informed consent was obtained from all patients in the study.

Exclusion criteria were previous ocular surgery or trauma, corneal dystrophies, a low baseline endothelial cell count (ECC), pseudoexfoliation syndrome, and a preoperative Snellen visual acuity of less than 0.05.

Patient Assessment

Preoperatively, the eye to be operated on was examined at the slitlamp and optical biometry was performed (IOLMaster 500, Carl Zeiss Meditec AG). The ECC was measured with a specular microscope (Seaeagle, Rhinetec) preoperatively and 1 month postoperatively. In addition, the CCT (ACMaster, Carl Zeiss Meditec AG) and aqueous flare (Kowa FM-600, Kowa Co. Ltd.) were measured preoperatively and 1 hour, 1 day, and 1 month postoperatively. For all parameters, at least 3 consecutive measurements were performed and the median was used for further analysis.

Surgical and Measurement Technique

Surgery was performed under topical anesthesia in all cases by the same experienced surgeon (O.F.). A self-sealing clear corneal incision was created with a 2.8 mm single-beveled steel blade. This was followed by the injection of sodium hyaluronate 1.4% (Healon GV), capsulorhexis, phacoemulsification using a horizontal chopping technique, and coaxial irrigation/aspiration of cortical material. In all cases, a hydrophobic acrylic IOL (ZCB00, Abbott Medical Optics, Inc.) was implanted after ophthalmic viscoelastic device instillation using an injector (Abbott Medical Optics, Inc.). Postoperative therapy within the first month was bromfenac 0.9% (Yellox) twice daily.

During surgery, continuous intraoperative anterior segment OCT imaging was performed and digitally recorded. In short, the setup was a prototype that allows continuous intraoperative measurements and that consists of a time-domain OCT system (Visante) connected to the operating microscope (OPMI 200) (both Carl Zeiss Meditec AG).5,6 To ensure imaging was at the center of the cornea during cataract surgery, a crosshair was introduced into the eyepiece of the operating microscope so that the surgeon knew the exact location of the OCT scan during surgery. During surgery, the surgeon guided the patient to look straight into the operating microscope light and then centered the crosshair on the corneal apex and the Purkinje I reflex. For further analysis, the continuous intraoperative OCT measurements were recorded together with the synchronized “2-dimension (2-D) view” video of the surgeon’s view without the crosshair. To detect whether the 2-D video was aligned in relation to the reticle, the 2-D image was analyzed in an additional step. Therefore, a reticle printed on a transparent template was aligned to the center of the 2-D image. The 2-D image was considered aligned only if the center of the 2-D image and the origin of the reticle were aligned (Figure 1); only these images were further analyzed. Otherwise, the image was considered off center and not analyzed. The center of the 2-D image was defined as the center of the limbus.

Figure 1
Figure 1:
Screenshot of continuous intraoperative OCT device taken during phacoemulsification. The upper 2 images are continuous intraoperative OCT screenshots taken at the same time, but 90 degrees separated from each other (90 degrees and 180 degrees). The lower image shows the simultaneously recorded 2-D video of the surgeon’s view. Before each image was analyzed, the 2-D view screenshot was checked for alignment, as shown by the dotted blue arrows.

All continuous measurements (videos) were analyzed after screenshots were taken at the timepoints of interest. To measure the distances in pixels within the scans, all screenshots were imported into Photoshop CS4 (Adobe Systems, Inc.) and the values were then converted into millimeters.

Fragment Score

The aim of the study was to develop a score that allows prediction of the loss of endothelial cells caused by swirling lens fragments during cataract surgery. Therefore, templates were created using 5 randomly chosen reference continuous intraoperative OCT cataract surgery videos. These 5 videos were analyzed to define typical characteristics of swirling lens fragments hitting the endothelium.

Position of the Swirling Fragment Hit

The position of the swirling fragment hit was defined as central or peripheral. Therefore, the mean corneal diameter in the study sample (10.25 mm length) was used and divided into 3 equally large sectors (Figure 2A).

Figure 2A
Figure 2A:
Same setting as in Figure 1, but using a transparency to classify swirling fragments hitting the cornea centrally, or peripherally.

Fragment Size

Swirling fragments larger than 1.0 mm were classified as large, and fragments 1.0 mm or smaller were classified as small (Figure 2B, top).

Figure 2B
Figure 2B:
Top: Example of a large swirling fragment. The red double-arrow depicts 1 mm. Bottom: Example of the size of a swirling fragment’s contact area with the cornea.

Size of the Contact Area on the Cornea

The size of the contact area was pinpoint or larger (Figure 2B, bottom).

Duration of the Hit

The hit duration of the swirling fragment and the endothelium was classified as brief (up to 1 second contact time) or long (contact time of more than 1 second).

Statistical Analysis

For statistical analysis, Excel 2011 for Mac software (Microsoft Corp.) with a Statplus:mac plug-in (version, Analystsoft, Inc.) and an Xlstat 2012 plug-in (Addinsoft) was used. In addition, SPSS software (version 21.0, International Business Machines Corp.) was used for boxplots. For missing data, observations were excluded from analysis. Descriptive data are always shown as the mean, standard deviation, and range. Furthermore, multiple linear regression modeling was used to assess the predictive power of explanatory variables.


Forty eyes of 40 patients were recruited for this study and 5 patients were lost to follow-up. Of the remaining 35 patients, 17 (49%) were women and 18 (51%) were men and the mean age was 75.0 ± 10.8 years (range 41 to 91 years). More right eyes than left eyes (21 versus 14) were included in this study.

In total, 104 swirling fragments coming into contact with the corneal endothelium were observed and analyzed. A mean of 2.6 swirling fragments (range 0 to 6 fragments) were observed per eye. Table 1 lists all 104 swirling fragments and their classification.

Table 1
Table 1:
Listing of all 104 swirling fragments and their classification.

Endothelial Cell Count

The mean ECC was 2325 ±347 cells/mm2 (range 1457 to 2882 cells/mm2) preoperatively and 2276 ± 371 cells/mm2 (range 1452 to 2948 cells/mm2) 1 month postoperatively (Figure 3). The mean ECC loss was 49 cells/mm2 (mean 2.1% loss ± 290%; range −703 to +414 cells/mm2). This change in cell count difference was not statistically significant (P = .32, Wilcoxon signed-rank test).

Figure 3
Figure 3:
Endothelial cell count over time.

Multiple linear regression for postoperative ECC was found to be significant for position (P = .03) and fragment size (P = .01). Including these 2 parameters, the regression coefficient was r2 = 0.6 (Figure 4).

Figure 4
Figure 4:
Regression model for postoperatively measured endothelial cell change including fragment size and fragment position (ECC = endothelial cell count).

Central Corneal Thickness

Table 2 shows the mean CCT at different timepoints. The mean CCT change from baseline to 1 hour postoperatively, from baseline to 1 day postoperatively, and from baseline to 1 month postoperatively was 49 μm (mean increase 9.0% ± 33.0%; range 7 to 138 μm), 3.8 μm (mean increase 1.1% ± 12.3%; range −25 to 32 μm), and 2.3 μm (mean increase 0.5% ± 13.2%; range −22 to 58 μm) (Figure 5). The mean CCT increased significantly from preoperatively to 1 hour postoperatively and decreased significantly from 1 hour to 1 day postoperatively (both P < .01). None of the fragment score parameters significantly predicted the postoperative CCT in our model (P > .05).

Figure 5
Figure 5:
Central corneal thickness change over time.
Table 2
Table 2:
Central corneal thickness and flare values at different time points.


Table 2 shows the mean flare at different timepoints. The mean flare change from baseline to 1 hour postoperatively, from baseline to 1 day postoperatively, and from baseline to 1 month postoperatively was 6.6 photons/msec (mean increase 127% ± 6.6%; range −5.6 to 24.3 photons/msec), 3.9 photons/msec (mean increase 77.0% ± 6.0%; range −3.4 to 23.1 photons/msec), and −0.1 photons/msec (mean change 30% ± 9.8%; range −18.0 to 24.7 photons/msec), respectively (Figure 6). The mean flare value increased significantly (P < .01) from preoperatively to 1 hour postoperatively and decreased significantly from 1 hour to 1 day postoperatively (P = .03). None of the fragment score parameters significantly predicted the postoperative flare in our model (P > .05).

Figure 6
Figure 6:
Flare change (photons/msec) over time.


The aim of this prospective study was to observe the influence of swirling lens fragments hitting the corneal endothelium during phacoemulsification. Although postoperative complications have been extensively described in the literature, to our knowledge nothing has been reported on the influence of swirling lens fragments coming into contact with the corneal endothelium during phacoemulsification.7

In this study, the mean postoperative ECC showed no significant difference from the preoperative value. These findings are in agreement with those of Storr-Paulsen et al.8 and Assaf and Roshdy.9 In contrast, Sobottka Ventura et al.4 measured a significant endothelial cell loss 3 months postoperatively. It remains unclear why Sobottka Ventura et al. observed an endothelial cell loss of more than 350 cells/mm2, which is 7 times higher than the loss in our study. Unfortunately, they did not give information about the intensity of cataract or phaco time in their paper.

We included consecutive patients in this study without selecting cataract intensity. For this reason, numerous patients with predominantly cortical cataracts or little nuclear cataracts were enrolled. In these cases, the phacoemulsification process is relatively straightforward with little energy use and the swirling lens fragments are probably less harmful because they are softer. Our mean phaco time and phaco power were 1.7 seconds and 11.2%, respectively.

In another study comparing different phaco tips,10 a significant loss of endothelial cells was observed, although not as severe as the loss observed by Sobottka Ventura et al.4 Faramarzi et al.10 used different intraoperative settings (ie, infusion bottle height) and did not measure swirling lens fragments. The height of the infusion bottle was found to be a factor in the endothelial cell loss. Suzuki et al.11 showed that a low bottle height is less harmful to the corneal endothelium in porcine eyes.

One difference between the Sobottka Ventura et al.4 and Faramarzi et al.10 studies and our study was the postoperative measurement follow-up. The other 2 studies evaluated the ECC for up 3 months after surgery, in contrast to our study’s 1-month postoperative follow-up. However, Kohlhaas et al.12 found no significant changes in the ECC after the first postoperative month. However, ECC measurements are not perfectly reliable. In some patients, the postoperative ECC was higher than the preoperative one, most likely as a result of a measurement error.

In this study, we observed a significant increase in CCT 1 hour postoperatively and complete recovery of the CCT on the first postoperative day. Therefore, it is likely that the corneal edema is a direct reaction to phacoemulsification and not of inflammation, which would show the largest effect many hours or days after surgery. Chen et al.13 observed similar findings when comparing torsional and conventional phacoemulsification techniques. Unfortunately, in their study CCT measurements were performed 1 day and 3 months after surgery only; therefore, no conclusion concerning the recovery time can be drawn. Sobottka Ventura et al.4 reported a larger increase in the CCT on the first postoperative day; however, no measurements were performed 1 hour after surgery. Mathew et al.14 observed a slower recovery of the CCT in diabetic patients and in nondiabetic patients; however, measurements were performed 1 week, 6 weeks, and 3 months after surgery only and with US pachymetry. One main factor that has to be taken into account is the strong correlation between corneal postoperative swelling and phaco energy, with the swelling resulting from thermal injury. However, the swelling usually takes place in the vicinity of the incision and not in the corneal center.15

Highest flare values were found immediately after surgery and returned to baseline at the 1-month follow-up. These findings agree with those in a previous study using the same flare meter and the same surgical setup16 as well as in another study using an older version of the same flare meter17 in which the flare values were back to baseline on the 28th postoperative day. Abell et al.18 reported a similar decrease in flare within the first postoperative weeks; however, their 1-day postoperative values were higher than ours, possibly because of the use of a femtosecond laser. Recently, a high correlation between postoperative flare and cystoid macular edema was reported.19 These findings suggest it would be enlightening to further investigate the factors that reduce postoperative flare, such as using a cooled intraocular solution.20

The main outcome in this study concerning fragments was that small swirling fragments hitting the center of the cornea had the most significant influence on postoperative endothelial cell loss. A possible explanation for this finding could be that smaller swirling fragments move faster and hit the endothelium with higher velocity than large swirling fragments. Alternatively, these fragments can cause a pinpoint defect, which a larger fragment with a more diffuse contact zone might not.

Possibly, the effect would have been more pronounced had more eyes with dense cataracts been included. Because grading of the density of cataract was not performed preoperatively, we came to the conclusion that very few dense cataracts were included in this study based on the low phaco time and energy used. The effect of more dense cataracts might be interesting to examine in another study.

A validated fragment score could have several advantages. Mainly, for research purposes, it provides the possibility of allowing direct measurement of the influence of intraoperative parameters, such as infusion bottle height, phaco techniques, phaco devices, phaco energy levels, and other parameters, on the morphology and behavior of swirling lens fragments. In addition, a fragment score could also provide a valuable teaching tool for surgeons. The possibility to perform and evaluate intraoperative OCT measurements of swirling lens fragments allows adaptation to each surgeon’s phacoemulsification technique and fluidic settings to improve the outcome. This could provide a different approach to cataract surgery, such as avoiding small fragments and trying to operate at a larger distance to the cornea. Another use of the fragment score is to document intraoperatively special cases, such as patients with cornea guttata.

Although the findings in this study could help improve cataract surgery, some shortcomings should be addressed. The frame rate and resolution of the OCT prototype used were not ideal. Some swirling fragments were not perfectly detected because of the low frame rate of the time-domain OCT. An increased frame rate would be valuable to identify each swirling fragment hitting the cornea, and a better frame rate would provide a more accurate measurement of the duration of the impact. A possible solution to this problem would be to use Fourier-domain OCT instead of time-domain OCT. The greatest advantages of Fourier-domain OCT are the speed in which scans are acquired as well as their resolution. Eventually, fragments could be shown in 3-dimensional images. This would allow much better measurement of fragment position and size relative to the cornea and the condition of the cornea itself. A system that is already integrated into an OCT device could score every fragment automatically and objectively evaluate each fragment during surgery and at the moment of collision.

Another limitation was that because of its low resolution, the OCT system we used cannot measure microparticles colliding with the endothelium. With higher resolution OCT, further grading of the fragment size can be defined and correlated with endothelial cell damage.

Another interesting thing to explore would be the phaco tip’s position in relation to the corneal endothelium and its impact on endothelial cells. Because of the shadowing effect of the phaco tip in the OCT recording, it was not possible to determine the exact position in relation to the endothelium; therefore, we could make no objective assessment of the phaco tip.

Contrary findings on endothelial cell loss between studies suggest that there are major influencing factors that are not well understood. Overall, the question of why dense cataracts are associated with higher endothelial trauma is complex and multifactorial. Some factors might be the longer US time, increased irrigation fluid volume, prolonged overall procedure time, possibly the greater proximity of the phaco tip to the cornea, and the endothelial bombardment with macrofragments and microfragments. Some of those factors are not easy to assess in vivo with our current setup Therefore, further studies and more sophisticated techniques are required add to our findings.

The fragment score we developed could help us better understand the main parameters influencing postoperative endothelial cell loss. Our findings suggest that the mechanical trauma from the swirling lens fragments touching the center of the corneal endothelium might be more relevant than the contributions of phacoemulsification energy or the changes in the biochemical milieu during surgery.

What Was Known

  • Many factors can cause corneal damage.

What This Paper Adds

  • Small swirling fragments hitting the center of the cornea had the most significant influence on the postoperative endothelial cell loss.
  • Fourier-domain intraoperative OCT can be used by surgeons to assess their technique.


1. Bourne RRA, Minassian DC, Dart JKG, Rosen P, Kaushal S, Wingate N. Effect of cataract surgery on the corneal endothelium; modern phacoemulsification compared with extracapsular cataract surgery. Ophthalmology. 2004;111:679-685.
2. Linebarger EJ, Hardten DR, Shah GK, Lindstrom RL. Phacoemulsification and modern cataract surgery. Surv Ophthalmol. 1999;44:123-147.
3. Takahashi H. Free radical development in phacoemulsification cataract surgery. J Nippon Med Sch. 72, 2005, p. 4-12, Available at: Accessed May 16, 2016.
4. Sobottka Ventura AC, Wälti R, Böhnke M. Corneal thickness and endothelial density before and after cataract surgery. Br J Ophthalmol. 85, 2001, p. 18-20, Available at: Accessed May 16, 2016.
5. Hirnschall N, Norrby S, Weber M, Maedel S, Amir-Asgari S, Findl O. Using continuous intraoperative optical coherence tomography measurements of the aphakic eye for intraocular lens power calculation. Br J Ophthalmol. 2015;99:7-10.
6. Hirnschall N, Amir-Asgari S, Maedel S, Findl O. Predicting the postoperative intraocular lens position using continuous intraoperative optical coherence tomography measurements. Invest Ophthalmol Vis Sci. 54, 2013, p. 5196-5203, Available at: Accessed May 16, 2016.
7. Hengerer FH, Dick HB, Buchwald S, Hütz WW, Conrad-Hengerer I. Evaluation of corneal endothelial cell loss and corneal thickness after cataract removal with light-adjustable intraocular lens implantation: 12-month follow-up. J Cataract Refract Surg. 2011;37:2095-2100.
8. Storr-Paulsen A, Jφrgensen JS, Norregaard JC, Thulesen J. Corneal endothelial cell changes after cataract surgery in patients on systemic sympathetic α-1a antagonist medication (tamsulosin). Acta Ophthalmol. 92, 2014, p. 359-363, Available at: Accessed May 16, 2016.
9. Assaf A, Roshdy MM. Comparative analysis of corneal morphological changes after transversal and torsional phacoemulsification through 2.2 mm corneal incision. Clin Ophthalmol. 7, 2013, p. 55-61, Available at: Accessed May 16, 2016.
10. Faramarzi A, Javadi MA, Karimian F, Jafarinasab MR, Baradaran-Rafii A, Jafari F, Yaseri M. Corneal endothelial cell loss during phacoemulsification: bevel-up versus bevel-down phaco tip. J Cataract Refract Surg. 2011;37:1971-1976.
11. Suzuki H, Oki K, Shiwa T, Oharazawa H, Takahashi H. Effect of bottle height on the corneal endothelium during phacoemulsification. J Cataract Refract Surg. 2009;35:2014-2017.
12. Kohlhaas M, Stahlhut O, Tholuck J, Richard G. Entwicklung der Hornhautdicke und -endothelzelldichte nach Kataraktextraktion mittels Phakoemulsifikation. Changes in corneal thickness and endothelial cell density after cataract extraction using phacoemulsification, Ophthalmologe 1997;94:515-518.
13. Chen X, Ji Y, Lu Y. Comparison of clear corneal incision injuries between torsional and conventional phacoemulsification. Graefes Arch Clin Exp Ophthalmol. 2013;251:2147-2154.
14. Mathew PT, David S, Thomas N. Endothelial cell loss and central corneal thickness in patients with and without diabetes after manual small incision cataract surgery. Cornea. 2011;30:424-428.
15. Murano N, Ishizaki M, Sato S, Fukuda Y, Takahashi H. Corneal endothelial cell damage by free radicals associated with ultrasound oscillation. Arch Ophthalmol. 126, 2008, p. 816-821, Available at: Accessed May 16, 2016.
16. Maedel S, Hirnschall N, Chen Y-A, Findl O. Effect of heparin coating of a foldable intraocular lens on inflammation and capsular bag performance after cataract surgery. J Cataract Refract Surg. 2013;39:1810-1817.
17. Kruger A, Schauersberger J, Findl O, Petternel V, Svolba G, Amon M. Postoperative inflammation after clear corneal and sclerocorneal incisions. J Cataract Refract Surg. 1998;24:524-528.
18. Abell RG, Allen PL, Vote BJ. Anterior chamber flare after femtosecond laser–assisted cataract surgery. J Cataract Refract Surg. 2013;39:1321-1326.
19. Ersoy L, Caramoy A, Ristau T, Kirchhof B, Fauser S. Aqueous flare is increased in patients with clinically significant cystoid macular oedema after cataract surgery. Br J Ophthalmol. 2013;97:862-865.
20. Findl O, Amon M, Kruger A, Petternel V, Schauersberger J. Effect of cooled intraocular irrigating solution on the blood–aqueous barrier after cataract surgery. J Cataract Refract Surg. 1999;25:566-568.
© 2016 by Lippincott Williams & Wilkins, Inc.