Multimodal Imaging-based and OCT-guided Navigated Retinal Laser Photocoagulation

Results from a prospective, interventional study

IGOR KOZAK, MD, PhD • SHARIF EL-EMAM, MD • JAY CHHABLANI, MD

Retinal laser photocoagulation of the posterior pole has undergone some revolutionary changes, such as semiautomated delivery, micropulse and selective application, and retinal navigation, with the integration of imaging and laser delivery.^1-3 These new concepts have increased the speed, safety, and accuracy of laser photocoagulation procedures.

Our diagnostic abilities are also continuously improving, thanks to advances in retinal imaging. These developments have included increases in image resolution, as well as the development of various new imaging modalities that were not part of clinical assessment few years ago.

A very popular approach to the diagnostic workup is multimodal imaging, in which each imaging study contributes to bettering the overall characterization of the disease process. If we can enhance diagnostics using multimodal imaging approaches, then perhaps therapy could also benefit.

NAVIGATED LASER PHOTOCOAGULATION

The Navilas navigated laser photocoagulator (OD-OS, Inc., Irvine, CA) integrates retinal imaging, fluorescein angiography, and retinal laser. Photocoagulation is executed based on previously acquired images from the same system and treatment plan.

The physician prepares the treatment plan, based on either angiography or fundus photos, and then overlays the plan on live retina so laser photocoagulation can follow ocular movements during treatment without contact lenses.^4-6

The Navilas system not only collects pretreatment diagnostic information, but it also eliminates the need to remember precisely where all laser applications should be applied or have been applied. The treatment plan is executed completely, which is not always the case when the plan and treatment are not integrated.

Navilas represents a photocoagulation system that performs image-guided treatments based on its own retinal images. Navigation systems depend on image registration to relate images acquired from different sources or time points.

In this study, we examined the feasibility of integrating retinal images from outside imaging sources for the purpose of navigated laser photocoagulation.

METHODS

Patients

This was an interventional case series of eyes with retinovascular disease. The study was approved by the Institutional Review Board of the University of California at San Diego and was performed at the UCSD Jacobs Retina Center. The spectrum of diagnoses included DME (n=12) (Figures 1 and 2), AMD (n=2) (Figures 3 and 4), and radiation retinopathy (n=1) (Figures 5 and 6, page 58).

Figure 1. A Navilas treatment plan (left) based on fluorescein angiogram of an eye with diabetic retinopathy and macular edema showing numerous microaneurysms and both focal and diffuse dye leakage. The large yellow circle represents the “no treatment zone,” and the small yellow circles are target laser spots. The retinal thickness map (right) from an OCT scan imported into the Navilas system, overlaid on the FA image, and aligned on the foveal center of the same eye. False color (red) shows retinal thickening.

Figure 2. A Navilas treatment plan (left) based on fluorescein angiogram of an eye with diabetic retinopathy that had been treated previously and that shows residual dye leakage. The large yellow circle represents the “no treatment zone,” and the small yellow circles are target laser spots. The retinal thickness map (right) from an OCT scan imported into the Navilas system, overlaid on the FA image, rotated, and aligned on the foveal center of the same eye. False color (red) shows retinal thickening.

Figure 3. Color fundus photo (left) of an eye with AMD that was unresponsive to intravitreal pharmacotherapy, showing retinal exudation and scarring. A Navilas treatment plan (right) for grid laser maculopexy, based on retinal thickness map from an OCT scan of the same eye imported into the Navilas system, overlaid on the FA image, and aligned on the foveal center of the same eye. False color (red) shows retinal thickening.

Figure 4. Fluorescein angiogram (left) of the same eye in Figure 3, showing exudation and scarring. A Navilas treatment plan (right) for grid laser maculopexy based on the FA image.

Figure 5. Color fundus photo (left) of an eye with circinate retinopathy that underwent previous laser photocoagulation following radiation for choroidal melanoma (black). Early-phase fluorescein angiography image (right) showing retinal scars from previous laser photocoagulation.

Figure 6. Late-phase fluorescein angiography image of the same eye in Figure 5, showing late leakage next to retinal scars from previous laser photocoagulation. The large yellow circle represents the “no treatment zone,” and the small yellow circles are target laser spots. A Navilas treatment plan (right) for grid laser, based on a retinal thickness map from an OCT scan of the same eye imported into the Navilas system, overlaid on the FA image. False color (red) shows retinal thickening, although overt leakage is absent on FA.

Image Integration Technology

All consecutive eyes in this study underwent preoperative fluorescein angiography and SD-OCT imaging (Heidelberg Engineering, Vista, CA). We downloaded these images in uncompressed format, onto an external memory source.

Once imported into the navigated laser system, a registration tool accessed the image to be aligned with a reference image, which was acquired by the navigated laser system. The semiautomatic registration used landmarks identified by the operator in both images to calculate a multidimensional transformation matrix for registration.

Correct registration could be achieved with only three corresponding landmarks, depending on the source images. Any additional landmark would have provided the software with more information to recalculate and would have potentially improved the image transformation to compensate for image distortions. Five landmarks spread over the entire image area typically result in quite an accurate overlay. Nine registration points is the theoretical optimum.

A transformation and warping algorithm in the application scaled and rotated the resultant layered image, providing both visual and internal confidence scales to determine adequate alignment.

We used the resultant image to plan treatments that specifically corresponded to the images acquired by the laser delivery system or to serve as a second layer to complement diagnostic information before planning treatment.

RESULTS

In our study, a total of 15 eyes from 15 patients (ages 44 to 78 years) underwent SD-OCT retinal thickness map imaging. We selected both FA and OCT images that were aligned and integrated.

We imported these images into the navigated laser system, where baseline color fundus photos and FA images were also obtained. All of the images were of good quality. The imported OCT images were successfully overlaid and registered onto the selected FA images, allowing for treatment plans to be executed.

On average, six landmarks were necessary to overlay the image successfully. We could adjust the transparency of the overlaid OCT image to allow us to see the underlying FA template.

We were able to create treatment plans, and all patients received standard photocoagulation treatment, which occurred without any complications.

The learning curve for OCT-guided treatment is typically steep. Because the software automatically aligned the selected images, this step was straightforward. As reported previously, orientation with a camera-based system, as opposed to a slit-lamp system, is crucial for the execution of treatment. Image integration did not delay overall treatment planning.

We were able to execute both planning and treatment as rapidly as previously reported, irrespective of the number of laser applications and even with less cooperative patients.⁷

DISCUSSION

Integrating Our Information

Imaging and photocoagulation have traditionally been considered two separate processes, performed with two or more different instruments. Integrated navigated retina laser technology brings both diagnostics and treatment closer than ever before, essentially merging them.

In this study, we evaluated a new feature of the navigated laser photocoagulation system, which allows for the importation of images from various diagnostic imaging devices, including OCT.

Currently, ophthalmologists review imaging studies before procedures — most frequently FA, because it provides two-dimension spatial information on the distribution of retinal lesions and dye leakage (retinal function).

The traditional classification of DME into focal and diffuse types mandates the use of FA in deciding which areas require treatment. The physician can then supplement the treatment plan with information from the z-axis from OCT regarding the amount of retinal thickening, coupled with SLO imaging, which precisely localizes the areas of thickening (retinal structure).

The information from color fundus photography can only be used rarely. Virtual treatment plans based on fundus photos must be executed by correctly remembering all of the steps and retinal landmarks. They can easily be lost during ancillary procedures, such as placing a contact lens on the ocular surface, focusing on the fundus, or titrating the power of laser burn. Incorrect judgment of laser target location can result in suboptimal results or unnecessary trauma to adjacent anatomical structures.

Putting the Information to Work

Navigated posterior-pole photocoagulation is a process that requires a great deal of accuracy in image registration. Image registration is the process of transforming images acquired at different time points or with different imaging modalities into the same coordinate system. It is an essential part of any surgical planning and navigation system because it facilitates the combining of images with important complementary structural and functional information, based on which critical decisions are made.⁸

As a result, integrating a second level of a diagnostic template, such as OCT, requires perfect alignment of the retinal structures to annotate correctly the spots on the retina that must be treated.

We achieved this integration by selecting an average of six corresponding landmarks between FA and SLO templates from OCT retinal thickness maps. Eventually, we could transform three-dimensional information into two-dimensional platforms.

The most significant advantage of integrating OCT images into treatment plans is the ability to identify areas of retinal thickening that are not visible on FA and that otherwise would not be treated if only FA were used as a treatment template.

Some diffuse leakage is visible on FA, but it is not necessarily associated with retinal thickening or cyst formation on OCT because it can be intracellular.^9,10 OCT adds important structural and clinical information to the treatment plan, with which we can construct fully integrated plans that will maximize the benefits of retinal photocoagulation.

Multimodal imaging has the advantage of complementary views of the retinal architecture, and their integration can significantly enhance treatment potential in retinal photocoagulation.

CONCLUSION

External image integration is now a feasible feature of navigated photocoagulation. Laser treatments can now be custom-planned, guided, and performed using multiple imaging modalities, including OCT.

These developments could provide an alternative to FA guidance or serve as an adjunct to FA for macular laser photocoagulation treatment for retinal vascular disorders. RP

REFERENCES