IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 41. NO. 2, FEBRUARY 1994 20 I Preliminary Results on Reflectance Feedback Control of Photocoagulation In Vivo Maya R. Jerath, Ravi Chundru, Steven F. Barrett, H. Grady Rylander 111 and Ashley J. Welch Abstract-The size of therapeutic laser-induced retinal lesions is crit- ical for effective treatment and minimal complications. Due to tissue variability, the size of a lesion that results from a given set of laser irradiation parameters cannnt be predicted. Real time feedback control of lesion size is implemented based on two-dimensional reflectance images acquired during irradiation. Preliminary results of feedback controlled lesions formed in pigmented rabbits demonstrate an ability to produce uniform lesions despite variations in tissue ahsorption or changes in laser power. I. INTRODUCTION Argon laser-induced retinal lesions find many therapeutic applica- tions in ophthalmology [I, 2, 31. Due to tissue variability, lesions that result from a given set of laser irradiation parameters are non- uniform. It has been shown that the local absorption in the pigment epithelium of light at the argon laser wavelengths varies by a factor of four from fundus to fundus and by a factor of two within the same fundus [4]. Thus, tissue optical properties at any given location are not known. The laser power and spot size at the fundus are also unknown quantities. They are preset at the cornea, but are affected by the ocular optics. Due to these inhomogeneities, the extent of photocoagulation cannot be predicted a priori. Work on improving the accuracy of photocoagulation has been ongoing for many years. Birngruber et al. suggested that measurement of fundus reHectance during photocoagulation could be used to control laser photocoag- ulation [6]. This work was further extended when Weinberg et al.. using a single photodetector to measure reflectance during photoco- agulation, attempted to correlate total reflectance to the volume of tissue coagulated [7]. Unfortunately a high correlation could not be obtained. Yang et al. [8] and Jerath et al. 191 have shown that two- dimensional reflectance images provide parameters that can be used to successfully control lesion sire in an eye phantom. This paper presents results obtained in i,iivi with a two-dimensional reflectance-based feedback system that ~(introla lesion size in real time by adjusting the exposure he. It must be noted that in these experiments, laser parameters were chosen to slow down the lesion formation process to provide the 'real time' system sufficient temporal resolution. As a reault, the parameters chosen are not similar to those used clinically. However proof of the concept is shown. and with the appropriate fast hardware, the use of clinical laser parameters will be possible. Central reflectance from two-dimensional images acquired during irradiation is used to control lesion depth in pigmented rabbits. Measurement of reflectance is non-invasive and monitoring lesion growth in real time allows the feedback control system to compensate for any kind of variability-both in the tiwue and the irradiation Manuscript received October 19, 1992: revised October 6. lYY3. Thi\ work supporled in part by the Texas Coordinating Board and in part by the Oflice of Naval Research under grant N(n)014-91-J-1564. The authors are with the Biomedical Engineering Program. ENS 63Y. The University of Texas at Auatin, Austin, TX 78712. IEEE Log Number 9214364. o"sol . . . om12 0.0 0 I 0 2 0.3 0.4 0.5 coagulated layer ~cb (cm) Fig. 1. Reflectance increases as the scattering layer thickness increases, linally levelling off to an R,. parameters. The laser is simply shut off when lesions of the 'right' size are obtained. It has been shown previously that central reflectance correlates strongly with the lesion depth [9]. The essential concept is as follows. In the one-dimensional case, as the thickness of a scattering layer (coagulated tissue) increases, reflectance increases monotonically until the scattering layer becomes infinitely thick optically. At this point the reflectance levels off to an R,. See Fig. 1. Thus, it is possible to control the coagulated layer thickness based on reflectance as long as reflectance is still increasing. I!. METHODS A. Animal Preparation Lesions were formed in pigmented, cross-bred (Californian and New Zealand) 3 kg rabbits. The rabbits were anesthetized intra- muscularly with a mixture of Ketamine (35 mgkg) and Rompan (5.9 mg/kg). The eyelids were held open with a speculum. A suture inserted in the medial rectus muscle allowed the eyeball to be rotated. The pupil was dilated with a 1% Tropicamide solution. The cornea was kept moist with a 0.9'4 saline drip. B. E.vprrimen tal Appuratus The experimental system for delivery of argon light to a target and measurement of reflectance images is shown in Fig. 2. A 5 watt rated Coherent Argon laser, functioning at all argon lines (primary lines are 488 nm and 514.5 nm), is fed to a tiber and serves as the coagulating light murce. A CCD camera (30 frames/s), coaxially aligned with the laser beam, collects the reflectance signal from the forming lesion. An Olympus fundus camera provides the illumination source for the reHectance images. In order to prevent interference between the reflectance signal and the reflected laser light a slotted spinning wheel alternately shutters thr laser and the camera paths. Thus the camera never 'sees' the laser beam. The irradiation and CCD interrogation duty cycles are each 33%. To ensure uniform illumination intensity in the acquired frames, the spinning wheel rotates at a harmonic of the frame rate. Frames are acquired by an IT1 15 I Image Processor controlled by a Sun 3/260 workstation. The laser can be opened and closed by the computer within 7 ms with a ITL signal. C. Image Prowssing and Control .%$?ware The image processor grabs the reflectance images of the lesion recorded by the CCD camera (512 x 512 pixels, 30 frameds) and 00i8-Y2Y4/94$04.00 0 1YY4 IEEE202 IEEE TRANSACTIONS ON RIOMEDICAL ENGINEERING, VOL. 41. NO. 2. FEBRUARY 1994 512rs12 u- . -! Fig. 2. size.