928
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 38. NO. 9, SEPTEMBER 1991
[IO] H . H. Pennes, “Analysis of tissue and arterial blood temperatures in the resting human forearm,” J . Appl. Physiol., vol. 1, pp. 93-122,
1948. [ I l l J . P. Casey and R . Bansal, “The near field of an insulated dipole in a dissipative dielectric medium,” IEEE Trans. Microwave Theory Tech., vol. MTT-34, pp. 459-463, 1986. 1121 J. W. Strohbehn, B. S . Trembly, and E. B. Douple, “Blood Row effects on the temperature distributions from an invasive microwave antenna array used in cancer therapy,” IEEE Trans. Biomed. E n g . , vol. BME-29, pp. 649-661, 1982.
Laser Doppler Velocimetry Stabilized in One Dimension Michael T. Milbocker Abstract-Stabilization of the incident laser beam against eye movement improves the reliability of laser Doppler velocimetry in measuring retinal blood speed.
an image of reflected light from the tracking strip onto a detecting element (CCD), electronics for repeatedly and rapidly scanning the intensity profile of the image strip, and additional electronics for analyzing the scanned intensity profile [ 5 ] .The analyzing electronics provide iterative correction signals to stabilize the tracking laser beam and the LDV illuminating beam at a fixed position on the fundus. The laser Doppler portion of the instrument is configured as a bidirectional system. The procedure involves collecting the light scattered by the RBC’s in two narrowly defined cones separated by a relatively large known angle in the plane of the vessel. The solid angle of a collection cone is defined by the aperture of a coherent fiber optic bundle located at a fixed position conjugate to a virtual pupillary plane. The difference between the Afmax obtained from the two collection angles is used to calculate an absolute measure of V,,,. As a prelude to stabilized measurement of blood velocity in humans, this communication presents preliminary measurements of spectra obtained from polystyrene spheres flowing in model retinal vessels using a prototype one-dimensional stabilized LDV. TRACKING PERFORMANCE
INTRODUCTION Laser doppler velocimetry (LDV) has been applied to retinal blood vessels providing a noninvasive, direct, and absolute measurement of the centerline red blood cell velocity V,,,,, in individual arteries and veins [ 11. Vmaxis measured by obtaining a spectrum of Doppler-shifted laser light scattered from red blood cells (RBC’s). Light scattered from the vessel wall serves as a local oscillator in the optical mixing spectroscopy of the Doppler-shifted light from the RBC’s so that V,,, is measured relative to the vessel wall. The spectrum exhibits large fluctuations in power density up to a clearly measurable frequency shift Af,,, which is directly proportional to V,,,, the velocity of RBC’s moving down the center of the vessel. From V,,, and knowledge of the spatial distribution of velocities within the vessel the mean blood velocity is determined [2]. Routine use of LDV for measurement of retinal hemodynamics has proven difficult. One reason is the susceptibility of the measurement to eye movement. Valid data are often distributed among data obtained while the beam is decentered or completely off the vessel. One remedy is to develop complex rejection criteria in an effort to circumvent the problem [3], [4]. In this communication we discuss a more direct solution, the stabilization of the incident laser beam on the target retinal vessel. INSTRUMENT The principal object of an LDV stabilized in one dimension is to maintain the incident laser beam at the center of a vessel, allowing for normal, fixated eye movement along the vessel axis. The advantage inherent to the stabilized LDV is a continuous stream of laser Doppler-shifted light originating from RBC’s flowing down the vessel center. The stabilized LDV tracks movement of the vessel perpendicular to its axis and repositions the incident laser to the vessel’s center. The tracking means (see Fig. 1) include a laser source which projects a tracking strip of light on the fundus, optics for producing Manuscript received October 29, 1990. This work was supported in part by NE1 Grant EY01303. The author is with the Eve Research Institute, Boston, MA 02 114. IEEE Log Number 9101993.
There are two categories of factors affecting the accuracy and precision of the position of the incident laser Doppler beam relative to a fixed location at the fundus. The first concerns the performance of the eye tracker as an eye movement detection device. The second concerns the ability of the beam directing mechanics to respond accurately to a measured eye movement. The eye movement detector accuracy is determined by: 1) the contrast of the retinal feature (vessel) against the local fundus background as 543 nm, 2) the intensity of the 543 nm tracking beam, and 3) noise sources within the detector. The accuracy of the mechanical response is determined by: 1) the step size, 2) the time to the next position update, and 3) the magnitude of the eye movement. Experiments designed to determine the minimal detectable displacement of a model target vessel, as well as the accuracy of the automated repositioning of the tracking strip were conducted. A 100 pm diameter model vessel was illuminated with a 450 pm long strip of green (543) nm) HeNe laser light oriented perpendicular to its axis. The power density was 0.50 nW/cm2. The strip was positioned such that its center was coincident with the center of the model vessel to an accuracy of k 2 pm. The model eye was rotated parallel to the CCD detecting elements until the tracking electronics indicated that decentering had occurred. The mean minimal detectable displacement ws 4 . 6 2 pm. The automated redirection of the tracking strip resulted in a mean recentering error of 5.6 2 pm. The minimal detectable displacement and recentering error increased with decreasing power density of the tracking strip. For a power density of 22 pW/cm2, the mean recentering error was 7.2 f 2 pm. The prototype instrument has been successful in tracking retinal vessels of individuals over a wide range of fudus pigmentations. During tracking, a record of electronic signals sent to the galvanometer is obtained which accurately reproduces the magnitude and direction of eye movements. The accuracy of the system as a whole was tested on humans by requiring the eye to fixate on targets separated by a known angle. Three small LED’s were mounted in a shield spaced 2 ” apart. An eye fixated on each of the LEDs in a sequential fashion produced eye movement records corresponding to 2” jumps to an accuracy of f 1.5%, the ratio of the LED halfwidth to the spacing distance.
*
0018-9294/91/0900-0928$01.OO
*
0 1991 IEEE
~
1ll l1Il 1l1l l lll
lul1lllIl
~~
I I Il III I I I Il
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 38, NO. 9, SEPTEMBER 1991
c> c> Beam Expander
929
c> ICCD
Attenuator
1
Laser 543 nm
633 nm
Fig. 1. A schematic of a laser Doppler velocimeter stabilized in one dimension. There are three sources of illumination (1) a yellow HeNe laser beam which illuminates f 10" of the fundus, (2) a green HeNe laser fundus tracking beam, and (3) a red HeNe laser Doppler beam. The instrument is comprised of a beam-directing and image rotation system (4 and 5 ) , an intensified charge-coupled imaging device (6). a laser Doppler collection system (7), and viewing optics (8). The two deflectors in system 4 indicated by bold lines are mirrors attached to galvanometers mounted orthogonal to each other. These two galvanometer-driven mirrors provide manual (both axes) and electronically automated (one axis) positioning of the tracking and diagnostic beams at the fundus. Scattered light from the fundus is collected by an ophthalmic lens and passes back through system 5 on the reverse side of the galvanometer-controlled mirrors. Red light is collected at two small mirrors behind the image rotator and directed into the PMT assembly. Green light is deflected into the intensified CCD assembly. Yellow light passes through to the observer.
SPECTRUM C
SPECTRUM B
Y
3
-
4 0
AI (kHz)
10
Fig. 2. Power spectra were obtained from red (633 nm) HeNe laser light scattered from a dilute suspension of 1 pm diameter polystyrene spheres flowing at a known rate through a 200 pm inner diameter glass tube. The tube was located in the retinal plane of an Eyetech model eye. The model eye was mounted on a galvanometer such that the plane of motion was perpendicular to the orientation of the glass tube. Spectrum A is a long time integration (30 s) with the artificial eye stationary. Spectrum B is the same time integration while the artificial eye oscillates +lo0 p n at 100 Hz. Spectrum C is taken under the same conditions, but with the eye tracker on. The bold line indicates the frequency shift calculated from the known cenerline velocity. Spectrum B illustrates the accumulated effect of the sweeping beam as it samples components of the flow < V,,,,,. Spectrum C illustrates the ability of the tracking system to correct for eye movements.
LDV RESULTS Laser Doppler spectra were obtained from 1 pm diameter polystyrene spheres flowing through a 200 pm diameter glass tube located at the fundus plane of an Eyetech model eye. The model eye was mounted on a General Scanning galvanometer and made to oscillate in a plane perpendicular to the vessel direction. A laser spot positioned on the vessel when stationary would sweep across
the vessel at a fixed frequency when the galvanometer was powered. Several frequencies were tested. In all cases V,,, exhibited a time dependent behavior which corresponded to the frequency tested and the position of the incident laser beam in the velocity distribution of the Row. Fig. 2 shows the effect of eye movement on a Doppler shift spectrum and the ability of the system to correct for eye movements.
930
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING. VOL. 38. NO. 9, SEPTEMBER 1991
DISCUSSION
Normal fixated eye movement procedures decentering of the incident LDV beam. A valid measurement of V,,, corresponds to measuring the fastest component of flow, even slight decentering of the beam can bias the measurement of V,,, to smaller values. Stabilization of the beam on the vessel center provides a more reliable measure of Vmax. REFERENCES
G. T . Feke, D. G. Goger, H . Tagawa, and F. C . Delori, “Laser Doppler technique for absolute measurement of blood speed in retinal vessels,” IEEE Trans. Biomed. Eng., vol. 34, pp. 673-680, 1987.
[2] G. T. Feke, H. Tagawa, D. M. Deupree, D. G . Goger. J . Sebag, and J. J. Weiter, “Blood flow in the normal human retina.” Invest. Ophth. Vision Sci., vol. 30, no. I , pp. 58-65, 1989. [3] M. T. Milbocker, G . T. Feke, and D. G. Goger, “Automated determination of centerline blood speed in retinal vessels for laser Doppler spectra,” presented at Top. Meet. Noninvasive Assessment of the Visual System, 1988 Tech. Dig. Series, Vol. 3. Washington, DC: Opt. Soc. Amer., 1988, pp. 162-165. [4] B. L. Petrig and C. E. Riva, “Retinal laser Doppler velocimetry: Towards its computer-assisted clinical use.” Appl. Opt., vol. 27, no. 6. pp. 1126-1 134, 1988. [5] M.T. Milbocker, K . P. Pflibsen, G . T . Feke, F. Rogers, and D. G. Goger, “A modular eye fundus tracker and image stabilizer,” presented at Top. meet. Noninvasive Assessment of the Visual System,” 1989 Technical Digest Series, Vol. 7. Opt. Soc. Amer., 1989, pp. 50-53.
Call for Applications and Nominations for Editor IEEE TRANSACTIONS O N MEDICAL IMAGING Applications and nominations for the position of Editor, IEEE TRANSACTIONS ON MEDICAL IMAGING, should be sent to: Robert C. Waag Chairman, Steering Committee IEEE TRANSACTIONS ON MEDICAL IMAGING c/o Diagnostic Radiology, Box 648 University of Rochester Medical Center 601 Elmwood Avenue Rochester, NY 14642 Supporting material should include a statement of interest in the position, goals, and qualifications. The deadline for submissions is October 31, 1991 and a final decision is expected in December 1991 for the assumption of duties.in January 1992.
IlI l Il1l l I)I I 111
lul1lllIl
I I Il III I I I Il
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 38. NO. 9, SEPTEMBER 1991
[IO] H . H. Pennes, “Analysis of tissue and arterial blood temperatures in the resting human forearm,” J . Appl. Physiol., vol. 1, pp. 93-122,
1948. [ I l l J . P. Casey and R . Bansal, “The near field of an insulated dipole in a dissipative dielectric medium,” IEEE Trans. Microwave Theory Tech., vol. MTT-34, pp. 459-463, 1986. 1121 J. W. Strohbehn, B. S . Trembly, and E. B. Douple, “Blood Row effects on the temperature distributions from an invasive microwave antenna array used in cancer therapy,” IEEE Trans. Biomed. E n g . , vol. BME-29, pp. 649-661, 1982.
Laser Doppler Velocimetry Stabilized in One Dimension Michael T. Milbocker Abstract-Stabilization of the incident laser beam against eye movement improves the reliability of laser Doppler velocimetry in measuring retinal blood speed.
an image of reflected light from the tracking strip onto a detecting element (CCD), electronics for repeatedly and rapidly scanning the intensity profile of the image strip, and additional electronics for analyzing the scanned intensity profile [ 5 ] .The analyzing electronics provide iterative correction signals to stabilize the tracking laser beam and the LDV illuminating beam at a fixed position on the fundus. The laser Doppler portion of the instrument is configured as a bidirectional system. The procedure involves collecting the light scattered by the RBC’s in two narrowly defined cones separated by a relatively large known angle in the plane of the vessel. The solid angle of a collection cone is defined by the aperture of a coherent fiber optic bundle located at a fixed position conjugate to a virtual pupillary plane. The difference between the Afmax obtained from the two collection angles is used to calculate an absolute measure of V,,,. As a prelude to stabilized measurement of blood velocity in humans, this communication presents preliminary measurements of spectra obtained from polystyrene spheres flowing in model retinal vessels using a prototype one-dimensional stabilized LDV. TRACKING PERFORMANCE
INTRODUCTION Laser doppler velocimetry (LDV) has been applied to retinal blood vessels providing a noninvasive, direct, and absolute measurement of the centerline red blood cell velocity V,,,,, in individual arteries and veins [ 11. Vmaxis measured by obtaining a spectrum of Doppler-shifted laser light scattered from red blood cells (RBC’s). Light scattered from the vessel wall serves as a local oscillator in the optical mixing spectroscopy of the Doppler-shifted light from the RBC’s so that V,,, is measured relative to the vessel wall. The spectrum exhibits large fluctuations in power density up to a clearly measurable frequency shift Af,,, which is directly proportional to V,,,, the velocity of RBC’s moving down the center of the vessel. From V,,, and knowledge of the spatial distribution of velocities within the vessel the mean blood velocity is determined [2]. Routine use of LDV for measurement of retinal hemodynamics has proven difficult. One reason is the susceptibility of the measurement to eye movement. Valid data are often distributed among data obtained while the beam is decentered or completely off the vessel. One remedy is to develop complex rejection criteria in an effort to circumvent the problem [3], [4]. In this communication we discuss a more direct solution, the stabilization of the incident laser beam on the target retinal vessel. INSTRUMENT The principal object of an LDV stabilized in one dimension is to maintain the incident laser beam at the center of a vessel, allowing for normal, fixated eye movement along the vessel axis. The advantage inherent to the stabilized LDV is a continuous stream of laser Doppler-shifted light originating from RBC’s flowing down the vessel center. The stabilized LDV tracks movement of the vessel perpendicular to its axis and repositions the incident laser to the vessel’s center. The tracking means (see Fig. 1) include a laser source which projects a tracking strip of light on the fundus, optics for producing Manuscript received October 29, 1990. This work was supported in part by NE1 Grant EY01303. The author is with the Eve Research Institute, Boston, MA 02 114. IEEE Log Number 9101993.
There are two categories of factors affecting the accuracy and precision of the position of the incident laser Doppler beam relative to a fixed location at the fundus. The first concerns the performance of the eye tracker as an eye movement detection device. The second concerns the ability of the beam directing mechanics to respond accurately to a measured eye movement. The eye movement detector accuracy is determined by: 1) the contrast of the retinal feature (vessel) against the local fundus background as 543 nm, 2) the intensity of the 543 nm tracking beam, and 3) noise sources within the detector. The accuracy of the mechanical response is determined by: 1) the step size, 2) the time to the next position update, and 3) the magnitude of the eye movement. Experiments designed to determine the minimal detectable displacement of a model target vessel, as well as the accuracy of the automated repositioning of the tracking strip were conducted. A 100 pm diameter model vessel was illuminated with a 450 pm long strip of green (543) nm) HeNe laser light oriented perpendicular to its axis. The power density was 0.50 nW/cm2. The strip was positioned such that its center was coincident with the center of the model vessel to an accuracy of k 2 pm. The model eye was rotated parallel to the CCD detecting elements until the tracking electronics indicated that decentering had occurred. The mean minimal detectable displacement ws 4 . 6 2 pm. The automated redirection of the tracking strip resulted in a mean recentering error of 5.6 2 pm. The minimal detectable displacement and recentering error increased with decreasing power density of the tracking strip. For a power density of 22 pW/cm2, the mean recentering error was 7.2 f 2 pm. The prototype instrument has been successful in tracking retinal vessels of individuals over a wide range of fudus pigmentations. During tracking, a record of electronic signals sent to the galvanometer is obtained which accurately reproduces the magnitude and direction of eye movements. The accuracy of the system as a whole was tested on humans by requiring the eye to fixate on targets separated by a known angle. Three small LED’s were mounted in a shield spaced 2 ” apart. An eye fixated on each of the LEDs in a sequential fashion produced eye movement records corresponding to 2” jumps to an accuracy of f 1.5%, the ratio of the LED halfwidth to the spacing distance.
*
0018-9294/91/0900-0928$01.OO
*
0 1991 IEEE
~
1ll l1Il 1l1l l lll
lul1lllIl
~~
I I Il III I I I Il
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 38, NO. 9, SEPTEMBER 1991
c> c> Beam Expander
929
c> ICCD
Attenuator
1
Laser 543 nm
633 nm
Fig. 1. A schematic of a laser Doppler velocimeter stabilized in one dimension. There are three sources of illumination (1) a yellow HeNe laser beam which illuminates f 10" of the fundus, (2) a green HeNe laser fundus tracking beam, and (3) a red HeNe laser Doppler beam. The instrument is comprised of a beam-directing and image rotation system (4 and 5 ) , an intensified charge-coupled imaging device (6). a laser Doppler collection system (7), and viewing optics (8). The two deflectors in system 4 indicated by bold lines are mirrors attached to galvanometers mounted orthogonal to each other. These two galvanometer-driven mirrors provide manual (both axes) and electronically automated (one axis) positioning of the tracking and diagnostic beams at the fundus. Scattered light from the fundus is collected by an ophthalmic lens and passes back through system 5 on the reverse side of the galvanometer-controlled mirrors. Red light is collected at two small mirrors behind the image rotator and directed into the PMT assembly. Green light is deflected into the intensified CCD assembly. Yellow light passes through to the observer.
SPECTRUM C
SPECTRUM B
Y
3
-
4 0
AI (kHz)
10
Fig. 2. Power spectra were obtained from red (633 nm) HeNe laser light scattered from a dilute suspension of 1 pm diameter polystyrene spheres flowing at a known rate through a 200 pm inner diameter glass tube. The tube was located in the retinal plane of an Eyetech model eye. The model eye was mounted on a galvanometer such that the plane of motion was perpendicular to the orientation of the glass tube. Spectrum A is a long time integration (30 s) with the artificial eye stationary. Spectrum B is the same time integration while the artificial eye oscillates +lo0 p n at 100 Hz. Spectrum C is taken under the same conditions, but with the eye tracker on. The bold line indicates the frequency shift calculated from the known cenerline velocity. Spectrum B illustrates the accumulated effect of the sweeping beam as it samples components of the flow < V,,,,,. Spectrum C illustrates the ability of the tracking system to correct for eye movements.
LDV RESULTS Laser Doppler spectra were obtained from 1 pm diameter polystyrene spheres flowing through a 200 pm diameter glass tube located at the fundus plane of an Eyetech model eye. The model eye was mounted on a General Scanning galvanometer and made to oscillate in a plane perpendicular to the vessel direction. A laser spot positioned on the vessel when stationary would sweep across
the vessel at a fixed frequency when the galvanometer was powered. Several frequencies were tested. In all cases V,,, exhibited a time dependent behavior which corresponded to the frequency tested and the position of the incident laser beam in the velocity distribution of the Row. Fig. 2 shows the effect of eye movement on a Doppler shift spectrum and the ability of the system to correct for eye movements.
930
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING. VOL. 38. NO. 9, SEPTEMBER 1991
DISCUSSION
Normal fixated eye movement procedures decentering of the incident LDV beam. A valid measurement of V,,, corresponds to measuring the fastest component of flow, even slight decentering of the beam can bias the measurement of V,,, to smaller values. Stabilization of the beam on the vessel center provides a more reliable measure of Vmax. REFERENCES
G. T . Feke, D. G. Goger, H . Tagawa, and F. C . Delori, “Laser Doppler technique for absolute measurement of blood speed in retinal vessels,” IEEE Trans. Biomed. Eng., vol. 34, pp. 673-680, 1987.
[2] G. T. Feke, H. Tagawa, D. M. Deupree, D. G . Goger. J . Sebag, and J. J. Weiter, “Blood flow in the normal human retina.” Invest. Ophth. Vision Sci., vol. 30, no. I , pp. 58-65, 1989. [3] M. T. Milbocker, G . T. Feke, and D. G. Goger, “Automated determination of centerline blood speed in retinal vessels for laser Doppler spectra,” presented at Top. Meet. Noninvasive Assessment of the Visual System, 1988 Tech. Dig. Series, Vol. 3. Washington, DC: Opt. Soc. Amer., 1988, pp. 162-165. [4] B. L. Petrig and C. E. Riva, “Retinal laser Doppler velocimetry: Towards its computer-assisted clinical use.” Appl. Opt., vol. 27, no. 6. pp. 1126-1 134, 1988. [5] M.T. Milbocker, K . P. Pflibsen, G . T . Feke, F. Rogers, and D. G. Goger, “A modular eye fundus tracker and image stabilizer,” presented at Top. meet. Noninvasive Assessment of the Visual System,” 1989 Technical Digest Series, Vol. 7. Opt. Soc. Amer., 1989, pp. 50-53.
Call for Applications and Nominations for Editor IEEE TRANSACTIONS O N MEDICAL IMAGING Applications and nominations for the position of Editor, IEEE TRANSACTIONS ON MEDICAL IMAGING, should be sent to: Robert C. Waag Chairman, Steering Committee IEEE TRANSACTIONS ON MEDICAL IMAGING c/o Diagnostic Radiology, Box 648 University of Rochester Medical Center 601 Elmwood Avenue Rochester, NY 14642 Supporting material should include a statement of interest in the position, goals, and qualifications. The deadline for submissions is October 31, 1991 and a final decision is expected in December 1991 for the assumption of duties.in January 1992.
IlI l Il1l l I)I I 111
lul1lllIl
I I Il III I I I Il
