FPGAs Make Retinal Disease Treatment Faster and Safer

Medical and electronics technologies join forces to treat retinal disease with 100% reliability.

Reliability is an important aspect for every product on the market, but no industry values it more than the manufacturers of medical devices. When reliability directly affects the health and safety of human beings, manufacturers must spend extra time and resources evaluating every possible concern and adding redundancies for a 100% success rate.

More than 50% of Americans diagnosed with diabetes are at risk of developing diabetic retinopathy, a retinal disease that commonly leads to blindness. Using laser burns to remove abnormal blood vessels from the retina, ophthalmologists can free up oxygen to the rest of the eye and slow down or even prevent the blindness caused by this retinal disease.

Traditionally, physicians administer laser burns one shot at a time using a joystick and foot pedal to aim and fire. With up to 2,000 burns per treatment, a typical procedure would consist of at least three 15-minute sessions. By semiautomating a series of patterned laser shots, PASCAL can complete the procedure in about 5 minutes.

PASCAL fires patterns of 56 burns in less than 600 ms. Before every laser fire, the safety circuitry verifies the exact position of the laser, monitors the power of the beam, checks the current state of the physician�s foot pedal, and communicates with the doctor�s user interface. To accomplish the precision and speed needed for this application, OptiMedica turned to FPGA technology for testing and monitoring all safety circuitry in parallel.

Upon receiving the coordinates, the galvanometer sends back two voltages, providing feedback for the actual dual-axis position of an output beam that allows PASCAL to test and verify the actual response. If the mirrors are not at the correct angle, the laser is not permitted to fire. Both the response time and the position accuracy of the galvanometer can be characterized and monitored over time to detect any possible degradation of the system.

As with all medical devices, PASCAL incorporates other sensors to add redundancy for an increased level of safety. In addition to feedback built into the galvanometer, the laser beam is split and directed to an internal photo detector, which is constantly read by another analog input channel on the DAQ board.

This photo detector allows PASCAL to constantly test and monitor the laser power coming from the output beam. If the power is too low or too high, a second galvanometer deflects the beam away from the final lens and prevents faulty laser beams from entering into the patient�s eye.

PASCAL also uses this internal photo detector to ensure that the power is off when necessary, such as during the start-up sequence or when the system is in standby mode. When a physician is using a weaker aiming beam to centralize the pattern on a certain part of the retina, the internal photo detector verifies that the actual surgical beam is inactive.

Although the laser control logic and safety logic are embedded within the same FPGA on the DAQ board, dedicated blocks of I/O and silicon operate independently. These dedicated blocks of I/O allow PASCAL to fully take advantage of the parallelism that FPGAs offer.

Engineers at OptiMedica chose to use LabVIEW FPGA for all software development of the host and FPGA applications. Each part of the LabVIEW FPGA application can function simultaneously with all other pieces of the application, allowing redundant logic to become more than just another set of commands that procedurally execute. The FPGA enables PASCAL to continuously test and monitor itself and immediately identify possible component failures before undesirable results have occurred.

Another level of safety is enabled by the physician performing the procedure as he uses the foot pedal. With PASCAL, a physician uses a weak aiming laser that displays the entire pattern without any harm to the patient�s eye. He can position the aiming laser to damaged parts of the retina with a mechanical joystick and use the foot pedal to administer the pattern.

When performing this type of surgery, there�s no such thing as being too close to your mark. Through a rigorous calibration procedure, PASCAL is able to hit a spot on a patient�s retina within a few microns.

A slightly modified LabVIEW FPGA application is loaded into hardware and reconfigured to provide feedback on laser output accuracy. Laser position and laser power are the two main things that are tested and calibrated.

Position accuracy is determined mostly by looking at the difference between the output voltage to the X-Y galvanometer and the actual resulting angle of the mirrors. Any kind of deviation from the expected position values is corrected using offset and scaling constants, and alignment values are filed away and continuously referenced until the next scheduled maintenance.

Laser output power plays an important role in the effectiveness of this medical device. Once the X-Y positions have been realigned, the output power of the laser must be verified over the full range of possibilities and adjusted if necessary. This crucial process is completely automated by self-generated test signals and an external power meter.

The output power of the laser goes through a self-diagnostic procedure with various values at different spot sizes. The external power meter measures the results, and a curve of power values is plotted over various intensities.

While FPGA hardware is the heart of the control and safety system, the host PC is responsible for storing and transferring laser patterns based on selections from the user interface (Figure 1). Another level of safety is achieved by constantly monitoring a pulse train from the FPGA digital I/O line, which provides a digital heartbeat from the FPGA application. This ensures that the host PC knows when the FPGA is active and running before sending any commands or laser patterns. If the digital pulse train ever pauses or changes frequency, the host immediately knows that an unexpected event has occurred and prevents the physician from sending any more laser patterns until the problem has been resolved.

Conclusion
Imagine visiting an ophthalmologist and being diagnosed with diabetic retinopathy. The physician advises that unless a retinal surgical operation is performed, there�s a possibility of going blind. The operation would have to be done in at least three installments over the next two weeks and involve up to 2,000 laser burns administered one shot at a time.

Now think about being diagnosed in the doctor�s office and having the option to complete the whole procedure immediately. In five minutes, the whole process would be done.

In addition to being faster, initial results show that patients experience less discomfort than the traditional method due to shorter laser bursts at higher power. The PASCAL method of laser photocoagulation has taken a proven procedure and made it faster and easier for both patients and physicians.

About the Author
Vineet Aggarwal is the marketing engineer for S Series and R Series data acquisition products at National Instruments. He began his career at NI as an intern in 2002 and then joined the company full-time through the Engineering Leadership Program. Mr. Aggarwal also spent three months on a branch assignment in Tokyo. He holds a B.S. degree in electrical engineering from Ohio State University. National Instruments, 11500 North Mopac Expwy., Austin, TX 78759, 512-683-0100, e-mail: [email protected]

December 2006