By: Sara Hoover
Dr. Khan Iftekharuddin may not be restoring vision to the blind yet, but the associate
professor of electrical and computer engineering in the Herff College of Engineering
and his collaborator, Dr. Pinakin Gunvant, are doing the next best thing. They’re
catching glaucoma more accurately in patients than ever before thanks to an old technique
applied in a new way.
For the past seven years, Gunvant, assistant professor at the Southern College of
Optometry, has focused on improving diagnostic techniques used in detecting glaucoma.
Early detection is a key to treating glaucoma, according to Iftekharuddin.
“There was a need to look at the entire region of the back of the eye rather than
only an isolated region around the back of the eye as done by most analyses. That
required a different approach,” Gunvant said.
He began researching engineering schools in the Memphis area to find a collaborator
and came across Iftekharuddin’s Web page and realized his fractal techniques might
be the perfect fit. The collaboration began in Fall 2008.
“Why don’t we apply fractal analysis and see if it works?” said Gunvant. “We just
got lucky that we got a quick breakthrough automatically rather than requiring the
technique to be adjusted for multiple things.”
Fractal analysis identifies a pattern in a particular image or signal. It has been
shown to have a higher accuracy rate of identifying glaucoma indicators than standard
Iftekharuddin serves as the “engineering brain” behind the research. Gunvant brings
the medical training and an acute understanding of glaucoma, a disease in which the
optical nerve becomes damaged, leading to progressive and irreversible loss of vision.
In Gunvant’s glaucoma testing, light is beamed through the back of the retina to measure
the thickness of retinal tissue at certain points, which can indicate glaucoma. He
collects the data sets from patients and sends the images to Iftekharuddin’s lab for
“Our job is to analyze these images, write the computer and mathematical algorithms
to analyze and interpret the results and summarize it,” said Iftekharuddin, whose
other fractal analysis work includes finding infrared targets for the Army, face recognition
development for the National Science Foundation and brain tumor research with St.
Jude Children’s Research Hospital.
“It’s higher than standardized machine classifiers,” said Gunvant of Iftekharuddin’s
fractal analysis. “We can’t claim that this necessarily beats out all the other techniques
available. However, when compared to the standard machine classifier, fractal analysis
quite significantly beats the classifier in the range of 13 percent.”
Adds Iftekharuddin, “Commonly available medical devices, such as scanning laser polarimeters,
already come with built-in algorithms. Those algorithms give accuracy around 85 percent.
Using our fractal technique, we are able to bring it up to around 97 to 98 percent.
The accuracy has improved using our technique.”
The medical application of fractal analysis is novel.
Dr. Khan Iftekharuddin, associate professor of electrical and computer engineering
in the Herff College of Engineering, is researching ways to enhance testing
of glaucoma in an effort to better manage the disease.
“This is breakthrough. Fractal analysis has never been utilized for glaucoma management,”
said Gunvant. “The technique was established before. However, in glaucoma — to the
best of our knowledge — this has been the first attempt of the application of this
Iftekharuddin adds, “The goals are to see if we can do anything better than what’s
already out there. If we’re successful, we should be able to predict glaucoma earlier
on. Like any other disease, if you detect it earlier, the chances are the prognosis
is better, treatment is better.”
Gunvant says the disease affects about 3 percent of the world’s population.
“Glaucoma has been described as a silent thief of vision,” he said. “By the time a
person is identified with having the disease, they’ve already lost about 30 to 50
percent of their retinal cells. This is irreversible vision loss. So any early identification
technique will be a great asset. Our research is going to be quite beneficial.”
The team would like to not only identify who does and does not have glaucoma, but
the disease’s progression.
“You want to find those individuals who have the disease first,” said Gunvant. “And
as a subsequent bigger goal, we want to identify the individuals who are having progressive
damage. Because this is such a slow, progressive disease, it takes about seven to
10 years before you can identify the progression. Any technique that can identify
patients who are undergoing progression of the disease is very useful.
“We could then act upon trying to save the vision of those individuals. That’s the
next phase to the application of this procedure.”
The next step for the research also includes moving beyond two-dimensional data into
“In some respects, we are yet to make our biggest breakthrough in this new technique,”
said Gunvant. “The most important aspect of this analysis would be to apply this to
three-dimensional data of the back of the eye. For the next two years, we are going
to apply a similar technique to the retinal nerve fiber layer at the back of the eye
in trying to better predict glaucoma.”
The pair applied for a patent in September, along with a third investigator, U of
M doctoral student Paul Kim.
“This is really when (students) learn the tricks of the trade, how to do research
and also how to guide other people,” said Iftekharuddin. “I’m mentoring him at the
technical part, whereas my colleague is looking at the clinical part, so (Kim) gets
With a patent pending, the team has presented at conferences and has applied for funding.
Gunvant received a grant from the Assisi Foundation of Memphis, which partially covers
Kim’s tuition and stipend for the next two years. They hope to eventually apply for
grants from the NIH’s National Eye Institute.
Since the technique works for identifying glaucoma so well, there is a good chance
it may also work for other degenerative retinal diseases.
View a video of the research here.