Do you remember something called COVID-19? It was this coronavirus that emerged at the end of 2019 and we’re pretty sure it didn’t have that much of an impact on our lives… well, perhaps, just slightly. Okay, fine, it was one of the most transformative experiences of our lives and we will all likely live in its shadow for the rest of our lives. Media MICE was no less affected in a number of ways, from disruption to our tea spread across the world, to the themes of the content we produced during that time.
One of the topics that dominated the medical headlines during much of the COVID-19 pandemic was telemedicine and how clinics across the healthcare spectrum, from cardiology and oncology to our own ophthalmology, were using it to continue treatments during stringent social distancing. Then the discussion moved from using the technology as a stop-gap measure to alleviate patient backlogs, toward becoming an accepted standardized option in ocular treatment.
One of the conditions that has benefited the most from the increased options that telemedicine can offer is cataract, one of the most common conditions ophthalmologists face on a day-to-day basis. Cataract accounts for 51% of all eye diseases in the U.S. and as a progressive disease, early diagnosis, intervention and treatment are absolutely crucial in achieving the best possible patient outcomes.1 Thus, the first line of defense is imaging technology, the means by which the first cataract diagnosis can be made — so, what if we could find a way to make screening more accessible than ever before by using smartphones?
Howdy Partner, That’s a Mighty Fine Smartphone You’ve Got There
That was the question a group of researchers based in the lone star state of Texas wanted to examine in their paper Detecting Cataract Using Smartphones. The Texans resolved to apply an efficient approach to identify cataract disease by adopting luminance features using a smartphone. The problem they said they had encountered in other studies was that results could vary from a wide variety of factors including camera quality and processing power (among others), and as a result, they needed to identify a methodology using luminance technology as it is not color-based, and thus dependent on camera sensor characteristics and environmental conditions.1
To justify their position our Texas rangers (of science) pointed to a study comparing standard hue, saturation, value (HSV) — also known as red-blue-green (RBG) — with luminance technology. The study concluded that the luminance-based method had 86.67 % accuracy, while the HSV color-based method had only 33.4% accuracy in detecting cancer cells.2 Based on these findings, they moved on to outlining their cataract methodology.
In the Texan tests, subjects would sit in a stable position and align their eye with a smartphone’s rear camera, which was located 10-50 cm from the eye with autofocus. After the images were captured, the smartphone processed them and presented the results. All in all, 100 eye model images were captured, 50 from healthy eye models, and 50 from diseased eye models.1
Let’s Round Up Them Steers (I Mean Patients)
The researchers found that changing the camera angle, distance and smartphone had 2.2%, 3.3%, and 3% impact on luminance values and 9.2%, 13.3%, and 8.5% impact on RGB values, respectively. However, changing the ambient light had a 36% difference impact on the luminance values, which was similar to the 32% difference impact it had on the RGB values.1 So we can therefore see (baddum tish) that the luminance technique is arguably more reliable, but how about accurate?
Our top Texan team reported that of the 100 eyes that were screened as part of their study they were able to achieve accurate results in 96.6% of cases, while also achieving 93.75% sensitivity and 93.4% specificity. While this is important, what also needs to be noted is that changing environmental factors had very limited impact (at an average of 2.8%) on the outcome results.
This means that the luminance technique could be applied effectively in almost any setting, significantly improving the accessibility of screening. No doubt that will help improve access to treatment in the lone star state, which is absolutely huge by the way, and also around the world as smartphone screening can get almost anywhere. Kudos!
- Askarian B, Ho P, Woon Chong J. Detecting Cataract Using Smartphones. IEEE J Transl Eng Health Med. 2021; 9: 3800110.
- Vaghela H, Modi H, Pandya M, Potdar MB. A Comparative Study of the HSV Color Model and YCbCr Color Model to Detect the Nucleus of White Cells. Int. J. Comput. 2016; 150(8):38-42.