June 2020
Volume 61, Issue 7
Free
ARVO Annual Meeting Abstract  |   June 2020
Detecting Cataract Using Smartphone.
Author Affiliations & Notes
  • behnam Askarian
    Electrical and Computer Engineering, Texas Tech University, Lubbock, Texas, United States
  • Darrin Peterson
    Lubbock Eye Clinic, Lubbock, Texas, United States
  • Peter Ho
    Lubbock Eye Clinic, Lubbock, Texas, United States
  • Jowoon Chong
    Electrical and Computer Engineering, Texas Tech University, Lubbock, Texas, United States
  • Footnotes
    Commercial Relationships   behnam Askarian, None; Darrin Peterson, None; Peter Ho, None; Jowoon Chong, None
  • Footnotes
    Support  NSF I-CORPS
Investigative Ophthalmology & Visual Science June 2020, Vol.61, 474. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      behnam Askarian, Darrin Peterson, Peter Ho, Jowoon Chong; Detecting Cataract Using Smartphone.. Invest. Ophthalmol. Vis. Sci. 2020;61(7):474.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

Purpose : A novel image processing and machine learning technique combined with an inclusive and affordable screening tool to detect cataracts using smartphones is proposed. The proposed method provides accurate detection of over 98.2% accuracy for cataract detection. The proposed method provides an affordable, easy-to-use and versatile method (application) that could be used in remote areas with medical shortage for detecting cataracts.

Methods : We recruited 50 subjects and recorded their eye data using our proposed smartphone-based cataract detection tool. Here, ophthalmologists’ diagnosis of subjects’ eyes is used as a gold-standard. Following the Texas Tech University Institutional Review Board (IRB#: IRB2018-964), we analyzed the de-identified eye image. The images were cropped to extract lens from the background. The proposed method uses a novel image processing method for measuring luminance reflection and color features of the lens combined with a Convolutional Neural Network (CNN) as a classifier to detect cataracts.

Results : Luminance reflection and color features of the lens has been extracted. The image from a diagnosed cataract with all features is shown in Fig.2. Fig.2c shows the luminance components of a healthy eye and Fig.2d shows the components of a cataract. The mean value of the colors for the healthy participant and cataract one is derived. From all the images that we fed into our system, our method could diagnose diseased eye from the healthy eye by a 98.2% accuracy, 97.8% specificity, and 97.2% sensitivity.

Conclusions : In this study, we investigated the feasibility of detecting cataracts using smartphones. Our study proposes an accurate, portable way to detect cataracts. We implemented our method using iPhone X camera pictures from 50 participants of which 25 had cataracts and 25 were healthy. Images were acquired from a participant’s eye using the smartphone, a special gadget, and flashlight with autofocus and maximum resolution. To improve the performance of our proposed method, we designed and manufactured a specially made gadget to control lighting conditions and avoid ambient light and reflection.

This is a 2020 ARVO Annual Meeting abstract.

 

Figure 1. Add on a device developed using 3D printing attached to an iPhone X (Left), Image acquisition setup (right).

Figure 1. Add on a device developed using 3D printing attached to an iPhone X (Left), Image acquisition setup (right).

 

Figure 2. (a) Healthy eye, (b) Eye with cataract, (c) result of cataract detection on healthy eye, and (d) result of cataract detection on the diseased eye.

Figure 2. (a) Healthy eye, (b) Eye with cataract, (c) result of cataract detection on healthy eye, and (d) result of cataract detection on the diseased eye.

×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×