# Automated Smartphone based System for Diagnosis of Diabetic Retinopathy

**Authors:** Misgina Tsighe Hagos, Shri Kant, Surayya Ado Bala  
**arXiv:** [2004.03408](https://arxiv.org/abs/2004.03408)  
**Date:** 2020-04-07  
**Summary page:** https://www.summarizepaper.com/en/arxiv-id/2004.03408v1/

## Key points

- Authors propose and implement an inception-based convolutional neural network and a binary decision tree-based ensemble of classifiers for the detection and classification of diabetic retinopathy
- Challenges in diagnosing diabetic retinopathy in rural areas include shortage of trained ophthalmologists, limited availability of healthcare centers, and expensive diagnostic equipment
- Deep learning-based automatic diagnosis techniques have been implemented but fail to provide point-of-care diagnosis
- Authors develop a smartphone application that incorporates their proposed method, providing an offline and automatic system for diagnosing diabetic retinopathy
- Proposed solution is to deploy automated diagnoses on smartphones to provide instant diagnosis to diabetic individuals residing in remote areas
- Increasing usage of smartphones worldwide presents an opportunity to reach individuals in remote areas who lack access to specialized healthcare services
- By leveraging smartphone technology, the proposed system aims to bridge the gap in early diagnosis and treatment of diabetic retinopathy
- This approach has the potential to improve healthcare outcomes for diabetic individuals living in rural regions by enabling non-experts to conduct accurate diagnoses using their smartphones

## Summary

In this paper titled "Automated Smartphone based System for Diagnosis of Diabetic Retinopathy," authors Misgina Tsighe Hagos, Shri Kant, and Surayya Ado Bala propose and implement an inception-based convolutional neural network and a binary decision tree-based ensemble of classifiers for the detection and classification of diabetic retinopathy. The existing summary discusses the challenges in diagnosing diabetic retinopathy in rural areas due to a shortage of trained ophthalmologists, limited availability of healthcare centers, and expensive diagnostic equipment. It also mentions that although deep learning-based automatic diagnosis techniques have been implemented, they still fail to provide point-of-care diagnosis. The authors further develop a smartphone application that incorporates their proposed method, providing an offline and automatic system for diagnosing diabetic retinopathy. The proposed solution is to deploy automated diagnoses of diabetic retinopathy on smartphones to provide instant diagnosis to diabetic individuals residing in remote areas. The study highlights the increasing usage of smartphones worldwide as an opportunity to reach individuals in remote areas who lack access to specialized healthcare services. By leveraging smartphone technology, the proposed system aims to bridge the gap in early diagnosis and treatment of diabetic retinopathy. This approach has the potential to improve healthcare outcomes for diabetic individuals living in rural regions by enabling non-experts to conduct accurate diagnoses using their smartphones. Overall, this research presents a promising solution for addressing the challenges associated with diagnosing diabetic retinopathy in rural areas. By utilizing smartphones as a platform for automated diagnoses, it offers a convenient and accessible tool that can significantly impact the management and treatment of this disease among underserved populations.