Outline

  • Abstract
  • Keywords
  • I. Introduction
  • II. Method
  • A. Image Acquisition
  • B. Optic Disk Localization
  • C. Macular Region Detection
  • D. Drusen Detection
  • III. Experiments and Results
  • IV. Conclusion and Recommendations
  • References

رئوس مطالب

  • چکیده
  • کلید واژه ها
  • 1. مقدمه
  • 2. روش
  • الف. گرفتن تصویر
  • ب. مکان یابی دیسک بینایی
  • ج. آشکارسازی منطقه لکه دار
  • د. آشکار سازی نقاط روشن
  • 3. آزمایشات و نتایج
  • 4. نتیجه گیری و پیشنهادات

Abstract

The human eye is a vital organ of vision which can be affected by many diseases. One of the most common diseases that have affected people of over 50 years of age is Age Related Macular Degeneration. Patients in their fifties or more or who have undergone surgery for cataract and glaucoma should have their eyes examined annually. Inspection provides a large number of Fundus images and medical professionals has to continually peruse them spending valuable time and energy. Current methods of detection and assessment are manual, expensive and potentially inconsistent. Thus, it would be more cost effective and beneficial if the initial task of analysing retinal photographs is automated. The proposed solution would act as an early warning system, where an ophthalmologist will be able to analyse numerous images within a brief period and spend more time on those patients who are actually in desideratum of their expertise.

Keywords: - - -

Conclusions

A method for ARMD diagnosis is discussed in this paper. In the approach first the optic disk is detected using threshold and using thee geographical features macular area is detected and the drusens are found using the canny edge detector. The reasons for choosing these technologies are also discussed in the paper. Some existing methods were tried on a data set and accurate method was found for segmentation which is the most important feature.

Then using another data set the implemented features were experimented and 92% accuracy was obtained. The main features and important requirements identified were developed. But there are some limitations in this product which are listed down below,

· Image captured with the old generation fundus camera cannot be supported by the system due to the lack in quality.

· Only colour retinal fundus images can be supported by the system.

· One of the major drawbacks in this proposed system is that this system is not connected with the database and the data is maintained in files and folder system which is not considered to be and efficient way of handling the data.

· Though it was said that the performance was up to the standard what was expected , it has not reached up to 100% precision in segmentation, due to the differentiation of drusens and different images

· Another drawback in this system is that only one image can be uploaded at a time for analysis purposes.

The following are some recommendation where the product can be improved and more accuracy can be obtained in future.

Since this proposed system is a Desktop Applications it is expected to press forward this system by constructing it a as web application.

Adding artificial intelligence is also recommendation as it could increase the accuracy level.

Here only fundus images are considered. Processing of auto-fluorescence image along with fundus image and also incorporating OCT report will give more accurate results.

دانلود ترجمه تخصصی این مقاله دانلود رایگان فایل pdf انگلیسی