Scribd
Upload
53 views25 pages

RM Final Presentation

This document discusses using image processing and machine learning techniques for diabetic retinopathy detection. It begins with an introduction to diabetic retinopathy, the research question of using medical image analysis instead of manual detection, and the dataset used. The methodology section describes the pre-processing, augmentation, and machine learning models used - PNN, Bayes Classifier, and SVM. Expected results show SVM achieving the highest accuracy of 95.38%. Future work is identified in detecting proliferative diabetic retinopathy using medical image analysis and deep learning.

Uploaded by

ahsan puri
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
53 views25 pages

RM Final Presentation

This document discusses using image processing and machine learning techniques for diabetic retinopathy detection. It begins with an introduction to diabetic retinopathy, the research question of using medical image analysis instead of manual detection, and the dataset used. The methodology section describes the pre-processing, augmentation, and machine learning models used - PNN, Bayes Classifier, and SVM. Expected results show SVM achieving the highest accuracy of 95.38%. Future work is identified in detecting proliferative diabetic retinopathy using medical image analysis and deep learning.

Uploaded by

ahsan puri
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 25

Diabetic Retinopathy Detection

Using Image Processing and Machine Learning

Muhammad Nouman Ashraf [MSCSF21M530]

Muhammad Ahsan Tahir puri [MSCSF21M526]


Detail Introduction Of (DR)
• What is Diabetic Retinopathy (DR):
• Cause’s Of (DR):
• Normal Fundus Images :
• Comparison of effected and non effected samples
• How Medical Image Analysis Reduces Burden in DR detection
Research Question
• Can we have a better efficiency using Medical image Processing and Machine
Learning techniques instead of using manual processes.
Dataset
Normal Images
• 250 Images
Diabetic effected eyes images
• 100 Images
Literature Review
• Diagnosis of diabetic retinopathy using machine learning techniques
• Detecting clinical features of diabetic retinopathy using image processing
• Diagnosis of diabetic retinopathy using deep neural networks
• Identification and classification of microaneurysms for early detection of diabetic
retinopathy
• A deep learning ensemble approach for diabetic retinopathy detection
•Pre-Processing
METHODOLOGY
• Size Normalization
• Shape Normalization
• Color Normalization
•Augmentation
• Flip the image horizontally
• Flip the image vertically
• Randomly rotate the image in the range of [−25,25] degrees
• Randomly zoom in or out in the range of [0.85,1.15]
• Randomly distort the image

•Machine Learning Models


• Probabilistic Neural network (PNN)
• Bayes Classifier
• Support Vector Machine (SVM)
Diabetic Retinopathy
•Diabetic retinopathy is an eye disease caused by
diabetes.
Normal Fundus Image
Macula

The disk has sharp margins and is

Vess
normal in color, with a small central cup.

el
• Blood Vessels

s
• Optic Disc
• Macula

Optic Disc
Diabetic Retinopathy
•When high blood sugar levels cause damage to blood
vessels in the retina.
• These blood vessels can swell and leak.
• Sometimes abnormal new blood vessels grow on the
retina
Types of DR
Diabetic Retinopathy (DR)

Non-Proliferative Diabetic Proliferative Diabetic RetinoPathy


RetinoPathy (NPDR) (PDR)

Moderate
Mild NPDR Severe NPDR
NPDR
NPDR Types
Mild NPDR Moderate NPDR Severe NPDR
How MIA is Effective in DR?

• Manually a comprehensive dilated eye exam


• MIA Reduces Time in detection
• MIA saves from human error and improve accuracy
• MIA saves patient by chemical effects
Technical discussion
• Size Normalisation
• Shape Normalization
• Optic disc detection
• Blood vessels segmentation
by ( Morphology and contours)
• Macula segmentation by
provided ground truth
Technical discussion
Technical discussion

• Morphological processing
• CLAHE
• Canny edge detection
• Thresholding
• Contours/blobs
Technical discussion
Technical discussion
Technical discussion
Technical discussion
Technical discussion
Expected results

Machine Learning Models


• PNN has an accuracy of 87.69%
• Bayes Classifier has an accuracy of 90.76%
• SVM has an accuracy of 95.38%.
Future Work
• PDR Detection using (MIA and DL) Needed attention
References
1. American Academy of Opthamology
2. Colombia Opthomology
3. Opthomology Expert services
4. Deep Learning based Detection Model
5. Feature Extraction Code
6. Dataset1.zip

You might also like