Detection of Glucoma and Diabetes through Image Processing and Machine Learning Approaches

In the last few decades, glaucoma became the second biggest leading cause of irreversible vision loss. Because of its asymptotic growth, it is not properly diagnosed until the relatively late stage. To stop the severe damage by glaucoma it is needed to detect glaucoma in its early stages. Surprisingly diabetes also be the greatest cause of glaucoma. In the modern era, artificial intelligence makes great progress in the medical image processing field. Image analysis based on machine learning gives a huge success in diagnosis glaucoma without any misdiagnosis. The aim of this proposed paper is to create an automated process that can detect glaucoma and diabetic retinopathy. Here various Machine Learning models are used and results of these methods are presented.


INTRODUCTION
Glaucoma is a persistent condition that gradually damages the optic nerve of the eyes. It is the second largest cause of blindness in the world. It is a neurodegeneration disease and the ultimate conclusion is lost vision. When the intraocular pressure increases in the eye area, it starts to damage the optic nerves of the eyes gradually. Finally, when this damage is too high then the optic nerves stop sending visual images to the brain. And it causes total vision loss. To stop the severe damage, it is needed to detected glaucoma in the early stages. Glaucoma was diagnosed by four examinations: (1) measuring elevated intraocular pressure (IOP), (2) percentage damage to the optic disc by calculating the cup-to-disc ratio (CDR), (3) identifying decreased retinal nerve fiber layer (RNFL) thickness and (4) detection of characteristic visual field defects. The clinical test like optical coherence tomography (OCT) and visual field (VF) test display the indicators to diagnose glaucoma. But these conventional methods have higher chances of being misdiagnosed. In the modern era, artificial intelligence makes a great impact in the medical image processing field.
Machine learning and image analysis give the greatest opportunity to detect glaucoma without being misdiagnosed. The main objective of this experiment is to create an automated glaucoma diagnosis system. So that the system can predict glaucoma at any stage very accurately and efficiently.

RELATED WORKS
To avoid the diagnosis the researchers, use artificial intelligence to detect glaucoma. They develop different machine learning models to create an automated glaucoma detection procedure. In today's world, machine learning became the leading tool for predicting and treating this severe disease.
Researchers develop a machine learning predictive model that can select the five data features of the patients 1) visual field test,2) a retinal nerve fiber layer optical coherence tomography (RNFL   Figure 2 shows the normal vs glaucoma vs diabetic glaucoma fundus images [16].

PROPOSED METHODOLOGICAL WORK
The proposed methodology follows the three main steps. Figure 3 shows the main three steps of the proposed methodology. The input image database contains 15 image sets of healthy patients, diabetic retinopathy patients, and glaucoma patients [16]. All the input images are in RGB color space.

b. Image Pre-Processing
The pre-processing step is divided into four parts. Figure 4 shows the steps involve in the pre-processing. For image classification, different pre-trained convolution neural network models ResNet50 [17], VGG16 [18], VGG19 [19], InceptionV3 [20], and Xception [21]will be used. In order to get the best result to detect glaucoma, it is needed to fine-tune the required layers of the pre-trained models and also needed to adjust the epochs. The proposed algorithm of this experiment is given below.

d. Algorithm
Step1: Load the input RGB images.
Step2: Convert RGB to Gray Scale.
Step3: Remove the noise of the image using Discrete Wavelet transformation techniques.
Step4: Convert the noise-free gray-scale images into RGB images.
Step5: Resize the images according to the network model's requirements.
Step6: Apply the pre-trained models to train the new data.
Step7: Classify the test data and predict the outcome. The proposed method is further analysis through Machine Learning models and their results are shown for indicating the betterment of the Machine Learning approaches.
Pima Indian women dataset from uci machine learning repository [22] is utilized for predicting diabetic tendency of a patient. The dataset can be formulated as collection of attributes that include several criterions for detecting diabetic tendency such as number of pregnancies, patient's age, blood pressure, insulin taken, BMI percentage, Glucose level, Diabetes Pedigree function, Skin Thickness, Outcome (Diabetic/Non-diabetic). However, the attribute 'outcome' is utilized as a target class of the prediction. The dataset is partitioned into training set and testing dataset. 80% of the dataset is allocated for the training set whereas testing dataset need 20% of the dataset. Table 1 and

CONCLUSIONS
Glaucoma is a neuropathic disease. It degenerates the ganglion cells of the eyes. The erosion of the eyecup enlargement damages the optic nerve and finally causes vision loss. It is one of the greatest challenges to detect glaucoma before severe stages. In this experiment, a method is proposed to detect glaucoma automatically. The proposed method helps to detect various glaucoma automatically and precisely. The method is used here ML and Stacking Ensemble Method shows the best results. https://associatedretinaconsultants.com/diagram-of-the-eye/