Image processing for early detection of Alzheimer’s disease using Iridology

Alzheimer’s disease is caused by genetics, personal lifestyle and other environmental factors. It is an irreversible disease that slowly destroys the brain memory cells. There are no speci ic methods for the detection of Alzheimer’s disease. The primary symptoms of Alzheimer’s disease are memory loss, dificulty in thinking, a problem in writing and speaking and others. Iridology is alternative research that has gained more popularity in recent years, which studies the alterations of the iris in correspondence with the organs of the human body. The combination of digital image processing with Iridology gives an excellent opportunity to explore and learn about different neuronal diseases, speci ically Alzheimer’s disease. In this work, MATLAB software is applied to determine the colour, pattern and other factors that show the existence of Alzheimer’s disease. The noise in the iris image is removed by the Gaussian ilter, followed by histogram analyses and cropping. The Hough circle transform is used to identify the region of interest and to convert the circular iris image into rectangle form. In the training methods, the SVM and CNN classi iers are used to classify whether the person has Alzheimer’s disease. Finally, the results are compared with the real-time images.


INTRODUCTION
Iridology is an elective research that has increased greater prevalence in recent years. It studies the alterations of the iris in correspondence with the organs of the human body (Pergad and More, 2015). Alzheimer's disease can't be identi ied in the beginning period until the individual show's symptoms like memory loss, problem in writing and learning, dif iculty in thinking and other symptoms (Thakare and Pawar, 2016). This appears when the disease has grown. So, it is hard to identify the disease in the beginning time. The combination of digital image processing with Iridology gives a great opportunity to study and learn about different neuronal diseases, speci ically Alzheimer's disease (Escudero et al., 2013). In this paper, we will be discussing the condition of the brain region through the iris. The iris picture of the patient is utilized as an input for detecting the condition of the brain region. This study allows us to analyze and observe several characteristics like color, texture, pupil range of the iris image. Using this data with MATLAB we are going to conclude whether the patient has Alzheimer's. This whole process is performed as graphical user interface in MATLAB for easy operation. The sample steps used in graphical user interface are shown in Figure 7. The region of interest (brain region) is taken from the iridology chart as shown in Figure 1.

Methodology
Alzheimer's can be detected in an early stage by processing iris images in MATLAB. This procedure incorporates different steps. The iris image taken from the patient is used as an input image in MAT-LAB (Trokielewicz et al., 2015). Image preprocessing, image segmentation and normalization is performed on the iris image from this ROI (Region of interest) is obtained. Feature extraction is performed on the ROI and classi ication is done using an SVM (Support Vector Machine) and CNN (Convolutional Neural Network) classi ier. The output of the classi ication is mean, entropy and standard deviation of the pixel values in the iris image. By comparing these values with a healthy patient, we can detect Alzheimer's disease.

Input Image
The image that is used as the input system comes from the eye clinic database or real-time images of the patient. The input image is the primary procedure to recognize the brain region on the iris image (Ramlee and Ranjit, 2009). The real-time images should be photographed within highly compelled, capturing surroundings so that the iris is visible (Trokielewicz et al., 2014). The iris image can be captured with the help of the iris scanner or a high de inition camera (Sarika and Madhuri, 2016). The iris scanner is a movable opening (gap), which controls the measure of light getting through the lens

Preprocessing of Image
In preprocessing of an image, the image is subjected to two changes that are the noise in the image is removed and image enhancement is performed. The noise is removed by using the Gaussian ilter and the image enhancement is performed to improve the visibility or highlight any portion or feature of the image. These processes improve the details of the image. Histogram analyses are also performed on the iris image. This process improves the complexity of an image to acquire a uniform histogram intensity levels overall the image. The graphical user interface of image preprocessing is shown in Figure 3.

Segmentation of Image
Segmentation is performed to eliminate the unwanted surrounding around the iris region. So that the image is clearer and easier to understand and analyze. There are two types of segmentation required 1. Iris segmentation 2. Pupil segmentation. The output of the segmentation is only the circular part of the eye that is the iris and the pupil (Kashani et al., 2015). The iris and pupil segmentation of sample image is shown in Figures 4 and 5.

Normalization of Image
Normalization converts the circular shape iris image into a rectangular shape (Xing et al., 2015). This is performed so that various sizes of irises of every patient can be changed over into standard size and shape (Shree and Sheshadri, 2014).

Extraction of ROI
The iris image is converted into a standard rectangle shape to identify the region of interest. The region of interest is the brain region. The region of interest for Alzheimer's is shown in Figure 6.

Extraction of Feature
The features and characteristics of the iris image are different for the normal iris image and abnormal iris image. So, feature extraction is performed to extract the features like mean and entropy. The graphical user interface of feature extraction is show in Figure 8. The mean and entropy values of Alzheimer's and Non-Alzheimer's iris images are given in Table 1.

Classi ication
The classi ication is the last step to detect Alzheimer's disease (Umesh et al., 2016). There are different types of classi iers available (Hernández et al., 2018). All these classi iers are appropriate for classi ication. We need to choose a suitable classi ier among them based upon our requirement.

CONCLUSIONS
This paper provides a clinical feature analysis for the detection of Alzheimer's disease. In this project, the abnormal image is stored in the database along with the normal iris image. The image is iltered using the Gaussian ilter, the iris pattern from eye image is normalized and increased that iris pattern and extracted some characteristics like mean, variance, entropy. By comparing these values with the normal iris image values, we can conclude the presence of Alzheimer's disease.

ACKNOWLEDGEMENT
I would like to express my deepest gratitude to my guide, Asuntha A. Her valuable guidance consistent encouragement, personal caring, timely help and providing me with an excellent atmosphere for doing research. All through the work, in spite of her busy schedule, she has extended cheerful and cordial support to me for completing this research work.

Funding Support
There is no funding associated with this study. Data is collected from online database.

Con lict of Interest
The authors declare that there is no con lict of interests regarding the publication of this paper.