A JOURNAL OF COMPOSITION THEORY ISSN : 0731-6755
Hardware & software tools used in the detection of glaucoma eye diseases in human beings
1Fazlulla Khan, 2Dr. Ashok Kusagur
1Research Scholar (Part-Time), USN: 4UB15PEJ04, VTU Research Centre, Govt. BDT College of Engg., EEE Dept., Davanagere, Karnataka & Asst. Prof., ECE Dept., HMS Inst. of Tech., Tumakuru, Karnataka
2Associate Professor & Head, Electrical & Electronics Engg. (EEE) Dept., Govt. BDT College of Engg., Davanagere, Karnataka
Abstract – This paper gives a brief review about the easily. But, in the glaucomatic eye, the trabecular meshwork various types of hardware & software tools used in the shown in the figure is blocked and hence the flow of the fluid design, development & implementation of algorithms for the is blocked, the pressure build up is shown and hence there is automatic detection of glaucoma in human beings. The a change in the optic nerve size, it bursts, the connectivity different types of hardware tools used are FPGA, HDL Kit between the eye and the brain is lost, which is leading to the & the software tools that are used are ModelSim, Xilinx, loss of the vision in the human eye [11]-[15]. Matlab, Simulink, Python, etc…. In this paper, a small review is given w.r.t. the said tools, which being used in the Here, in this work, the detection of glaucoma is planned to glaucoma detection, topic of research taken by us as a part be done using image processing algorithms on eye fundus of the Ph.D. programme. images, like watershed algorithms in an FPGA, the HOG algorithm using SVM, the genetic algorithms & the Keywords – Glaucoma, Eye, Disease, Normal, Affected, convolution neural network using the ANNs. The image processing techniques have already been starting to be used Blocks, Pressure, Tool, Hardware, Implementation, FPGA, in detection of various ailments such as tumors etc. So, Xilinx, Modelsim, Matlab, Simulation. detection of this eye diseases can empower ophthalmologist
to be more efficient in their work [16]-[20]. 1. Introduction
One of the danger of the glaucoma is that it is gradual and Glaucoma causes permanent blindness due to the damage of causes no pain and symptoms to the patients in its early optic nerves. It is an incurable disease so the diagnosis of the stage. Glaucoma causes loss of vison in the peripheral and disease is very imperative. The data shows that it is one of spreads do the center if it is not treated. Data from the WHO the leading cause for the blindness in the world, and also the (World Health Organization) shows is the cause of 15% of second most eye disease in huge numbers across the total blindness cases reported making it the second leading universe. So, the diagnosis of this disease in the early stage, cause for blindness. Projections show that, in the next and patient monitoring for long time and arriving with an decade, it can potentially affect ~80 million people [21]- applicable remedy in an appropriate time window by an [25]. ophthalmologist is a imperative [1]-[5].
When circulation of liquid called aqueous humor, in the front part of the eyes is not proper it results in a high fluid pressure inside the eyes. This is the cause for glaucoma. Tis is illustrated in the Fig. 1. The aqueous humor in normal eyes, discharges out of the eye the channel resembling mesh. When this channel for whatever reason gets blocked, the liquid density increases, and thus causes in the high pressure. This causes Glaucoma. In the Fig. 1 shown, there is a lot of difference which is shown between the anatomy of the Fig. 1 : Anatomy of normal and glaucoma eye normal eye & the glaucomatic eye, i.e., the eye which is
affected with the glaucoma disease [6]-[10]. How the glaucoma occurs in the early stages is shown in the Fig. 2.
In the normal eye, the trabecular meshwork is shown which is responsible for the flow of the aqueous fluid, there is a channel shown, which is free and hence the fluid can flow
Volume XII Issue X OCTOBER 2019 Page No: 600 A JOURNAL OF COMPOSITION THEORY ISSN : 0731-6755
Fig. 2 : Vision of normal and glaucoma eyes consists of modules and the language allows Behavioral, Dataflow and Structural Description. VHDL (Very High 2. Types of Glaucoma Speed Integrated Circuit Hardware Description Language) is standardized by IEEE1164. The design is composed of There are a large number of types of glaucoma which occurs entities consisting of multiple architectures. SystemC is a in the human eyes both at the primary level, secondary level language that consist a set of C++classes and macros. It & at the tertiary level, which is termed as the normal, allows electronic system level and transaction modeling. moderate & the severe level in context of severity of the disease. Here, in this section, different types of glaucoma Need for HDLs are discussed as below. These are marked by an increase of intraocular pressure (IOP) or pressure inside the eye [1]- The Moore’s Law in the year 1970 has brought a drastic [25]. change in the field of IC technology. This change has made Open-Angle Glaucoma the developers to bring out complex digital and electronic Angle-Closure circuits. But the problem was the absence of a better Normal Tension Glaucoma programming language allowing hardware and software Congenital Glaucoma codesign. Complex digital circuit designs require more time Primary Glaucoma for development, synthesis, simulation and debugging. The Secondary Glaucoma arrival of HDLs has helped to solve this problem by allowing each module to be worked by a separate team. Neo-vascular glaucoma
Neo-vascular glaucoma All the goals like power, throughput, latency (delay), test Exfoliate Glaucoma coverage, functionality and area consumption required for a Pigmentary Glaucoma design can be known by using HDL. As a result, the designer Chronic Glaucoma can make the necessary engineering tradeoffs and can Traumatic Glaucoma develop the design in a better and efficient way. Simple In the research work that we are going to take up further is syntax, expressions, statements, concurrent and sequential going to deal with the primary glaucoma detection along programming is also necessary while describing the with the hardware implementation of the same so that both electronics circuits. All these features can be obtained by at the simulation level & at the implementation level, it using a hardware description language. Now would have been validated. while comparing HDL and C languages, the major 3 Hardware & software tools used for research work difference is that HDL provides the timing information of a
The various h/w & s/w tools that are used in the course of design. the research work are explained as follows one after the other HDL Structure & Design in a nut-shell. Generally, HDL structure consist a textual description Hardware description language (HDL) is a specialized involving many inputs, outputs, signals operators, computer language used to program electronic and digital components, multiple architectures, and comments. logic circuits. The structure, operation and design of the Concurrent and sequential way of programming style is circuits are programmable using HDL. HDL includes a possible in HDL. Each and every HDL uses a different textual description consisting of operators, expressions, structure and design method. The examples shown below statements, inputs and outputs. Instead of generating a using VHDL and Verilog will help to get an overall idea computer executable file, the HDL compilers provide a gate about HDL structure and design. map. The gate map obtained is then downloaded to the programming device to check the operations of the desired HDL Simulation & Debugging circuit. The language helps to describe any digital circuit in the form of structural, behavioral and gate level and it is HDL Simulation involves the need of a test bench. A test found to be an excellent programming language for FPGAs bench is an environment used to check the correctness of the and CPLDs. design. The test bench involves the inputs, outputs and the procedure to be done. The HDL simulator will execute the The three common HDLs are Verilog, VHDL, and test bench; thereby the designer can verify the function of SystemC. Of these, SystemC is the newest. The HDLs will the circuit on the waveform analyzer provided by the HDL allow fast design and better verification. In most of the simulator. Many simulators are there for HDLs. OR Gate industries, Verilog and VHDL are common. Verilog, one of Simulation Waveform from Xilinx ISim is shown in the Fig. the main Hardware Description Language standardized as 3 & the AND Gate Simulation Waveform from Xilinx ISim IEEE 1364 is used for designing all types of circuits. It is shown in the Fig. 4.
Volume XII Issue X OCTOBER 2019 Page No: 601 A JOURNAL OF COMPOSITION THEORY ISSN : 0731-6755
Array Logic (PAL) devices. Finally, System Verilog is an extension to Verilog.
MATLAB 2017b Fig. 3 : OR Gate Simulation Waveform from Xilinx ISim
MATLAB® is a programming platform designed Debugging can be done before and after simulation. If errors specifically for engineers and scientists. The heart of are present, the HDL simulators usually give an error report MATLAB is the MATLAB language, a matrix-based and thereby corrections can be done. Also, by checking the language allowing the most natural expression of simulation result the errors can be found and rectified. computational mathematics. The language, apps, and built- ModelSim, Xilinx ISim, Aldec Active - HDL are some in math functions enable you to quickly explore multiple examples of HDL simulators. approaches to arrive at a solution. MATLAB lets you take
your ideas from research to production by deploying to enterprise applications and embedded devices, as well as integrating with Simulink® and Model-Based Design.
Fig. 4 : AND Gate Simulation Waveform from Xilinx ISim Millions of engineers and scientists in industry and academia use MATLAB. You can use MATLAB for a range of Different Types of HDLs applications, including deep learning and machine learning, signal processing and communications, image and video Different HDLs are available for describing analog circuits, processing, control systems, test and measurement, digital circuits and PCBs. computational finance, and computational biology. HDL Coder™ generates portable, synthesizable Verilog® and HDLs for digital circuit design VHDL® code from MATLAB® functions, Simulink® models, and Stateflow® charts. The generated HDL code Other than Verilog, VHDL and SystemC many HDLs are can be used for FPGA programming or ASIC prototyping available for digital circuits. Among them,Advanced and design. Boolean Expression Language (ABEL) is better for programming PLDs (Programmable Logic Devices). It Vision HDL Toolbox™ provides pixel-streaming includes sequential and concurrent logic formats, truth table algorithms for the design and implementation of vision formats and describes test vectors also. Altera Hardware systems on FPGAs and ASICs. It provides a design Description Language (AHDL), a language from Altera is framework that supports a diverse set of interface types, used for CPLDs and FPGAs. All language constructs can be frame sizes, and frame rates, including high-definition used in AHDL providing more control and (1080p) video. The image processing, video, and computer support. Bluespec, a high level functional programming vision algorithms in the toolbox use an architecture HDL language is developed to handle chip design and appropriate for HDL implementations. The toolbox automation systems. Bluespec System Verilog(BSV) uses algorithms are designed to generate readable, synthesizable syntax similar to Verilog HDL. C-to-Verilog is generally a code in VHDL and Verilog (with HDL Coder™). The converter which helps to convert C to Verilog language. generated HDL code can process 1080p60 in real time. Constructing Hardware in a Scala Embedded Language (Chisel), MyHDL and HHDL are HDLs used to support Toolbox capabilities are available as MATLAB System advanced hardware design. objects™ and Simulink® blocks. Computer Vision System Toolbox™ provides algorithms, functions, and apps for Confluence and CoWareC are the two languages which were designing and simulating computer vision and video discontinued after the arrival of the System C processing systems. You can perform feature detection, language. Compiler for Universal Programming extraction, and matching, as well as object detection and Language (CUPL) is generally used for logic device tracking. For 3D computer vision, the system toolbox programming. Handel-C is used for programming supports single, stereo, and fisheye camera calibration; FPGA’s. Hardware Join Java (HJJ) helps in reconfigurable stereo vision; 3D reconstruction; and 3D point cloud computing applications. Impulse C, subset of C supports processing. parallel programming. Just-Another Hardware Description Language (JHDL) uses an object oriented programming Algorithms for deep learning and machine learning enable approach. LavaHDL helps in specifying layout of circuits you to detect faces, pedestrians, and other common objects mainly. Lola HDL is basically used for synchronous digital using pretrained detectors. You can train a custom detector circuits. M, is another HDL from Mentor using ground truth labeling with training frameworks such Graphics. PALASM is used as a HDL for Programmable as Faster R-CNN and ACF. You can also classify image
Volume XII Issue X OCTOBER 2019 Page No: 602 A JOURNAL OF COMPOSITION THEORY ISSN : 0731-6755
categories and perform semantic segmentation. Algorithms Python’s developers strive to avoid premature optimization, are available as MATLAB® functions, System objects™, and reject patches to non-critical parts of CPython that and Simulink® blocks. For rapid prototyping and embedded would offer marginal increases in speed at the cost of system design, the system toolbox supports fixed-point clarity.When speed is important, a Python programmer can arithmetic and C-code generation. move time-critical functions to extension modules written in languages such as C, or use PyPy, a just-in-time compiler. Python interpreters are available for many operating Cython is also available, which translates a Python script systems. CPython, the reference implementation of Python, into C and makes direct C-level API calls into the Python is open source software and has a community-based interpreter. development model, as do nearly all of Python's other implementations. Python and CPython are managed by the Syntax and semantics non-profit Python Software Foundation. Python is a multi- paradigm programming language. Object-oriented Python is meant to be an easily readable language. Its programming and structured programming are fully formatting is visually uncluttered, and it often uses English supported, and many of its features support functional keywords where other languages use punctuation. Unlike programming and aspect-oriented programming (including many other languages, it does not use curly brackets to by metaprogramming and metaobjects (magic methods)). delimit blocks, and semicolons after statements are optional. Many other paradigms are supported via extensions, It has fewer syntactic exceptions and special cases than C or including design by contract and logic programming. Pascal.
Python uses dynamic typing, and a combination of reference Python uses whitespace indentation, rather than curly counting and a cycle-detecting garbage collector for memory brackets or keywords, to delimit blocks. An increase in management. It also features dynamic name resolution (late indentation comes after certain statements; a decrease in binding), which binds method and variable names during indentation signifies the end of the current block. Thus, the program execution. Python’s design offers some support for program's visual structure accurately represents the functional programming in the Lisp tradition. It has filter(), program's semantic structure. This feature is also sometimes map(), and reduce() functions; list comprehensions, termed the off-side rule. Statements and control flow of dictionaries, and sets; and generator expressions. The Python's statements include (among others) are standard library has two modules (itertools and functools) that implement functional tools borrowed from Haskell and The assignment statement (token ‘=’, the equals sign). Standard ML. The language's core philosophy is This operates differently than in traditional imperative summarized in the document The Zen of Python (PEP 20), programming languages, and this fundamental which includes aphorisms such as: mechanism (including the nature of Python’s version of variables) illuminates many other features of the 1. Beautiful is better than ugly language. 2. Explicit is better than implicit Assignment in C, e.g., x = 2, translates to “typed 3. Simple is better than complex variable name x receives a copy of numeric value 2”. 4. Complex is better than complicated The (right-hand) value is copied into an allocated 5. Readability counts storage location for which the (left-hand) variable name is the symbolic address. The memory allocated to the Rather than having all of its functionality built into its core, variable is large enough (potentially quite large) for the Python was designed to be highly extensible. This compact declared type. modularity has made it particularly popular as a means of In the simplest case of Python assignment, using the adding programmable interfaces to existing applications. same example, x = 2, translates to “(generic) name x Van Rossum's vision of a small core language with a large receives a reference to a separate, dynamically allocated standard library and easily extensible interpreter stemmed object of numeric (int) type of value 2”. This is termed from his frustrations with ABC, which espoused the opposite binding the name to the object. Since the name’s storage approach. While offering choice in coding methodology, the location doesn’t contain the indicated value, it is Python philosophy rejects exuberant syntax (such as that of improper to call it a variable. Perl) in favor of a simpler, less-cluttered grammar. As Alex Names may be subsequently rebound at any time to Martelli put it: “To describe something as ‘clever’ is not objects of greatly varying types, including strings, considered a compliment in the Python culture”. Python’s procedures, complex objects with data and methods, etc. philosophy rejects the Perl “there is more than one way to do Successive assignments of a common value to multiple it” approach to language design in favor of “there should be names, e.g., x = 2; y = 2; z = 2 result in allocating storage one—and preferably only one—obvious way to do it”. to (at most) three names and one numeric object, to
Volume XII Issue X OCTOBER 2019 Page No: 603 A JOURNAL OF COMPOSITION THEORY ISSN : 0731-6755
which all three names are bound. Since a name is a a generator function, and from Python 3.3, the information generic reference holder it is unreasonable to associate can be passed through multiple stack levels. a fixed data type with it. However, at a given time a name will be bound to some object, which will have a TensorFlow type; thus there is dynamic typing. The if statement, which conditionally executes a block TensorFlow is an open source software library for numerical of code, along with else and elif (a contraction of else- computation using data-flow graphs. It was originally if). developed by the Google Brain Team within Google's The for statement, which iterates over an iterable object, Machine Intelligence research organization for machine capturing each element to a local variable for use by the learning and deep neural networks research, but the system attached block. is general enough to be applicable in a wide variety of other The while statement, which executes a block of code as domains as well. It reached version 1.0 in February 2017, long as its condition is true. and has continued rapid development, with 21,000+ The try statement, which allows exceptions raised in its commits thus far, many from outside contributors. This attached code block to be caught and handled by except article introduces TensorFlow, its open source community clauses; it also ensures that clean-up code in a finally and ecosystem, and highlights some interesting TensorFlow block will always be run regardless of how the block open sourced models. TensorFlow is cross-platform. It runs exits. on nearly everything: GPUs and CPUs—including mobile The raise statement, used to raise a specified exception and embedded platforms—and even tensor processing units or re-raise a caught exception. (TPUs), which are specialized hardware to do tensor math The class statement, which executes a block of code and on. They aren't widely available yet, but we have recently attaches its local namespace to a class, for use in object- launched an alpha program. oriented programming. The def statement, which defines a function or method. TensorFlow execution model
The with statement, from Python 2.5 released on Graphs : Machine learning can get complex quickly, and September 2006, which encloses a code block within a deep learning models can become large. For many model context manager (for example, acquiring a lock before graphs, you need distributed training to be able to iterate the block of code is run and releasing the lock within a reasonable time frame. And, you'll typically want afterwards, or opening a file and then closing it), the models you develop to deploy to multiple platforms. allowing Resource Acquisition Is Initialization (RAII)- With the current version of TensorFlow, you write code to like behavior and replaces a common try/finally idiom. build a computation graph, then execute it. The graph is a The pass statement, which serves as a NOP. It is data structure that fully describes the computation you want syntactically needed to create an empty code block. to perform. This has lots of advantages: The assert statement, used during debugging to check It's portable, as the graph can be executed immediately for conditions that ought to apply. or saved to use later, and it can run on multiple The yield statement, which returns a value from a platforms: CPUs, GPUs, TPUs, mobile, embedded. generator function. From Python 2.5, yield is also an Also, it can be deployed to production without having operator. This form is used to implement coroutines. to depend on any of the code that built the graph, only The import statement, which is used to import modules the runtime necessary to execute it. whose functions or variables can be used in the current It's transformable and optimizable, as the graph can be program. There are three ways of using import: import transformed to produce a more optimal version for a
Support for distributed execution Python does not support tail call optimization or first-class
continuations, and, according to Guido van Rossum, it never TensorFlow’s high-level APIs, in conjunction with will. However, better support for coroutine-like computation graphs, enable a rich and flexible development functionality is provided in 2.5, by extending Python's environment and powerful production capabilities in the generators. Before 2.5, generators were lazy iterators; same framework. information was passed unidirectionally out of the generator.
From Python 2.5, it is possible to pass information back into
Volume XII Issue X OCTOBER 2019 Page No: 604 A JOURNAL OF COMPOSITION THEORY ISSN : 0731-6755
TensorFlow’s open source models
The TensorFlow team has open sourced a large number of models. You can find them in the tensorflow / models repo. For many of these, the released code includes not only the model graph, but also trained model weights. This means that you can try such models out of the box, and you can tune many of them further using a process called transfer learning. Here are just a few of the recently released models (there are many more):
The Object Detection API: It's still a core machine learning challenge to create accurate machine learning Fig. 6 : Asking ParseySaurus to parse a line from models capable of localizing and identifying multiple Jabberwocky objects in a single image. The recently open
sourced TensorFlow Object Detection API has Multistyle Pastiche Generator from the Magenta produced state-of-the-art results (and placed first in research work : “Style transfer” is what's happening the COCO detection challenge). under the hood with those fun apps that apply the style
of a painting to one of your photos. This Magenta model extends image style transfer by creating a single network that can perform more than one stylization of an image, optionally at the same time. (Try playing with the sliders for the dog images in this blog post.)
Xilinx Vivado
The out-of-the-box Object Detection model, derived from raneko via Flickr, CC BY-2.0 & asking ParseySaurus to Fig. 5 : The out-of-the-box Object Detection model, parse a line from Jabberwocky is shown in the Figs. 5 & 6 derived from raneko via Flickr, CC BY-2.0. respectively. Vivado Design Suite is a software suite produced by Xilinx for synthesis and analysis of HDL tf-seq2seq: Google previously announced Google designs, superseding Xilinx ISE with additional features for Neural Machine Translation(GNMT), a sequence-to- system on a chip development and high-level synthesis. sequence (seq2seq) model that is now used in Google Translate production systems. tf-seq2seq is an open Vivado represents a ground-up rewrite and re-thinking of the source seq2seq framework in TensorFlow that makes it entire design flow (compared to ISE), and has been easy to experiment with seq2seq models and achieve described by reviewers as “well-conceived, tightly state-of-the-art results. integrated, blazing fast, scalable, maintainable, and ParseySaurus is a set of pretrained models that reflect an intuitive”. Unlike ISE which relied on ModelSim for upgrade to SyntaxNet. The new models use a character- simulation, the Vivado System Edition includes an in-built based input representation and are much better at logic simulator. Vivado also introduces high-level synthesis, predicting the meaning of new words based both on with a toolchain that converts C code into programmable their spelling and how they are used in context. They are logic. Vivado has been described as a “state-of-the-art much more accurate than their predecessors, comprehensive EDA tool with all the latest bells and whistles particularly for languages where there can be dozens of in terms of data model, integration, algorithms, and forms for each word and many of these forms might performance”. never be observed during training, even in a very large corpus. 4. Conclusions
A brief review of the hardware & software tools that are going to be used by the authors for the sake of detection of glaucoma disease has been herewith presented in a nutshell in this paper as a result of which an overview or an brief idea is obtained about how the authors go ahead with the hardware implementation of the same as the hardware
Volume XII Issue X OCTOBER 2019 Page No: 605 A JOURNAL OF COMPOSITION THEORY ISSN : 0731-6755
implementation in line with the software implementation [13]. Donald L. Budenz, Robert T. Chang, Xiangrun Huang, leads to the validation of the work that is done by the Robert W. Knighton, James M. Tielsch, “Reproducibility of retinal nerve fiber thickness measurements using the researchers in the open market. This paper also gave a brief stratus OCT in normal and Glaucomatous eyes”, Jour. of insight into the research work on the detection of the second Investigative Ophthalmology & Visual Science, Vol. 46, largest disease in the world, i.e., the Glaucoma disease in the No. 7, pp. 2440-2443, Jul. 2005. human eyes @ the introductory level that is being taken up [14]. J. Liu, D.W.K. Wong, J.H. Lim, Huiqi Li, N.M. Tan, Z. Zhang, T.Y. Wong, R. Lavanya, “ARGALI-An automatic as the research work by us. cup to disc ratio measurement system for glaucoma analysis using level-set image processing”, Springer’s References 13th Int. Conf. on Biomed. Engg., pp. 559-562, CBME 2008, 3–6 Dec. 2008 Singapore. [1]. Ganesh Babu T.R., Ganesh S., Shenbagadevi, “Automatic [15]. Kavitha S., S. Karthikeyan, Dr. K. Duraiswamy, detection of glaucoma using fundus image”, European “Neuroretinal Rim Quantification in Fundus Images to Jour. of Scientific Res., ISSN 1450-216X, Vol. 59, Issue Detect Glaucoma”, IJCSNS Int. Jour. of Comp. Sci. & 1, pp. 22-32, 2011. Network Security, Vol. 10, No. 6, Jun. 2010. [2]. Murthi A. and M. Madheswaran, “Enhancement of optic [16]. Rozy Kumari, Narinder Sharma, “A Study on the Different cup to disc ratio detection in glaucoma diagnosis”, IEEE Image Segmentation Technique”, Int. Jour. of Engg. & Int. Conf. on Comp. Communication & Informatics Innovative Tech. (IJEIT), Vol. 4, Issue 1, ISSN 2277-3754, (ICCCI-12), Coimbatore, Tamil Nadu, India, pp. 1-5, Jan. pp. 284-289, Jul. 2014. 10-12, 2012. [17]. Krishna Kant Singh, Akansha Singh, “A Study of Image [3]. Tan N.M., J. Liu, D.W.K. Wong, F. Yin, J.H. Lim, T.Y. Segmentation Algorithms For Different Types of Images”, Wong, “Mixture model-based approach for optic cup Int. Jour. of Comp. Issues (IJCSI), Vol. 7, Issue 5, ISSN segmentation”, IEEE Annual Int. Conf. on Engg. in 1694-0784, Sept. 2010. Medicine & Biology Society (EMBC), Buenos Aires, [18]. Sujata Sainil and Komal Arora, “A Study Analysis on the Argentina, pp. 4817 – 4820, 31 Aug-4 Sept. 2010. Different Image Segmentation Techniques”, RI [4]. Yuji Hatanaka, Atsushi Noudo, Chisako Muramatsu, Publications Int. Jour. of Info. & Computation Tech., Akira Sawada, Takeshi Hara, Tetsuya Yamamoto, Hiroshi ISSN 0974-2239, Vol. 4, No. 14, pp. 1445-1452, 2014. Fujita, “Automatic measurement of cup to disc ratio based [19]. Shilpa Kamdi, R.K. Krishna, “Image Segmentation and on line profile analysis in retinal images”, 33rd Annual Int. Region Growing Algorithm”, Int. Jour. of Comp. Tech. & Conf. of the IEEE Engg. in Medicine & Biology Society, Electronics Engg. (IJCTEE), Vol. 2, Issue 1, ISSN 2249- pp. 3387 – 3390, Boston, MA, USA, 30 Aug.-3 Sept. 2011. 6343, pp. 103-108, Feb. 2012. [5]. Fengshou Yin, Jiang Liu, Damon Wing Kee Wong, Ngan [20]. T. Chanwimaluang and G. Fan, “An efficient blood vessel Meng Tan, Carol Cheung, Mani Baskaran, Tin Aung, Tien detection algorithm for retinal images using local entropy Yin Wong, “Automated segmentation of optic disc and thresholding,” IEEE Proceedings of the 2003 Int. Symp. optic cup in fundus images for glaucoma diagnosis”, 25th on Circuits & Systems, ISCAS’03, Bangkok, Thailand, IEEE Int. Symp. on Comp. Based Medical Systems Vol. 5, pp. 21–24, 25-28 May 2003. (CBMS), Rome, Italy, pp. 1-6, 20-22 Jun. 2012. [21]. C.A. Lupascu, D. Tegolo, and E. Trucco, “Fabc : Retinal [6]. Sandra Morales, Valery Naranjo, Jesús Angulo, Mariano vessel segmentation using adaboost,” IEEE Trans. on IT Alcañiz, “Automatic detection of optic disc based on PCA in Biomedicine, Vol. 14, No. 5, pp. 1267–1274, May 2010. and mathematical morphology”, IEEE Trans. on Medical [22]. Malay Kishore Dutta, Amit Kumar Mourya, Anushikha Imaging, Vol. 32, Issue 4, pp. 786-796, Apr. 2013. Singh, M. Parthasarathi, Radim Burget, Kamil Riha, [7]. Jun Cheng, Jiang Liu, Yanwu Xu, Fengshou Yin, Damon “Glaucoma Detection by Segmenting the Super Pixels Wing Kee Wong, Ngan-Meng Tan, Dacheng Tao, Ching- from Fundus Colour Retinal Images”, IEEE Int. Conf. on Yu Cheng, Tin Aung, Tien Yin Wong, “Superpixel Medical Imaging, m-Health & Emerging Commn. Syst. classification based optic disc and optic cup segmentation (MedCom), Greater Noida, UP, India, pp. 86 – 90, 7-8 for glaucoma screening”, IEEE Trans. on Medical Nov. 2014. Imaging, Vol. 32, Issue 6, pp. 1019-1032, Jun. 2013. [23]. Siddalingaswamy P.C., Gopalakrishna Prabhu K., [8]. Aquino, Arturo, Manuel Emilio Gegúndez-Arias, and “Automatic Localization and Boundary Detection of Optic Diego Marín, “Detecting the optic disc boundary in digital Disc Using Implicit Active Contours”, Int. Jour. of Comp. fundus images using morphological, edge detection, and Applications (IJCA), ISSN 0975 – 8887, Vol. 1, No. 7, feature extraction techniques”, IEEE Transactions on pp. 1-5, 2010. Medical Imaging, Vol. 29, Issue 11, pp. 1860-1869, Nov. [24]. Zhuo Zhang, Feng Shou Yin, Jiang Liu, Wing Kee Wong, 2010. Ngan Meng Tan, Beng Hai Lee, Jun Cheng, Tien Yin [9]. J. Liu, F.S. Yin, D.W.K. Wong, Z. Zhang, N.M. Tan., Wong, “Origa-light : An online retinal fundus image C.Y. Cheung, M. Baskaran, T. Aung, T.Y. Wong, database for glaucoma analysis and research,” Int. Conf. of “Automatic glaucoma diagnosis from fundus image”, IEEE Eng. in Med and Bio. Soc., Buenos Aires, Argentina, 2011 Annual International Conference of the IEEE pp. 3065–3068, 31 Aug.-4 Sept. 2010. Engineering in Medicine and Biology Society, Boston, [25]. Hafsah Ahmad, Abubakar Yamin, Aqsa Shakeel, Syed MA, USA, pp. 3383 – 3386, 30 Aug.-3 Sept. 2011. Ormer Gillani, Umer Ansari, “Detection of Glaucoma [10]. Rafael C. Gonzalez, Richard E. Woods, “Digital Image Using Retinal Fundus Images”, Int. Conf. on Robotics & Processing using Matlab”, Pearson Prentice Hall, New Emer. Allied Technol. in Engg., Islamabad, Pakistan, pp. York, USA, 4th Edn., 2004. 321 – 324, 22-24 Apr. 2014. [11]. Yong Yang, Shuying Huang, “Image segmentation by fuzzy c-means clustering algorithm with a novel penalty term”, Computing and Informatics, Vol. 26, No. 1, pp. 17– 31, 2007. Prof. Fazlulla Khan was born on 21st of July 1970 in Chitradurga [12]. Deepak S.K, “Automatic Assessment of Macular Edema Dist of Karnataka State. He completed his schooling in from Color Retinal Images”, IEEE Trans. on Medical Government High School in Chitradurga & later the Pre-University Imaging, Vol. 31, No. 3, pp. 766 – 776, Mar. 2012. Course in Government PU College in Chitradurga. He completed
Volume XII Issue X OCTOBER 2019 Page No: 606 A JOURNAL OF COMPOSITION THEORY ISSN : 0731-6755
his Bachelor of Engg. in Electronics & Communication Engg. from the reputed SJMIT (Kuvempu University, Shimoga), Chitradurga, followed by M.Tech. (Digital Electronics) from SSIT, Tumakuru (VTU, Belagavi) in the year 1996 & 2011 respectively in First Class. Currently he is working as a Assistant Professor in the Dept. of E & CE in HMSIT, Tumakuru since 18 years & at the same time, he is pursuing his Ph.D. programme in VTU as a research scholar in VTU Research Centre in the EEE Dept. of Govt. UBDT College of Engg., in Davanagere. He has got a teaching experience of more than 21 years in various engineering colleges across the state of Karnataka. His areas of interest are Microprocessor, Microcontrollers, Embedded Systems, RTS, Digital Electronics, SS, DSP, DIP & the relevant labs.
Ashok Kusagur was born in the year 1970 and he received the B.E. degree in Electrical & Electronics Engg. (EEE) from Bapuji Institute of Engg. & Tech. (BIET), Davanagere, Karnataka, India from Kuvempu University in the year 1996 and the M.Tech. Degree in Power Electronics from PDA College of Engg., Gulbarga, Karnataka, India from the VTU in the year 2001. He has completed his doctorate, Ph.D. from the reputed JNTU, Hyderabad under the guidance of Dr. S.F. Kodad & Dr. B.V. Sankar Ram in the year 2010. He has got a teaching experience of more than 20 years. Currently, he is working as Associate Professor & head of the department of EEE Dept. in UBDTCE, Davanagere, Karnataka and is also the Chairman- Department of Studies in E & E Engineering in the UBDTCE (a constituent college of the Visvesvaraya Technological University) He has published a large number of papers in various national & international conferences & journals apart from guiding a number of research scholars under his supervision. His areas of interests are neural networks, fuzzy logic, artificial intelligence, power electronics, field theory, networks, control systems, Matlab, etc.
Volume XII Issue X OCTOBER 2019 Page No: 607