A VLSI Image Processing Algorithm

for Biomedical Applications

[Link] Kumari 1 ✯, Alugonda Rajani 2 , Allanki Sanyasi
Rao 3
1
Department of Computer Science & Engineering, Malla Reddy
Institute of Technology, Secunderabad, Telangana, India.
2
Department of Electronics & Communication Engineering, University
College of Engineering Kakinada, JNTUK, Kakinada, Andhra Pradesh,
India.
3
Department of Electronics & Communication Engineering, Christu
Jyothi Institute of Technology & Science, Jangaon, Telangana, India.

Received: 16.05.2023 ❼ Accepted: 21.05.2023 ❼ Published Online: 31.05.2023

Abstract: The domain of signal processing that is dedicated to applications in the medical sector has shown
phenomenal expansion over the course of the last decade. It’s possible that developments in fields such as
computer-aided design (CAD), healthcare, and automated signal processing have played a key part in this
expansion. As a result of the current COVID-19 epidemic, there is a greater need for biomedical signal analysis,
and from an imaging point of view, biomedical research projects far exceed those that use more basic signal
processing methodologies. The fundamentals of image processing and the processing of biological pictures
have been the subject of a lot of studies, and the findings of these studies have shown a variety of possible
applications in a variety of settings. A examination of the existing literature reveals that there are a limited
number of solutions that are based on very large scale integration (VLSI). As a direct consequence of this, this
chapter devotes a significant amount of its focus to the investigation of biological image processing, particularly
in terms of its applications and VLSI implementations. It is hoped that researchers and investigators would
✯ Correspondence: Associate Professor, Department of Computer Science & Engineering, Malla Reddy Institute of
Technology, Secunderabad, Telangana, India. Email:lalithakumari4@[Link]
[Link]
Volume-1, Issue-5, PP:17-25 (2023)

17
This work is licensed under a Creative Commons Attribution 4.0 International License CC BY-NC-ND 4.0.
 International Journal of Scientific Methods in Engineering and Management (IJSMEM)

find this useful in better preparing a solution to this issue so that it may be put into action. Following an
analysis of the most recent developments in research, a number of possible directions for future research have
been contemplated.

Key words: Biomedical, Computer aided diagnosis, VLSI implementations, Image processing.

1. INTRODUCTION
The field of signal processing and the applications that are directly connected to it have almost
made ”image processing” into a household phrase. A digital picture is handled like a signal in the
process known as digital image processing (DIP), which involves a large variety of different operations.
DIP procedures may be used to a broad range of digital photos, including those that are utilised
in the medical industry, general-purpose landscape images, satellite images, remotely sensed images,
underwater images, photographs of various human and animal body parts and organs, photographs
of plants, and so on. The processing of images with the particular objectives of enhancement,
segmentation, and restoration [1] amongst others, has led to numerous applications in fields such
as biometrics medical imaging[2]. In recent years, the field of medicine has made substantial use
of image processing systems for a variety of goals, including increased performance and automated
diagnosis-[3]. As has been mentioned previously [4] medical image processing is a general-purpose area
of medical science that employs image processing concepts in the diagnosis and analysis of a variety
of medical image modalities. Medical imaging modalities include, but are not limited to, radiographs,
mammograms, magnetic resonance imaging, computed tomography, and x-rays. A few examples of
biological signals are electroencephalograms and ultrasounds. In the field of cognitive research, the
electroencephalogram (EEG) is one of the biological signal modalities that has received the most
attention and is used to the greatest extent. For the purpose of processing biomedical photographs
such as EEG and MEG, biological image signal processing is used. As can be seen in figure 1 each and
every biomedical image signal is an example of an imaging modality that goes through a certain set
of procedures. The most important stages are as follows:
❼ Categorization and post-processing.
❼ Image pre-processing; image segmentation; image acquisition.
❼ The extraction of features
To begin, a picture is captured by means of a high-tech sensing system made up of embedded and
intelligent sensors. Several researchers have stressed the importance of cleaning up sensor-generated
images of artefacts and unwanted effects before moving forward with image processing [5]. In the pre-
processing phase, any problems present in the original raw images are fixed, and artefacts and noise
are removed. A lot of extra work, such resizing and reformatting, is done during the pre-processing
step. There is a large range of image enhancement techniques and filters that may be applied in
the spatial and frequency domains. The median filter is widely regarded as one of the most effective
methods for improving picture quality. After photos have been pre-processed or enhanced, picture
segmentation is employed to identify the ROI. By focusing on what’s most important, we may avoid

18
 International Journal of Scientific Methods in Engineering and Management (IJSMEM)

Figure 1. Important stages of image processing

wasting time and resources by considering irrelevant factors. Whether the operation is pixel-based
or region-based, as well as the colour, texture, and other aspects of the image, segmentation can be
broken down into a wide variety of subtypes. In order to train photos and extract the necessary
characteristics, a training method is used that is fed the necessary portion of the whole picture. Many
other kinds of features may be selected, depending on factors like texture, intensity, size, form, and
so on. Some characteristics are trained repeatedly to distinguish them from others and improve their
use for the application. Post-processing, which may entail the classification application, is when the
positive traits are ultimately put to use, as discovered by [6]. A picture might show a person who
has been diagnosed with cancer or someone who is in the early stages of the disease and has not yet
been diagnosed. Electroencephalograms, sometimes known as EEGs, are the primary focus of research
and development in the field of biomedical signal processing. Philosophical discussions; and more all
fall under the umbrella of cognitive science. When studying the brain, electroencephalography (EEG)
is the most often used image modality Bran imaging signal analysis and its application have been
growing at a rapid rate during the whole pandemic as a research tool for the investigation of post-
infection psychological and mental consequences in patients with COVID19[7]. Here are some of the
most common applications of EEG and related brain signals:

19
 International Journal of Scientific Methods in Engineering and Management (IJSMEM)

2. CURRENT RESEARCH TRENDS

There is still a lot of work to be done in the fundamentals of biomedical image processing and
its related applications, as seen by the literature on VLSI implementations for biomedical signals
processing noted above. We reviewed the material that is currently being distributed to identify the
breadth and relevance of the activity being undertaken by the research and scientific communities
due to the restricted hardware support for the same [8]. This page gives a brief overview of the
most important findings from the related literature. De-noising, image pre-processing, and image
segmentation are some of the most crucial aspects of image. Medical imaging and diagnostics are the
second most widespread use of image processing used widely in all contemporary diagnostic facilities
and hospitals [9]. The use of imaging technologies is widespread. Robotics, autonomous vehicles,
automated inspection in industry, defect identification in metal, satellite-based weather prediction,
remote-sensing-based navigation [10]. Table 1.1, we list the many studies that have contributed to
our understanding of vital aspects of biological signal processing. Google Scholar and the ACM
(Association for Computing Machinery) Digital Library have been scoured to find these contributions.

Table 1. Identifies the major contributions to the literature in the field of image processing and its applications
during the last five years

Research contributions in the last 5 years (year 2016 onwards)

Research area
Google scholar ACM databases Science direct database Scopus database IOP research
Image processing and applications 1,831,010 167,129 713,004 43,190 600
Biomedical image processing and its applications 116,100 167,228 88,610 18,034 600
Medical image processing 1,371,010 151,907 299,042 24,570 600
VLSI implementations for biomedical image signal processing 14,410 159,778 171 949 17

Since 2016, academics have begun analysing the effects that these databases have had in the
disciplines to which they are most closely related. The little research that has been conducted on
VLSI and hardware solutions for biological signal processing is outlined in Table 1.1. Instead of
relying on real hardware implementation and manifestation, the actual quantity that is given in the
table below is derived from research studies and simulations that were carried out for general reasons
[11]. These were carried out for the purpose of gaining insight. There has been a significant amount
of research conducted on broad image progression; yet, you won’t find very much of it in academic
publications. Although it is possible to consider medical imaging and applications to be a subset of
general image processing, the latter is the focus of a far greater amount of research. This is due to
the fact that clinical applications and diagnostic imaging are both included in the scope of what is
considered to be generic image processing. In a similar vein, the field of biomedical image processing
has gotten a far less amount of attention than its medical equivalent. IOP, Scopus, and Science Direct
are three examples of prospective research databases that demonstrate how important simulation is in
comparison to the actual implementation of hardware in VLSI research. A flow architecture that was
based on microprocessors was introduced by[9].
It was proposed that this design may be used in the field of medical image processing. This strategy
might, in theory, assist in the development of suitable technologies for completely automated medical

20
 International Journal of Scientific Methods in Engineering and Management (IJSMEM)

image processing and diagnosis. The construction of a multiprocessor system that is geared towards
producing 3D graphics is now under progress. Hardware for medical imaging and applications is often
insufficient because of restrictions in real-time image collecting and processing are some of the citations
that may be found here [12]. The authors of an interesting research use a technique that is implemented
using VLSI to compute the Euler number of a picture. This approach is described in the paper. Digital
pictures provide issues for a wide range of applications due to their massive file sizes, storage needs,
and processing times.
The picture representation derived from this research, which is based on Euler numbers, will make
the implementation of large-scale image processing simpler and more efficient. It was hypothesised
that the concept may be used in VLSI design, and research was done to investigate the computational
challenges posed by this approach. It was discovered that the Euler number representation makes
use of far less processing resources than the direct image domains do. This technology is illustrative
of a contemporary trend in VLSI-based image processing architectures and has the potential for a
variety of applications. Some examples of these applications include medical imaging and biometric
security systems, although the list is not exhaustive. A very basic implementation of the numerical
representation is accomplished by using very large-scale integration, also known as VLSI.

3. APPLICATIONS
In this article, a wide variety of applications for biomedical image and signal processing are covered,
with a primary focus on the hardware implementations of those applications suggests a management
architecture that is based on field-programmable gate arrays (FPGA). This would be used to handle
medical signal processing. This configuration makes use of a multiprocessor system operating at 193
MHz. This research for a master’s thesis was intended to develop an algorithm that could be used in real
time to segment and improve medical pictures in order to make them easier to read and comprehend.
The central processing unit design was used in a wide range of image processing methods, and the
on-board memory of the CPU made it possible to save interim discoveries. These various ways each
produced a good amount of fruit. What makes this field of study so fascinating is the possibility
of creating image processing systems that run in parallel. Medical imaging has an extremely wide
range of applications, some of which include the detection, diagnosis, and treatment of a wide variety
of medical conditions. In order to achieve these objectives, the field of medical imaging makes use
of a wide variety of imaging techniques. These types of diagnostic instruments include, but are not
limited to, mammograms, computed tomography (CT) scans, x-rays, and radiography. The diagnosis
of cancer, the examination of brain diseases, the localization of tumours, the evaluation of cognitive
function, the development of brain-computer interfaces, and a great number of other medical image
analysis professions may all stand to profit from the use of machine learning and AI-based approaches.
The introduction of robotic surgery is only one illustration of how sensor technologies, the Internet of
Things (IoT), and ”smart healthcare” are having an impact on the detection and treatment of a wide
variety of illnesses and ailments. In a similar vein, it is possible that this technology was instrumental

21
 International Journal of Scientific Methods in Engineering and Management (IJSMEM)

in the development of innovative approaches to both diagnosis and treatment.

It is possible that this application might be miniaturised with the use of very large-scale integration
(VLSI) technology, which would make it easier to design innovative portable gadgets. The article
”Medical Image Implementation Using VLSI and FPGA-Based Technologies,” which was published in
2006 by Altera Corporation, explores the possible application of VLSI and FPGA-based technologies
in the development of medical pictures. The paper is titled ”Medical Image Implementation Using
VLSI and FPGA-Based Technologies.” This article will explain the many diagnostics and therapeutic
applications that may be accomplished with the use of medical imaging. This article will guide you
through the process of building algorithms for hardware and FPGA by making use of a variety of
➤
different programming languages. MATLAB is one example of such a programme. Studies on this
topic have been published by a number of different research teams over the last few years. A wider
FPGA implementation for medical image processing and other related applications has been made
possible as a result of advancements in digital signal processing (DSP) systems, soft computing tools,
and development methodologies.
Recent research has demonstrated that VLSI designs can be effective in high-performance image
processing applications, such as 3D imaging. It goes without saying that a significant amount of
investigation is necessary in order to come up with concepts that are both aesthetically pleasing and
useful. There are a few different examples of picture-guided interventions that can be found in the field
of image processing. Image segmentation. Figure 2 demonstrates a general FPGA implementation for
images. Digital / analog input/output, line processing are the three fundamental steps that comprise

Figure 2. FPGA implementation of an image

the actualization of FPGAs. A picture will function as both an output and an input at each step
of the implementation process. This is going to be the situation permanently. Because of this, the
programme can’t work without the ability to convert information between digital and analogue formats.

22
 International Journal of Scientific Methods in Engineering and Management (IJSMEM)

In recent years, the FPGA application has benefited from the widespread usage of real-time embedded
systems in biological settings. A sample FPGA circuit that can read and write picture data is seen
in Figure 3. Here’s a breakdown of the typical procedures this board follows: camera for machine
vision (MV), image capture via frame grabber, camera interface (SerDes), PCIe*, and picture output.
Throughout the design process, the FPGA board employs machine vision (MV) cameras for usage in
image processing software. According to research by [10] cameras are capable of gathering and partially
interpreting photographic evidence. SerDes provide a purpose similar to that of camera connections’
integrated circuits in that they allow for the rapid transfer of data. may perform a bidirectional
conversion of data between parallel and serial formats. The following method is used to collect the
picture frames and do some preliminary processing and noise removal. The integrated pre-processing
components of the frame capture are suitable for use with a wide variety of image data sources. The
PCIe express gateway allows the device, the motherboard, and the rest of the system to exchange
data at a blistering rate. Application-specific integrated circuit, or ASIC, design is also used for signal
processing purposes in the field of extremely large-scale integrated circuit design. The ASIC designs
may be modified to fit a variety of uses. Application-specific integrated circuit (ASIC) architecture is
shown from start to finish in Figure 3

Figure 3. Common FPGA board used in imaging applications.

Figure 4. A method for designing ASICs used in signal processing.

The following crucial processes are part of this flow:

❼ This requirement’s design specifications and design characteristics, such as area, power, speed, and
time, are similar to those for any signal processing design.

23
 International Journal of Scientific Methods in Engineering and Management (IJSMEM)

❼ Design of the architecture: The architecture is created in accordance with the given parameters.
❼ Real-time clock (RTC) modelling: This is used to create models for real-time clocks and element
synchronization.
❼ Synthesis.
❼ Physical design: The design is physically realized and produced.
❼ Layout and fabrication: The layout that will be used to create a chip is created and constructed.
❼ Chip: In the end, the chip is made for the image or signal processing task that uses the image as an
input for the ASIC design.

4. FUTURE RESEARCH DIRECTION

The research prospect and its potential are examined in this part in light of the study of recent literature
on the subject covered in this chapter. Highlighted the FPGA and CMOS based hardware in the form of
simulations and some hardware implementation to an observable level when implementing deep neural
network-based architecture in hardware for biomedical and healthcare applications. The computer
vision system was explored taking EMG (electromyography) inputs for biomedical analysis, which is
a crucial part of contemporary healthcare. The hardware devices proposed in this research and the
AI methodologies can both be of significant use in contemporary diagnostics. Design problems are
seen in a variety of hardware designs and applications related to image sensors, power consumption,
memory technologies, storage problems, system integration, robustness in segmentation tasks, etc. An
effort is being made in this direction so that the significant design problems, which are development
limitations, are addressed. In for image and video processing and their applications for constructing
various types of automated computer vision (CV) based systems and tools, hardware implementation
methods and associated design difficulties are examined. Because the problems differ from technology
to technology, FPGA and ASIC.

5. CONCLUSION
In this chapter, we provided an introduction to biomedical image processing and its applications. We
placed a particular focus on the present state of research on VLSI implementations of biomedical
application, which is discussed in more detail later in the chapter. The use of VLSI, ASIC design,
and FPGA applications have all been investigated in regard to the challenges that they now face
when it comes to image processing. In this chapter, we will go through the procedures that need to
be completed in order to integrate FPGA image processing into an application that does biomedical
image processing as an embedded system.

References

[1] Rafael C Gonzalez. Digital image processing. Pearson education india, 2009.

24
 International Journal of Scientific Methods in Engineering and Management (IJSMEM)

[2] Fei Kou, Weihai Chen, Changyun Wen, and Zhengguo Li. Gradient domain guided image filtering. IEEE
Transactions on Image Processing, 24(11):4528–4539, 2015.
[3] Kuk-Jin Yoon and In So Kweon. Adaptive support-weight approach for correspondence search. IEEE
transactions on pattern analysis and machine intelligence, 28(4):650–656, 2006.
[4] Sylvain Paris and Frédo Durand. A fast approximation of the bilateral filter using a signal processing
approach. In Computer Vision–ECCV 2006: 9th European Conference on Computer Vision, Graz,
Austria, May 7-13, 2006, Proceedings, Part IV 9, pages 568–580. Springer, 2006.
[5] Kaiming He, Jian Sun, and Xiaoou Tang. Guided image filtering. IEEE transactions on pattern analysis
and machine intelligence, 35(6):1397–1409, 2012.
[6] Kaiming He and Jian Sun. Fast guided filter. arXiv preprint arXiv:1505.00996, 2015.
[7] Rucha R Thakur, Swati R Dixit, and AY Deshmukh. Vhdl design for image segmentation using gabor
filter for disease detection. International Journal of VLSI design & Communication Systems, 3(2):211,
2012.
[8] Ms Priyanka, S Chikkali, and K Prabhushetty. Fpga based edge detection and segmentation. international
journal of advanced engineering sciences and technologies, 9(2):187–192, 2011.
[9] SR Dixit and Amol Y Deshmukh. Design strategies for edge based image representation system with vlsi.
In 2013 6th International Conference on Emerging Trends in Engineering and Technology, pages 115–116.
IEEE, 2013.
[10] Ashutosh Mishra, Pulak Mondal, and Swapna Banerjee. Vlsi-assisted nonrigid registration using modified
demons algorithm. IEEE Transactions on Very Large Scale Integration (VLSI) Systems, 23(12):2913–2921,
2015.
[11] Huan Li and Yan Xin. Modified cordic algorithm and its implementation based on fpga. In 2010 Third
International Conference on Intelligent Networks and Intelligent Systems, pages 618–621. IEEE, 2010.
[12] Linbin Chen, Fabrizio Lombardi, Jie Han, and Weiqiang Liu. A fully parallel approximate cordic design.
In 2016 IEEE/ACM International Symposium on Nanoscale Architectures (NANOARCH), pages 197–202.
IEEE, 2016.

25