ISSN (Online) 2581-9429

IJARSCT
International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)

Volume 2, Issue 1, September 2022

Impact Factor: 6.252

Medical Image Enhancement using Deep Learning

Mrs. Nilima Patil1, Dr. M. M Deshpande2, Dr. V. N. Pawar3
Research Scholar, Department of Computer Engineering1
Professor, Department of Computer Engineering2
Principal, A.C Patil College of Engineering, Kharghar, Navi Mumbai, Maharashtra, India3
A.C Patil College of Engineering, Kharghar, Navi Mumbai, Maharashtra, India
[email protected], [email protected], [email protected]

Abstract: Image Enhancement is one of many vital and complicated techniques in medical image
processing. The main aim of image enhancement is to boost the visual appearance of an image, or to offer
a “better rework'' illustration of the image. Numerous types of images like medical images from totally
different issues like poor distinction and noise. To enhance the contrast of an image and remove the noise is
necessary to maintain image quality. X-ray and ultrasound imaging are the most preferred medical
imaging technologies which are important for diagnosis of disease, but the edges and borders of the images
are not as clear as expected due to interference and low intensity in images. Speckle noise is the main
reason behind the degradation of quality in ultrasound images. It is mandatory to remove this noise to get
better images. Several techniques have already been projected and devised for enhancing the
digital images. Deep learning algorithms, particularly convolutional networks, have speedily become a
strategy of selection for analyzing medical images. The main aim of this paper is to provide an overview of
different techniques used in medical image enhancement and how deep learning provides the better
solution for enhancement.

Keywords: Medical imaging; Deep learning; Image enhancement; X-Ray, Ultrasound, Denoising.

I. INTRODUCTION
Image enhancement is one of the important step in image processing. It is mostly used to improve the image quality by
emphasizing some information and reducing some unnecessary information of an image according to users' need. There
are two purposes of image enhancement, one is to improve the visual effect, and the other is to make the image become
easier for the computer to process. Medical image enhancement is intense to boost the perceivability of the image.
Image enhancement offers many benefits like denoising, sharpening edges, increasing brightness, increasing contrast,
etc. which helps the doctors and the physicians to check fine details of the image more carefully. Enhancement methods
include contrast adjustment, histogram equalization, edge preserving filters, wavelet-based approaches, deblurring,
denoising etc. Medical images are often obtained with low contrast, low-intensity, noise, blurred, etc. due to limitations
in the medical devices. Such images have poor quality and cannot be used for diagnostic purpose. Thus, it is necessary
to process the images accurately to produce better-quality images with improved details for better analysis. In the past,
different strategies have been introduced by various researchers to deal with the low quality medical images. Numerous
search methods are discussed and it is necessary to process the images accurately to produce better-quality images with
improved details for better analysis. Some methods are discussed below in the literature survey section. The digital
image processing offers a significant effect on decision-making procedure based on some predictions. It gives better
features extraction and accuracy. Medical imaging is the rapidly growing research area that is used to diagnose a
disease for early treatment. The function of image processing within the health domain is relative to the growth. The
procedure of functioning assessment is complicated and contains numerous diverse properties. The digital image
processing techniques are implanted in many different computer systems. Artificial intelligence is the main domain and
machine learning and deep learning works under this domain. Medical imaging refers to completely different
technologies that are used to view the human body in order to diagnose, monitor, or treat medical conditions. It
includes techniques like X-Ray, MRI, ultrasound, CT. PET, endoscopy etc. Medical imaging is incredibly vital in aid
world-wide for physicians to diagnose, treat and prevent.

Copyright to IJARSCT DOI: 10.48175/IJARSCT-7076 106

www.ijarsct.co.in
 ISSN (Online) 2581-9429
IJARSCT
International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)

Volume 2, Issue 1, September 2022

Impact Factor: 6.252
II. LITERATURE REVIEW
2.1 Medical Image Processing
X-ray and ultrasound imaging are the most preferable medical imaging technologies which are important for diagnosis
of disease, but the edges and borders of the images are not as clear as expected due to interference and low intensity in
images. Speckle noise is the main reason behind the degradation of ultrasound images [1] It is mandatory to remove
speckle noise to get better images.. The goal of image enhancement is to improve the quality of image or certain
features. The accuracy of medical diagnosis depends on the accuracy of image recognition, so the details of the
processioning of medical images are more stringent, especially the edges that contain a lot of information need to be
enhanced.The various steps are performed on medical images before the detection of output. Initially, the medical
image is inserted as input to the machine and deep learning algorithms. After that, the image is fragmented into
different segments to recognize the interested area. Then, the features extraction performed through information
retrieval techniques. Next,the desired features are selected and the noise is removed. The classifier is used to classify
the extracted data and make predictions based on this classification. These steps are generally used in machine learning
experiment. The supervised, semi-supervised, unsupervised, reinforcement and active learning algorithms are the main
categories of machine learning. Further, the deep learning methods are basically advanced phases of machine learning
algorithms that classify data and predict more accurately using neural networks [2].There are different steps of machine
learning and deep learning algorithms [3] which are used in the medical imaging domain as shown in Figure 1.

Fig.1 Machine and Deep Learning algorithms workflow in medical images

This section highlights the literature survey of various approaches used for medical images (Specially X-ray and
Ultrasound image) enhancement by different researchers in the field of image processing. Researchers have provided
an overview of spatial domain techniques for image enhancement processing. Mostly, processing methods
categorization based on representative image improvement techniques. [4]Global histogram equalization does not work
effectively for images containing low contrast regions of right or dark areas. Adaptive histogram equalization (AHE)[5]
is the change to the histogram equalization, which applied on medical Chest and X-ray images. Compared to PSNR and
SSIM, the RLE technique showed higher values and better image quality. Authors developed a method of enhancement
by incorporating Laplacian, Sobel operator, addition operation, filter, product operation and power law transformation
techniques to enhance medical images. This developed method is tested on low contrast medical X-ray images and by
observing the results it can be said that these applied methods perform well for enhancing medical image. Author[6]
proposed an improved image enhancement method on digital chest radiography using the so-called N-CLAHE method,
which is based on global and local enhancement. This technique consists of two main steps. Firstly, intensity correction
of the raw image is encountered by the log-normalization function which adjusts the intensity contrast of the image
dynamically and Contrast Limited Adaptive Histogram Equalization (CLAHE) method is used for enhancing small
details like textures and local contrast of the images. This approach compared with conventional enhancement methods,
Copyright to IJARSCT DOI: 10.48175/IJARSCT-7076 107
www.ijarsct.co.in
 ISSN (Online) 2581-9429
IJARSCT
International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)

Volume 2, Issue 1, September 2022

Impact Factor: 6.252
such as histogram equalization, unsharp masking, CLAHE. The results show that the proposed N-CLAHE method
yields great improvement on the pre-processing correction for digital chest radiography.Zhou, Hui, and ShuaiWu [7]
combine various enhancement methods in the spatial domain and apply python software to medical image
enhancement. The experimental results show that the algorithm effectively retains the texture and details in the medical
image along with enhancing the contour and edge of the medical image, so that some details cannot be observed by the
naked eye. Python makes the algorithms more easy, effective, fast and more conducive to medical staff to observe the
focus of the medical image.

2.2 Ultrasound Imaging

Ultrasound imaging devices are more secure and real-time than other medical diagnosis devices. Therefore, it is
difficult to diagnose lesions using ultrasound images due to the speckle noise. In the case of medical ultrasound images,
said noise can produce uncertainty in the diagnosis because details, such as limits and edges, should be preserved.
Author [8] describestechniques that mainly focus on the smoothing or elimination of speckle noise in medical
ultrasound images. Speckle noise filtering methods implemented in US imageswhich has an impact on legal metrology
for healthcare services focused on patient safety. [9] The author has proposed an algorithm using speckle reducing
anisotropic diffusion (SRAD) and a guided filter for speckle noise reduction and edge preservation. Kumar, M. Ravi, J.
Lakshmi Prasanna& et al.[10] has used Median filtering, Fourier filters, Sobel filters, histogram equalization, Canny
Edge Detection techniques for improving image quality. In terms of edge safety, the performance of the proposed
method is calculated by a mean structural similarity in relation to the other methods. The results of the proposed
approach shows that the proposed approach can enhance the image quality of ultrasound images by improving image
quality while effectively minimizing speckle noise.Many researchers have worked on various machine learning
algorithms for medical image analysis as well as image enhancement techniques. X-ray non-destructive testing (NDT)
[11] is a primary detection technology in industrial fields, providing an effective detection for fragile and complex
structures without destructing components. In this study, author proposed the principle of convolution neural network
(CNN) and a Laplacian filter for edge detection model with improved performance. [12] Parallel computing is used to
accelerate the computations within the reduction of speckle noise in medical ultrasound images. The author has
projected a new hybrid algorithm for speckle noise reduction in medical ultrasound images and compared with existing
efficient methods. The experiments conducted [13] on various test images illustrate that the proposed algorithm
outperforms the various classic denoising algorithms including block matching 3-D (BM3D) and optimized Bayesian
nonlocal means filter. The author [14] provides novel approach low dimension local diagonal maxima -minima pattern
and local non-diagonal maxima-minima pattern edge detection techniques are developed for medical images such as
digital radiography, dental X-ray and ultrasound images. LDMMP and LNDMMP prove to be an efficient edge
detection technique for noisy medical images. In [15], the authors presented a method with multi resolution
Convolutional Neural Networks (CNN) within the framework of holomorphic filtering, which considers speckle noise
to be multiplicative; therefore, they implemented a logarithmic transformation of the data followed by the
exponentiation of the results.[16] Salem, Nema et. al, proposed HE, CHE, QDHE, and CLAHE algorithms and tested
them on retina, brain, endometrium, breast and knees images. Along with, it evaluated each of them using PSNR, SD
and MSE.HE spreads out intensity values along the total range of values in order to achieve higher contrast while CHE
has a better performance than HE. The QDHE [17] is a brightness-preserving algorithm that reduces limitations caused
by HE in contrast enhancement. The CLAHE is a recommended technique for images with non-uniform intensity
distribution over all areas of the image. In this way comparison of HE, CHE and CLAHE provided. In [18-19]author
had Compared different deep learning methods (e.g., Convolutional neural networks), transfer learning is characterized
by simplicity, efficiency and its low training cost, breaking the curse of small datasets. Comprehensive review is done,
including (i) structure of CNN, (ii) background knowledge of transfer learning, (iii) different types of strategies
performing transfer learning, (iv) application of transfer learning in various sub-fields of medical image analysis, and
(v) discussion on the future prospect of transfer learning in the field of medical image analysis. [20] In this paper, the
author proposed a novel and versatile bi-directional GAN, named Structure and illumination constrained GAN
(StillGAN), for medical image quality enhancement. This StillGAN treats low- and high-quality images as two distinct
domains, and introduces local structure and illumination constraints for learning both overall characteristics and local

Copyright to IJARSCT DOI: 10.48175/IJARSCT-7076 108

www.ijarsct.co.in
 ISSN (Online) 2581-9429
IJARSCT
International Journal of Advanced Research in Science,, Communication and Technology (IJARSCT)

Volume 2, Issue 1, September 2022

Impact Factor: 6.252
details. Extensive experiments on three medical image datasets (e.g., corneal confocal microscopy, retinal color fundus
and endoscopy images) demonstrate that this method performs better than both conventional methods and other dee deep
learning-based methods. In [21] four commonly used image enhancement techniques are explained, compared and
corresponding experimental results are shown. These enhancement techniques are as follows: I. Fuzzy Set Theory
Image Enhancement Method II. Histogram
Histogram Equalization Image Enhancement Method III. Histogram Matching Image
Enhancement Method IV. Equalized Histogram Equalization Image Enhancement Method. From the result it is said
that EHE method is the best method for image enhancement problems. HS method,
method, HE method and Fuzzy method are
the other better image enhancement methods respectively. However fuzzy method is a good enhancement method for
low contrast images only.Traditional medical image enhancement technologies [22-26]
[2 ] are generally divided into three
categories, namely the spatial domain methods, frequency domain methods, and deep learning methods. The method
about spatial domain mainly involves direct image processing including histogram transformation [[27], histogram
equalization [28], local gray level [29
29], edge extraction and smooth filtering [30].]. Meanwhile, the method about
frequency domain achieves enhancement via image transformation processing. The correlative algorithms include
homomorphic filter, high-frequency
frequency enhancement, wavelet methods,
methods, contour let method, and curvelet methods. The
methods using deep learning mainly achieve image enhancement by learning end-to-end
end end mapping from low resolution
to desired high resolution. The relevant algorithms include thermal image enhancement using C Convolutional Neural
Network (CNN) [31], multi-scale
scale mutual feature (MSMF)-convolutional
(MSMF) convolutional neural network (CNN)[(CNN)[32] etc. In recent
years, deep learning technology has been increasingly developing; one of the most well known CNNs has attained great
achievementss in text processing, voiceprint recognition, and image classification. Image
mage enhancement using CNN
method,, effectively enhanced the image quality. But still there is scope for medical image processing using machine
learning and deep learning.

III. PROPOSED SYSTEM

The reason why Deep Learning is much popular? In recent years, deep learning, a form of AI in which neural networks
learn patterns directly from raw data, has achieved remarkable success in image classification. Deep learning
algorithms, in particular Convolutional
onvolutional Neural Networks,
etworks, have rapidly become a methodology of choice for analyzing
medical images. We survey the use of deep learning for image classification, object detection, segmentation,
registration, and other tasks. The system consists ofof following flow for being operational.For this system development,
I used to take medical images (mostly, X-ray
X ray & ultrasound) as input and perform a deep learning approach to improve
the image quality for better understanding of images by radiologists to give better opinion about diseases. Figure 2, is
considered as a basic and simple representation of image enhancement processes.
The system consist of following flow being operational.
Step1. To begin with, upload the input image (X-ray/ Ultrasound) which is to be enhanced is taken into the model.
Step 2. Pre-processing
processing of image performed by Open CV methods
Step 3. Performed noise removal using deep learning judgment about choosing a proper image enhancement algorithm
is made.
Step 4. Last but not least, the image is processed using the selected image enhancement technique
Step 5. Analysis using different enhancement techniques with different parameter performed.
Step 6. lastly improved image by image enhancement is displayed as output.
output

Fig.2 System Architecture

Copyright to IJARSCT DOI: 10.48175/IJARSCT-7076 109

www.ijarsct.co.in
 ISSN (Online) 2581-9429
IJARSCT
International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)

Volume 2, Issue 1, September 2022

Impact Factor: 6.252
IV. CONCLUSION
This paper highlights the various image enhancement techniques which can be used particularly for medical image
enhancement which enable medical professionals. The paper briefly reiterated the facts that there still much
improvements are required in existing techniques to get better results. In order to overcome the limitations of earlier
techniques, deep learning algorithms, in particular convolutional neural networks, have rapidly become a methodology
of choice for analyzing medical images. This resulting system guarantees to be well-organized and precise,to give
better enhancement related to speckle noise reduction and image quality enhancement in X-ray and Ultrasound medical
images.

REFERENCES
[1]. Panayides, Andreas S., et al. "AI in medical imaging informatics: current challenges and future directions."
IEEE Journal of Biomedical and Health Informatics 24.7 (2020): 1837-1857.
[2]. Koonsanit, Kitti, et al. "Image enhancement on digital x-ray images using n-clahe." 2017 10th Biomedical
Engineering International Conference (BMEiCON). IEEE, 2017.
[3]. Latif, Jahanzaib, et al. "Medical imaging using machine learning and deep learning algorithms: a review."
2019 2nd International conference on computing, mathematics and engineering technologies (iCoMET).
IEEE, 2019.
[4]. Puttagunta, Muralikrishna, and S. Ravi. "Medical image analysis based on deep learning approach."
Multimedia Tools and Applications 80.16 (2021): 24365-24398.
[5]. Pourasad, Yaghoub, and Fausto Cavallaro. "A Novel Image Processing Approach to Enhancement and
Compression of X-ray Images" International Journal of Environmental Research and Public Health 18.13
(2021): 6724.
[6]. Islam, Shaikh Mahmudul, and Himadri Shekhar Mondal. "Image Enhancement Based Medical Image
Analysis." 2019 10th International Conference on Computing, Communication and Networking Technologies
(ICCCNT). IEEE, 2019.
[7]. Zhou, Hui, and Shuai Wu. "Design of medical image enhancement algorithm based on Python." 2021 IEEE
International Conference on Power Electronics, Computer Applications (ICPECA). IEEE, 2021.
[8]. Duarte-Salazar, Carlos A., et al. "Speckle noise reduction in ultrasound images for improving the
metrological evaluation of biomedical applications: an overview." IEEE Access 8 (2020): 15983-15999.
[9]. Choi, Hyunho, and Ji Chang Jeong. "Speckle noise reduction in ultrasound images using SRAD and guided
filter." 2018 International Workshop on Advanced Image Technology (IWAIT). IEEE, 2018.
[10]. Kumar, M. Ravi, J. Lakshmi Prasanna, Varanasi Sri Harsha Vardhan, Morla Bhaskar Vystanava Swamulu,
Gade Kalyan Kumar, Sreevardhan Cheerla, and Chella Santhosh. "Image enhancement for ultrasound images
using sobel edge detection." Journal of Critical review, ISSN- 2394-5125 Rev 7, no. 13 (2020)
[11]. Bharodiya, Anil K., and Atul M. Gonsai. "An improved edge detection algorithm for X-Ray images based on
the statistical range." Elsevier Heliyon 5.10 (2019): e02743.
[12]. Arnal, Josep, and Ilya Mayzel. "Parallel techniques for speckle noise reduction in medical ultrasound
images." Advances in Engineering Software 14 8 (2020): 102867.
[13]. Basha, Cmak Zeelan, et al. "Enhanced and Effective Computerized Classification of X-Ray Images." 2020
International Conference on Electronics and Sustainable Communication Systems (ICESC). IEEE, 2020.
[14]. Yadav, Navdeep, et al. "Local Diagonal Maxima-Minima Pattern-based Edge Detection Technique for
Ultrasound and Digital Radiography Images." IETE Journal of Research (2021): 1-11.
[15]. Xiao, Zimu, Ki-Young Song, and Madan M. Gupta. "Development of a CNN edge detection model of noised
X-ray images for enhanced performance of non-destructive testing." Measurement 174 (2021): 109012.
[16]. Salem, Nema, Hebatullah Malik, and Asmaa Shams. "Medical image enhancement based on histogram
algorithms." ScienceDirect Proceding Computer Science 163 (2019): 300-311.
[17]. Wang, Jian, et al. "A review of deep learning on medical image analysis." Mobile Networks and Applications
26.1 (2021)
[18]. Buettner, Ricardo, Marcus Bilo, Nico Bay, and Toni Zubac. "A Systematic Literature Review of Medical
Copyright to IJARSCT DOI: 10.48175/IJARSCT-7076 110
www.ijarsct.co.in
 ISSN (Online) 2581-9429
IJARSCT
International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)

Volume 2, Issue 1, September 2022

Impact Factor: 6.252
Image Analysis Using Deep Learning." In 2020 IEEE Symposium on Industrial Electronics & Applications
(ISIEA), pp. 1-4. IEEE, 2020.
[19]. Qiu, Tao, Chang Wen, Kai Xie, Fang-Qing Wen, Guan-Qun Sheng, and Xin-Gong Tang. "Efficient medical
image enhancement based on CNN-FBB model."IET Image Processing 13, no. 10 (2019): 1736-1744.
[20]. Ma, Yuhui, Jiang Liu, Yonghuai Liu, Huazhu Fu, Yan Hu, Jun Cheng, Hong Qi, Yufei Wu, Jiong Zhang, and
Yitian Zhao. "Structure and Illumination Constrained GAN for Medical Image Enhancement." IEEE
Transactions on Medical Imaging 40, no. 12 (2021): 3955-3967.
[21]. Irmak, Emrah. "A Comprehensive Reference Source for the Researchers Involved in Image Enhancement
Field–A Review." Journal of Psychological and Brain Sciences. Vol 2.5 (2017): 109-119.
[22]. J. Jai Jaganath Babu and G. Florence Sudha, “Adaptive speckle reduction in ultrasound images using fuzzy
logic on Coefficient of Variation,” Biomed. Signal Process. Control, vol. 23, pp. 93–103, 2016.
[23]. E. Irmak and A. H. Ertas, “A review of robust image enhancement algorithms and their applications” The 4th
IEEE International conference on Smart Energy Grid Engineering,August 21-24, 2016.
[24]. Gandhamal, A., Talbar, S., Gajre, S., et al.: ‘Local gray level s-curve transformation – a generalized contrast
enhancement technique for medical images’, Comput. Biol. Med., 2017, 83, (83), pp. 120–133
[25]. Daniel, E., Anitha, J.: ‘Optimum wavelet based masking for the contrast enhancement of medical images
using enhanced cuckoo search algorithm’, Comput. Biol. Med., 2016, 71, (71), pp. 149–155
[26]. Moreno, R., Smedby, Ö.: ‘Gradient-based enhancement of tubular structures in medical images’, Med. Image
Anal., 2015, 26, (1), pp. 19–29
[27]. Li, B., Xie, W.: ‘Adaptive fractional differential approach and its application to medical image
enhancement’, Comput. Electr. Eng., 2015, 45, (45), pp. 324–335.
[28]. Jiang, G., Lin, S.C., Wong, C.Y., et al.: ‘Color image enhancement with brightness preservation using a
histogram specification approach’, Opt.-Int. J. Light Electron. Opt., 2015, 12, (24), pp. 5656–5664
[29]. Isa, I.S., Sulaiman, S.N., Mustapha, M., et al.: ‘Automatic contrast enhancement of brain MR images using
average intensity replacement based on adaptive histogram equalization (AIR-AHE)’, Biocybern. Biomed.
Eng., 2017, 37, (1), pp. 24–34
[30]. Zeng, M., Li, Y., Meng, Q., et al.: ‘Improving histogram-based image contrast enhancement using gray-level
information histogram with application to X-ray images’, Opt.- Int. J. Light Electron. Opt., 2012, 123, (6),
pp. 511–520
[31]. Choi, Y., Kim, N., Hwang, S.: ‘Thermal image enhancement using convolution neural network’. IEEE/RSJ
Int. Conf. Intelligent Robots & Systems, Daejeon, Korea, October 2016
[32]. Liao, X., Zhang, X.: ‘Multi-scale mutual feature convolutional neural network for depth image denoise and
enhancement’. 2017 IEEE Visual Communications and Image Processing (VCIP), St. Petersburg, USA,
December 2017, pp. 1–4

Copyright to IJARSCT DOI: 10.48175/IJARSCT-7076 111

www.ijarsct.co.in