Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

REVIEW OF VARIOUS HISTOGRAM BASED MEDICAL

IMAGE ENHANCEMENT TECHNIQUES ∗

† ‡
Vidyasaraswathi H N Dr.M C.Hanumantharaju
Assistant Professor Associate Professor
Bangalore Institute of Technology, Dept of ECE BMS Institute of Technology
K R Road, Bangalore Avalahalli,Yelahanka Bangalore
Karnataka,India Karnataka ,India
vidya.gowdaec@gmail.com mchanumantharaju@gmail.com

ABSTRACT General Terms


Image enhancement is a processing on an image to make Algorithms,Theory
it more suitable for some applications. The main problem
addressed by this paper is the enhancement of medical im- Keywords
ages using efficient algorithms based on HE techniques. The Image enhancement, histogram equalization, FPGA, con-
paper involves analyzing and formulating different HE im- trast enhancement.
age enhancement techniques suitable for various medical ap-
plications. More precisely, proposed research will focus on 1. INTRODUCTION
the enhancement of medical images captured under poor In modern world due to the changing life style of human be-
illumination conditions, foggy situations and speckle noise ings, everyday new types of diseases are emerging. ItâĂŹs
etc.. Developing algorithms that would assist doctors to an everyday challenge for doctors to effectively diagnose
diagnose the disease in the beginning stage only. For ex- these diseases and provide remedy. Many of the diseases
ample the removal of speckle noise and artifacts with seg- like breast cancer, retinal fundus, kidney disorders, spinal
menting kidney from those images where kidney boundary card problems, heart problems result in enormous amount
are not much clear. Medical images are enhanced using ef- of digital data either in the form of sound, images, or videos.
ficient Histogram Equalization(HE) techniques such as It- Efficient analyses of these images are necessary for success-
erative dynamic HE, Dualistic sub-image HE, Background ful diagnosis of diseases. Sophisticated hardware and soft-
brightness preserving HE, Gray-level and gradient magni- ware knowledge is quite essential in this regard. Medical
tude HE. For improving the performance of image process- images like mammogram, retinal, ultrasound, bones, chest
ing systems the crucial solution is implementation of image etc., form a major diagnostic tool. All of these images can
processing techniques in hardware. Therefore for improving be captured by using hardware device like camera. These
the performance of image processing systems, for flexible de- images do not give the proper information required for anal-
sign development, more compact, low power, and high speed ysis of diseases due to electrical fluctuations, environmen-
and to reduce cost and time, gives the implementation of ef- tal change, not expert in using the device. Therefore, to
ficient histogram algorithms on Field Programmable Gate enhance the low contrast image and to highlight the im-
Array (FPGA). portant details of image, image enhancement is required.
Several researchers have proposed numerous image enhance-
ment algorithms like Histogram equalization (HE), gamma
Categories and Subject Descriptors correction, homomorphic filter etc. In recent years, digi-
I.4.3 [Enhancement]: Fpga [; []: ]
tal image enhancement have been widely used in many ap-
plications areas such as medicine, space exploration, auto-
∗(Does NOT produce the permission block, copyright
mated industry inspection, authentication etc. The image
information nor page numbering). For use with enhancement technique principal objective is to process a
ACM PROC ARTICLE-SP.CLS. Supported by ACM. given image to modify the attributes or features of an image
†Dr. Trovato insisted his name be first.
so that the result is most suitable than the original image
‡The secretary disavows any knowledge of this author’s ac- for a specific applications. It accentuates or sharpens im-
tions. age features such as edges, boundaries, or contrast to make
a graphic display more helpful for display and analysis to
diagnose a particular problem. The enhancement doesn’t
increase the inherent information content of the data, but it
increases the dynamic range of the chosen features so that
they can be detected easily. Low contrast image analysis is
a problem in medical image processing, since it reduces the
ability of pathologists in analyzing the images. Therefore,
a well known and straightforward scheme for contrast en-
hancement is HE. HE is a method that improves the contrast
in an image, in order to stretch out the intensity range. Due
to its simplicity, moderate and for better performance on im- 2.1 Histogram Equalization (HE)
ages HE method is used in most of the techniques. In this Histogram equalization is widely used for contrast enhance-
research work, study of various HE methods such as Global ment in a variety of applications like computer vision, re-
Histogram Equalization (GHE), Local Histogram Equaliza- mote sensing, medical image analysis etc., due to its simple
tion (LHE), Adaptive Histogram Equalization (AHE), Con- function and effectiveness. It works by using the cumulative
trast Limited Adaptive Histogram Equalization (CLAHE) density function of the image by flattening the histogram
etc is adopted and have to propose new algorithm to en- and stretching the dynamic range of the gray levels. One
hance medical images. HE methods improve the quality of problem of the histogram equalization is that the brightness
the information in an image; hence can be used in field of of an image is changed after the histogram equalization
medical where the quality of the scans can be enhanced to
diagnose the patientâĂŹs disease accurately and efficiently. 2.2 Adaptive HE Technique(AHE)
The Tumor images can be enhanced to segment the proper AHE differs from the normal HE method in the respect
boundary of the tumor which helps in accurate treatment that HE generates only one histogram whereas AHE method
for the patient. The eye images can be enhanced to get computes several histograms corresponding to a distinct sec-
proper view which helps to detect any eye diseases such as tion of the image and uses that to redistribute the intensity
Diabetic Macular edema, Glaucoma etc. For mismammo- values of the image. By using ADE method can improve
gram images to find exact location of cancerous regions and the detection of spiculations on dense mammographic back-
also for low dose CT images to find tiny anatomies like ves- grounds. Saeid et al.[3], proposed an adaptive HE method
sels, lugs nodules, airways, pulmonary fissures etc. Using for segmentation of blood vessels in color retinal images.In
HE improve the image contrast to enhance the image qual- this method, focus on two methods of retinal vessel segmen-
ity of retinal vessels used in diagnosis of diseases such as tation including first derivative of Gaussian matched filter
diabetes and high blood pressure. Similarly CT scans, MRI and Gaussian matched filter. AHE improves the image con-
scans, and Ultrasound and any other Bio-Medical related trast to enhance the image quality of retinal vessels used in
scans can be enhanced in concern of the accurate detection diagnosis of diseases such as diabetes and high blood pres-
and classification of diseases and also the treatment. The sure. This method showed an accuracy of 0.9353 a raise of
architectural approach used for implementation of image en- about 2 percent compared to existing methods. Analysis
hancement algorithms are classified into different categories of this approach shows that AHE method used in retinal
are General purpose processor(GPP), Digital signal proces- vessel segmentation is based on threshold. Therefore, this
sor(DIPs), Application specific integrated circuits(ASICs) approach is suitable for specific type of images
and Field programmable Gate Array(FPGA). Depending
on the time consumption, cost, speed, flexibility, memory
used, throughput and other requirements a particular ap-
2.3 Histogram Modified Contrast Limited Adap-
proach can be chosen. In GPP approach, the throughput tive Histogram Equalization (HM CLAHE)
of the system decreases and also it is limited in perfor- Sundaram et al. [26] proposed to adjust the level of con-
mance because its instructions are executed in sequence and trast enhancement, which in turn gives the resultant image
they are not suited for fast processing of high resolution a strong contrast and brings the local details for more rel-
of image enhancement methods. In DSP approach, only evant interpretation and this proposed technique provides
marginal improvement can achieved because parallelism and better contrast enhancement with preserving the local in-
pipelining incorporated in design are inadequate. The hard- formation of the mammogram images. Shibin et al. [7],
ware implementation of image enhancement technique has proposed a new algorithm for feature and contrast enhance-
been strongly influenced by the evolution of VLSI technol- ment of mammographic images based on multiscale trans-
ogy. The complex computation tasks such as 2D convolu- form and mathematical morphology. In this algorithm the
tion, filtering, averaging, normalization, smoothing and con- Laplacian Gaussian pyramid transform is applied first to
trast stretching operations encountered in image enhance- decompose the mammography into different multiscale sub-
ment techniques are well suited for VLSI implementations. band sub-images. Further the detail or high frequency sub-
In addition, VLSI based hardware implementation exploits images are equalized by using the Contrast Limited Adap-
pipelining and massive parallel processing operations, resul- tive Histogram Equalization (CLAHE) and low frequency
tinin increased speed and throughput of the image enhance- sub-images are processed by using mathematical morphol-
ment In ASIC implementation, the performance can be in- ogy. Finally, the enhanced image of feature and contrast
creased but it takes a large time to market and cost is high. is reconstructed from the Laplacian Gaussian pyramid coef-
So it is used for high volume productions. For high speed ficients modified by using CLAHE and mathematical mor-
and throughput FPGA becomes a right choice. The FPGA phology. Therefore, the performance evaluation of the this
implementation can use fully the data and I/O parallelism algorithm is based on, signal-noise-ratio (SNR) and contrast
for image enhancement applications. FPGA implementa- improvement index (CII). And the research result shows the
tion of image enhancement techniques is cost effective for algorithm yield better image contrast compared to AHE.
low volume productions. Therefore, in this research work
FPGAs are preferred in order to implement the histogram 2.4 Brightness Preserving Dynamic Histogram
based image enhancement algorithms.
Equalization (BPDHE)
In BPDHE method the original image is decomposed into
multiple sub images according to their local maxima, then
the dynamic histogram equalization is applied to each sub
2. CONVENTIONAL IMAGE ENHANCEMENT image and finally, the sub Images are combined. It di-
TECHNIQUES vides the histogram based on the local maxima. It pro-
duces the output image with the mean intensity almost equal ultrasound imaging detection of tissues and also to detect
to the mean intensity of the input, thus fulfils the require- structural abnormalities becomes difficult. In [11], proposed
ment of maintaining the mean brightness of the image. This a method to develop and implement a system that can seg-
method smoothes the input histograms with one dimensional ment human kidney from ultrasound images, usable during
Gaussian filter, and then partitions the smoothed histogram surgical operations like punctures but this method is not
based on its local maxima. After that it assigns new dy- suitable when image boundary is not clear. Therefore work
namic range to each partition. Then, the histogram equal- includes removal of speckle noise and artifacts with segment-
ization process is applied independently to these partitions, ing images where image boundaries are not much clear by
based on this new dynamic range and the output image is using different segmentation techniques. Ultrasound images
normalized to the input mean brightness. Nungsanginla et have poor contrast due to the presence of granular speckle
al. [17], provided an analysis and comparison of the various noise. In [13], proposed algorithm aims to improve the abil-
techniques of image enhancement through histogram equal- ity to differentiate between healthy and malignant condi-
ization are overviewed. The result is that Brightness pre- tions via the use of homomorphic filtering and OtsuâĂŹs
serving Dynamic Histogram Equalization (BPDHE) tech- gray-level histogram thresholding. Homomorphic filtering is
nique provides better and scalable brightness preservation based on its principle of reducing the dynamic range and
for images with poor contrast. The advantage of BBPHE thereby increasing the local contrast. But this method loses
is it allows higher level of brightness preservation to avoid some of the image texture. Therefore to improve trade-off
unwanted noises. between image contrast and resolution in context of preser-
vation of textural information and reduced blurring of sam-
ple the HE algorithms can be adopted.
2.5 Sharp Contrast Limited Adaptive Histogram
Equalization (SCLAHE) 4. METHODOLOGYS
Siti et al. [12], study on the identification of dental abnor- Iterative Dynamic HE technique In [2], proposed the itera-
malities in intra-oral dental radiographs. The abnormalities tive sub-histogram equalization method, which helps to en-
are Periapical Radiolucency (PA), widen periodontal liga- hance better contrast than any other brightness preserving
ment space (widen PDLs) and Loss of Lamina Dura (Loss of enhancing techniques.
LD). Image processing techniques such as Contrast enhance-
ment algorithms (CEAs) are accepted in dentistry in en- Dualistic Sub Image Histogram Equalization (DSIHE) Du-
hancing digital dental radiographs and assisting the dentists alistic sub-image histogram equalization (DSIHE) also sep-
during interpretation process. Thus it compares the perfor- arates the input histogram into two subsections. DSIHE
mance between original images and the images enhanced by chooses to separate the histogram based on gray level with
CEAs. The algorithms are Adaptive Histogram Equaliza- cumulative probability density .The result of the dualistic
tion (AHE), Contrast Limited Adaptive Histogram Equal- sub-image histogram equalization is obtained after the two
ization (CLAHE) and Sharp Contrast Limited Adaptive His- equalized sub images are composed into one image . Bright-
togram Equalization (SCLAHE). The researchers mentioned ness Bi-Histogram Equalization (BBHE) BBHE method is
that the future work will be conducted to focus on further used to decompose the original image into two sub-images,
investigation on the gray scale properties of the images that by using the image mean gray-level, and then apply the HE
contribute in detecting the abnormalities. method on each of the subimages. The BBHE method pro-
duces an output image with the value of brightness (mean
3. RELATED WORK gray-level) located in the middle of the mean of the input
Breast cancer is most common cancer in women. Due to the image. The BBHE technique is a hybrid method between
low contrast nature of mammographic images it is compli- mean brightness preserving histogram equalization methods
cated to identify signs such as bilateral asymmetry, archi- with clipped histogram equalization method. This technique
tectural distortion and masses etc [1]. Here Adaptive HE can enhance the images without producing unnecessary ar-
technique (AHE) is used to differentiate between benign or tifacts.
malignant tissues.AHE differs from the normal HE method
in the respect that HE generates only one histogram whereas By applying these techniques the problems explained in sec-
AHE method computes several histograms corresponding to tion 3 is overcome.
a distinct section of the image and uses that to redistribute
the intensity values of the image [9]. In [12], by using ADE 5. CONCLUSION
method can improve the detection of speculations on dense The main problem addressed by this paper is the enhance-
mammographic backgrounds. This is better enhancement ment of medical images using efficient algorithms based on
method for medical images and also incorporates histogram HE techniques. The proposed research involves analyzing
modifications as an optimization technique, but produces and formulating different HE image enhancement techniques
noise over amplification. For that reason have to extract suitable for various medical applications. More precisely,
more features of mass and classify to identify benign or ma- proposed research will focus on the enhancement of medical
lignant patients for early recognition of syndrome by using images captured under poor illumination conditions, foggy
other techniques. situations and speckle noise. Therefore design of algorithm
for histogram based medical image enhancement using the
In recent years ultrasound images used as imaging modalities techniques Iterative Dynamic HE, DSIHE, and BBHE can
to detect and diagnose human organ and tissue related disor- overcome the problems explained in section 3 which include
ders like to early detection of kidney disorders and diseases, the applications breast cancer, kidney diseases etc. Devel-
but due to the presence of speckle noise and low contrast in oping algorithms that would assist doctors to diagnose the
disease in the beginning stage only. For example the removal
of speckle noise and artifacts with segmenting kidney from
those images where kidney boundary are not much clear. For 10]Mengxu Chen, Junfeng Wang, Tianyi Zhan, ”Mammo-
improving the performance of image processing systems the grams contrast enhancement based on background homoge-
crucial solution is implementation of image processing tech- nization”, Shanghai JiaoTongUniversity,Shanghai, China 2013
niques in hardware. Also for improving the performance of IEEE.
image processing systems, for flexible design development,
more compact, low power, and high speed and to reduce cost
and time, gives the implementation of efficient histogram al- 11]Tanzila Rahman, Mohammad Shorif Uddin, ”Speckle Noise
gorithms on Field Programmable Gate Array (FPGA) with Reduction and Segmentation of Kidney Regions From Ultra-
pipelining and parallel processing. sound Image”, Jahangirnagar University 2013 IEEE.

6. REFERENCES 12] Etta D.Pisano, Shuquan Zong, Bradley M. Hemminger,


1]P.Spandana, Dr.KundaM.M.Rao, SMIEEE, Prof.B.V.V.S.N.Prabhakarrao,
Marla DeLuca, R.Eugene Johnston, Keith Muller, M. Pa-
Dr.Jwalasrikala,”Novel image processing techniques for early
tricia Braeuning, and Stephen M.Pizer, ”Contrast Limited
detection of breast cancer, mat lab and lab view implemen-
Adaptive Histogram equalization Image processing to Im-
tation”, 2013 IEEE Point-of-Care Healthcare Technologies
prove the detection of Simulated Spiculations in Dense Mam-
(PHT) Bangalore, India, 16 - 18 January, 2013
mograms”, Journal of Digital imaging,1998.

2] PingHsien, Chih-Tsung Shen, ”Contrast enhancement on


13]Nikhil J. Dhinagar and Mehmet Celenk, âĂIJUltrasound
color images using iterative histogram rearrangement”, 13th
medical image enhancement and segmentation using adap-
IEEE international symposium on consumer electronics.
tive homomorphic filtering and histogram thresholdingâĂİ,
2012 IEEE EMBS International Conference on Biomedical
Engineering and Sciences.
3] SaeidFazli, SevinSamadi, ParisaNadirkhanlou, ”A novel
retinal vessel segmentation based on local adaptive histogram
equalization”, 2013 8th Iranian Conference on Machine Vi-
14] Herwanto, AniatiMurniArymurthy, ”A system for com-
sion and Image Processing.
puter aided diagnosis of breast cancer based on mass analy-
sis”, 2013 International Conference on Robotics, Biomimet-
ics, Intelligent Computational Systems (ROBIONETICS)
4] Tarun Kumar Agarwal, Mayank Tiwari, Subir Singh Lamba,
Yogyakarta, Indonesia, November 25-27,2013.
”Modified histogram based contrast enhancement using ho-
momorphic filtering for medical images”, 2014 IEEE.
15]Kanchan Pandey, ”A comparative study of histogram equal-
ization techniques for image contrast enhancement”, IJE-
5] Senthilkumaran N, Thimmiaraja J, ”Histogram equaliza-
SAT [International Journal of Engineering Science Advanced
tion for image enhancement using mri brain images”, 2014
Technology], Volume-4, Issue-1, 023-027, 2014.
World Congress on Computing and Communication Tech-
nologies.
16]NungsanginlaLongkumer, Mukesh Kumar and RohiniSax-
ena, Contrast Enhancement Techniques using Histogram Equal-
6]V.KrishnaSree, P.Sudhakar Rao, ”Hardware implementa-
ization: A Survey, International Journal of Current Engi-
tion of enhancement of retinal fundus image using simulink”,
neering and Technology, Vol.4, No.3 (June 2014).
2013 IEEE Asia Pacific Conference on Postgraduate Re-
search in Microelectronics and Electronics.
17] NungsanginlaLongkumerÈe,, MukeshKumarÈe, and Ro-
7]Shibin Wu, Qingsong Zhu, and Yuhan YangYaoqinXie,
”Feature and contrast enhancement of mammographic im- hiniSaxenaÈe,
, ”Contrast Enhancement Techniques using His-
age based on multiscale analysis and morphology”, IEEE togram Equalization: A Survey”, International Journal of
International Conference on Information and Automation Current Engineering and Technology.
Yinchuan, China, August 2013.
18]LeilaMalihi, Karim Ansari-Asl, ”Malaria parasite detec-
8]S.S. CHONG, MEN k. S, ”Modified HL Contrast En- tion in giemsaâĂŞstained blood cell images”, 2013 8th Ira-
hancement Technique for Breast MR Images”, 2013 IEEE nian Conference on Machine Vision and Image Processing
International Conference on Signal and Irnage Processing (MVIP).
Applications (ICSIPA).

19]Yanling Zhang, Wenhao Zhou, Weirong Xu, Li Li, ”Lymph


9]Shelda Mohan, M. Ravishankar, ”Modified Contrast Lim- Node Image Segmentation Based on Fuzzy c-Means Cluster-
ited Adaptive Histogram Equalization Based on Local Con- ing and an Improved ChanâĂŞVese Model”, 2013 6th Inter-
trast Enhancement for Mammogram Images”, 2013. national Congress on Image and Signal Processing (CISP
2013). dari, ”Despeckle Filtering in Ultrasound Imaging”, Contem-
porary Engineering Sciences, Vol. 2, 2009, no. 1, 17 âĂŞ 36.

20]Herwanto, AniatiMurniArymurthy, ”A system for com-


puter aided diagnosis of breast cancer based on mass analy- 31]J. Ullom, M. Oelze and J.R. Sanchez, ”Speckle Reduc-
sis”, 2013 International Conference on Robotics, Biomimet- tion for Ultrasonic Imaging Using Frequency Compounding
ics, Intelligent Computational Systems (ROBIONETICS) and Despeckling Filters along with Coded Excitation and
Yogyakarta, Indonesia, November 25-27,2013. Pulse Compression”, Advances in Acoustics and Vibration,
Volume 2012, Hindawi Publishing Corporation.

21]Agung W. Setiawan, Tati R. Mengko, Oerip S. Santoso,


Andriyan B. Suksmono, ”Color retinal image enhancement 32]N.Otsu, ”A Threshold Selection Method from Gray-Level
using clahe”, Biomedical Engineering Research Group SEEI, Histograms”, IEEE Transactions on Systems, Man and Cy-
InstitutTeknologi BandungBandung, Indonesia. bernetics Vol, Smc-9, no. , January 1979.

22]AmitKamra, V K Jain (Member IEEE), ”Enhancement of 33]Zhang B1, Zhang L, Zhang L, Karray F.,”Retinal vessel
Subtle Signs in Mammograms Using Multiscale Morpholog- extraction by matched filter with first-order derivative of
ical Approach”, 2013 IEEE Point-of-Care Healthcare Tech- Gaussian”, ComputBiol Med. 2010 Apr;40(4):438-45. doi:
nologies (PHT) Bangalore, India, 16 - 18 January, 2013. 10.1016/j.compbiomed.2010.02.008. Epub 2010 Mar 3.

34] S. Chaudhuri, S. Chatterje e, N. Katz, M. Nelson, M.


23]VuThanh An and Nguyen Thanh Hai, ”Enhancement of Goldbaum, ”Detection of blood veessels in retinal images
CT Image Using Image Fusion”, The 2013 International Con- using two dimensional matched filters,” IEEE Trans. Med.
ference on Advanced Technologies for Communications (ATC’13). Imaging 8 (3)(1989),pp. 263âĂŞ269.

24]Shelda Mohan, M. Ravishankar, ”Modified Contrast Lim- 35]Jeong Hyun YOON , Yong Man RO ,” Enhancement of
ited Adaptive Histogram Equalization Based on Local Con- the Contrast in Mammographic Images Using the Homo-
trast Enhancement for Mammogram Images”, 2013. morphic Filter Method” IEICE TRANSACTIONS on Infor-
mation and Systems Vol.E85-D No.1 pp.298-303.

25] Etta D.Pisano, Shuquan Zong, Bradley M. Hemminger,


Marla DeLuca, R.Eugene Johnston, Keith Muller, M. Pa- 36] M. C Hanumantharaju, M. Ravishankar, D. R Ramesh-
tricia Braeuning, and Stephen M.Pizer, ”Contrast Limited babu, ”Design and FPGA Implementation of an 2D Gaus-
Adaptive Histogram equalization Image processing to Im- sian Surround Function with Reduced On-Chip Memory Uti-
prove the detection of Simulated Spiculations in Dense Mam- lization”, 2013 International Conference on Advances in Com-
mograms”, Journal of Digital imaging,1998. puting, Communications and Informatics (ICACCI).

26]Sundaram, M.Ramar, K.Arumugam, N.Prabin, G. , ” 37]Duan Zhu, Tian Hong, Sun Lei,”Research on Mammo-
Histogram based contrast enhancement for mammogram im- gram Enhancement”.
ages, in Signal Processing, Communication, Computing and
Networking Technologies (ICSCCN), 2011 International Con-
ference.

27]Wan M. Hafizah, EkoSupriyanto, ”Automatic Genera-


tion of Region of Interest for Kidney Ultrasound Images
Using Texture Analysis”, International Journal of Biology
and Biomedical Engineering, Issues 1, Volume 6, 2012.

28]EkoSupriyanto, Wan MahaniHafiza, Yeoh Jing Wui, Adee-


laArooj, ”Automatic Non Invasive Kidney Volume Measure-
ment Based On Ultrasound Image”, Recent Researches in
Computer Science, ISBN: 978-1-61804-019-0.

29] Gabor filter - wikipedia, the free encyclopedia

30] E. Nadernejad, M.R.Karami, S. Sharifzadeh and M. Hei-

You might also like