Table of Contents
Abstract
In this work we are trying to implement the segmentation of an x-ray image for cervical spondylosis detection. One of the major concerns of that particular disease is faster detection and identification of diseases in pervious stages. According to the opinion of clinical experts, today x-ray tomography method is most effective technology in medical science domain for easily diagnosis of particular cervical diseases. Segmentation is a kind of approach that is used to identify the unambiguous region from the particular x-ray image.
Today the diagnosis of cervical spondylosis is become one of the challenging work. MRI and CT scans used by a doctor for manual inspection which method are already available, so our proposed method which is automatically detect and analysis the cervical spondylosis using morphological segmentation and edge detection and classification based approach. The results of this study gaining more than 90% accuracy, sensitivity, for identifying and classifying the cervical diseases in x-ray images more accurately. Here the experimental performance shows better PSNR and MSE values for image quality measurement of detection of cervical spondylosis more accurately.
Introduction
Cervical Osteoarthritis in neck also known as Cervical Spondylosis. The “wear and tear” of the tendon and bones found in our Cervical Spondylosis. It is very much ex-tent to old age people and caused by different factors also.
127 our experts are availableright now
[bookmark: page2]In a large number of ways the cervical spondylosis can express itself. It can causes some neck and neurological problems of the sphincters and middle or the furthest points if there is spinal cord contribution[1]. Some other factors than maturing can build your of Cervical Spondylosis. Medical image processing is upcoming area of research for the advancement of new approaches and software releases for analysis. It plays a vital role in disease diagnosis and enhanced in medical area with patient care and helps medical practitioners to make decision in regard to the type of treatment.
Pankaj Pramod Chitte, Dr. U M. Gokhale[1] analyses that, “Different Methods for Identification and Classification of Cervical Spondylosis”. CNN Long-Short Team memory logic has been ended up being a powerful tool planned for course of action estimate issues with spatial data sources, like pictures or chronicles. Xinghu Yu, Liangbi Xiang[2] analyses that, In the Chinese analysis, clinical classification of Cervical Spondylosis mostly include some approaches using Fuzzy Calculation. They used Fuzzy logic to classifying Cervical Spondylosis as a part of some medical expart systems[1]. J.Anish Jafrin Thilak, Dr.P.Suresh, N.Subramani, S.Sathishkumar, V.V. Arunsankar [3] “Analysis of Human Neck Image Using Pro/E and MATLAB”. They take some approach of Modeling and Analysis of FSU. Diya Chudasama, Tanvi Patel, Shubham Joshi, Ghanshyam I. Prajapati[4] “Image Segmentation using Morphological Operations”. They use some morphological and edge detection method on some image for segmentation process.
Background Study
Degenerative cervical spondylosis is the very much familiar, moreover asymptomatic situation, happened due to age and some deteriorating changes in the cervical part of the body. Different clinical diseases like axial neck discomfort, cervical irritation etc. can be classified by the cervical spondylosis. A patient of that particular disease mostly having a mixed combination of axial neck discomfort, cervical irritation and cervical compression.
Scope Of Work
In general traditional diagnosis method, the symptoms are often seen that above situation for taking final decision and some necessary action to be taken. Careful clinical examination is required for proper diagnosis of diseases, also a thorough inter-view of the patient and relatives, X-ray, CT, and so forth. However, X-ray is repeatedly used as a complimentary diagnostic instrument in addition to the clinical discovery in the basic-level hospitals.
Due to the fact that CS has different classification, different types of CS need to be treated with different treatments and different types of CS have to be discriminated from the nominal disease specific changes in the X-ray images. The X-ray images nominal changes make diagnosis visually a critical job so it needs experienced personnel to solve. Also with it, the problem is unsolved and the potential of diagnosis of the disease has not been explored in this area. So our pro-posed methodology of segmentation has certain practical significance.
Types Of Disease
Different types stages of diseases regarding cervical spondylosis are-
- Postural Neck Pain
It occurs due to postural habits for long period of time. When the head alleged onward in poor posture, the spine of the cervical part must maintain the increasing quantity of weight. As a result it gives additional pressure on ligaments and muscles and causes the postural neck pain.
- Acute Neck Pain
It occurs due to unknown cause in a sudden advancement of the neck realizes extraordinary neck acknowledges uncommon neck torment. There are many cause behind the acute neck pain. One may feel toughness and locked condition on neck.
- Cervical Spondylosis
It occurs due to ‘wear and tear’ of Cervical Vertebrae and the intervertebral disease. In a large number of ways the cervical spondylosis can express itself. It is mostly common and degrades with oldness.
- Neck Osteitis
It is aggravation of bone. In general term Osteitisa is irritation of bone. This type of neck problem mostly depend on individuals bone structure or the skeleton structure and some other factors of body.
Proposed Methodologies
Here, the technique used different steps as like as-
Read Color CT Scan Image
Here some color CT scan images are used for this experimental study and dis-play the image in the very first part of our segmentation process. Basically these images consist of three color components as Red, Green and Blue.
Image Pre-Processing
Beginning step is the conversion of the color image into grayscale image because processing time taken by color image is more than grayscale image. The next step is resizing the image to a particular size. So requirement of preprocessing is to prepare better image quality more and make results of image. Then change the image with its corresponding gray scale image and show the image in our second step of work.
Normalization Of Image
We normalized the image by using some factors and command in our third step.
Filtering Of Image
Filtering is basically used for removing noise from the image. Different types of filtering techniques are used for reducing the noise for gaining better quality image for working. Then we filter the image using some filtering method in our fourth step of methodology.
Edge Detected Image
The use of canny edge detector to show the edge for our image in the fifth step of progress and found that canny edge detector is very useful for this purpose. It is an Edge detection operator to detect an extensive range of Edges in image. It uses a multi stage algorithm. It almost detects every Edge in image. Canny algorithm can identify Edges with noise suppressed at the same time.
Median Filtered Image
Here median filtering is used for removing the rest of the noise present in images. It is a nonlinear digital filtering method and used to decrease noise at the pre-processing stage for improving the result of processing.
Morphological Operations
Here some morphological operations are used to obtain the morphological image in this work. Erosion is best outfit as one of the simple operators in morphological operation. It can cause the elements to become lesser in size. Erosion simply corrodes away the borders of the foreground which outcomes in areas of those pixels lesser in size and holes of those areas turned out into larger[4].
Dilation also best as the simple operator in morphological operation[4]. It is basically used on binary image but can be used on grey scale image. It causes the elements to rise in size. The outcome of this operation will progressively increase the borders of the foreground pixels, thus areas raise in size and holes in that part turned out lesser.
Results And Analysis
[bookmark: page6]Here the results are shown in given images that the results are showing the intermediate steps of segmentation methodology whatever used for experimental study. The experimental result shows the region of diseases or changes in the disease area of the particular X-Ray images
The accuracy is measured with confusion matrix for supervised classification of data. The Sensitivity, Specificity, Accuracy parameter shows better classification accuracy with supervised learning methods with more than 90% of accuracy rates. Here the Accuracy of second image is 100% for better accuracy rates more than other images.
The Segmented Cervical image is required for statistical measurement of different parameter then confusion matrix parameters are used for measuring classification accuracy. Here, the early detection of the disease is done properly. The early detection is very important for clinical experts for taking decision about the cases of disease.
Measure Of Image Quality Parameter
In the experimental phase we have used the parameter PSNR (Peak Signal to Noise Ratio) for calculating the difference between the cover image and segmented image. The PSNR for an image of size NxN is given as follows:
PSNR = 10 log10 (255^2 / MSE) (dB) (1)
Where MSE= (1/N*N) ΣΣ (xij – x’ij) 2,
Similarity Index Calculation
The Structural Similarity Index (SSIM) is a perceptual metric that calculates image quality deprivation produced by handling such as data compression or by sufferers in data conduction. It is a complete reference metric that needs two images from the same image detention— a reference image and a handled image.
Our requirement is to get the higher PSNR value and lower MSE value for better quality performance for these cervical images. Here we got PSNR value of range 7 to 9.5 and MSE value of range 1 to 8.7 for detection of cervical spondylosis images.
Here, in the second case PSNR value is higher and MSE value is lower. The SSIM value is higher so it will produce best performance rather than other cases for cervical spondylosis detection.
Conclusion
[bookmark: page8]In this particular study, the methods for segmentation of different kinds of Cervical Spondylosis diseases related images have discussed. We have also discussed the basic concept of Cervical Spondylosis disease, type of Cervical Spondylosis related diseases and some algorithms related to the problem according to some Literature Review and also Proposed our Methodology using edge detection and morphological segmentation.
But from the above review paper we can conclude that some papers are showing partial automated diagnosis system, some authors tried to increase efficiency of the images by modifying the segmentation techniques. According to study of the problem domain, we will try to implement our proposed methodology of CERVICAL SPONDYLOSIS to the system using edge detection and morphological segmentation-that will be more effective and also the accuracy level of the particular set of data will be higher.
The used algorithm takes higher than 90% accuracy rates for classification, in terms of better segmentation and classification of cervical diseases. Here, in particular one case of cervical spondylosis detection is done with higher accuracy and yields the better image quality performance than other.
References
- Pankaj PramodChitte, Dr. U M Gokhale “Analysis of Different Methods for Identification and Classification of Cervical Spondylosis (CS): A Survey” International Journal of Application Engineering Research ISSN 0973-4562 volume 12, Number 21 (2017) pp. 11727-11737.
- J.AnishJafrin Thilak, Dr.P.Suresh, N.Subramani, S.Sathishkumar, V.V.Arunsankar “Analysis of Human Neck Image Using Pro/E and MATLAB” International Journal of Innovative Research in Science, Engineering and TechnologyVol.6,DOI:10.15680/IJIRSET.2017.0602159.
- Kudva, V., Prasad, K., &Guruvare, S. (2017). Detection of Specular Reflectionand Segmentation of Cervix Region in Uterine Cervix Images for Cervical Cancer Screening. IRBM, 1, 3–13. https://doi.org/10.1016/j.irbm.2017.08.003
- M. Anousouya Devi, S. Ravi and J. Vaishnavi S. Punitha Detection of Cervical Cancer using the Image Classification Algorithms, I J C T A, 9(3), 2016, pp.
- Diya Chudasama, Tanvi Patel, Shubham Joshi,Ghanshyam I. Prajapati “Image Segmentation using Morphological Operations”International Journal of Computer Applications (0975 – 8887) Volume 117 – No. 18, May 2015.
- Beyrem Jebri et al, “Detection of Degenerative Change in Lateral Projection Cervical Spine X-ray Images”, 2015.
- Xinghu Yu, Liangbi Xiang “Classifying Cervical Spondylosis Based on Fuzzy Calculation” volume 2014, pp 1-7, 2014.
- Er. Komal Sharma, Er. Navneet Kaur, “Comparative Analysis of Various Edge Detection Techniques”, International Journal of Advanced Research in Computer Science and Software Engineering, Volume 3, Issue 12, December 2013.
- T. Gajpal, Mr. S. Meshram “Edge Detection Technique Using Hybrid Fuzzy logic Method”, IJERT International Journal of Engineering Research & Technology, Vol. 2 Issue 2, Febuary-2013.
- Minavathi, Murali.S, M.S.Dinesh, “Classification of Mass in Breast Ultrasound Images using Image Processing Techniques”, Mysore University, India. IJOCA, vol.42.no.10, 2012.
- Kelly JC, Groarke PJ, Butler JS, Poynton AR, O’Byrne JM. The natural history and clinical syndromes of degenerative cervical spondylosis. Advances in Orthopedics. 2012; 2012.
- Christopher M. Bono et al, “Diagnosis and Treatment of Cervical Radiculopathy from Degenerative Disorders,” North American Spine Society Evidence Based Clinical Guidelines for Multidisciplinary Spine Care, 2010.
- Rao RD, Currier BL, Albert TJ. Degenerative cervical spondylosis: Clinical syndrome, pathogenesis and management. J Bone joint Surg. 2007;89(6):1360-78
- M Hochman, S Tuli, “Cervical Spondylotic Myelopathy: A Review,” The Internet Journal of Neurology, vol. 4, no. 1, 2004.
- https://www.google.com/search?q=cervical+spondylosis+images+dataset&tbm=isch&tbo=u&source=univ&sa=X&ved=2ahUKEwijgbfh3PvfAhUBfn0KHXwJDKgQsAR6BAgEEAE&biw=1242&bih=553
