You are on page 1of 32

VISVESVARAYA TECHNOLOGICAL UNIVERSITY

JNANA SANGAMA, BELGAVI - 590018

JSS ACADEMY OF TECHNICAL EDUCATION


JSS Campus, Uttarahalli - Kengeri Main Road, Bengaluru - 560060

DEPARTMENT OF ELECTRONICS AND COMMUNICATION

Project On
KIDNEY STONE DETECTION USING LEVEL SET
SEGMENTATION

Submitted by:

SI. No USN NAME

1. 1JS15EC002 Aarushi Saxena

2. 1JS15EC011 Anjali Krishna Singh


Under the guidance of:
3. 1JS15EC030 Garima
Dr. Veeramma Yatnalli
Assoc. Prof. ,Dept. of E&C,
JSSATE, Bengaluru-60
1
ABSTRACT

• The proposed work is used to detect the kidney stones by using Level Set
Segmentation.
• Computed Tomography(CT) scans are diagnostic tools used for many
purposes.
• Initially Input Images are pre-processed and the region of interest is
segmented.
• The pre-processing of the CT images is done.
• After pre-processing step the resulting image is segmented using Level Set
Segmentation Technique.
• Finally the segmented images are analyzed to detect the stone.

2
PROBLEM DEFINITION

Detection of Kidney Stone in the Computed Tomographic (CT) images


using Level Set Segmentation and few more algorithms, implemented
using MATLAB.

3
OUTLINE

 Introduction
 Objective
 Methodology
 Database Collection
 Pre-Processing
 Clustering
 Segmentation
 Result
 Conclusion
 References

4
INTRODUCTION

 Medical imaging is a technique that allows the physician to create a visual


representation interior organs.

 It reveals internal structure hidden by the skin and bones, as well as it diagnoses
and treat diseases

 The wide varieties of medical imaging modalities are available to treat and
diagnose the disease. Here, we have considered the CT imaging.

5
Kidney Stone

 The most common abnormality that occur in human urinary


system is kidney stone, also known as Renal Calculi.

 Stone is a pebble like structure formed by the crystal


aggregation by the dietary mineral components in the kidney.

 The stone diseases remain unnoticed in the initial stage,


which in turn damages the kidney as they develop.

6
OBJECTIVES

 Collection of images
 The pre-processing of data images
 To find the most effective Image segmentation method
 To select suitable methodology and algorithms to detect the presence of
stone

7
LITERATURE SURVEY
Author Akkasaligar PT, Biradar S, Kumbar V
Title of the Paper Kidney stone detection in computed tomography
images
Year of Publication In Smart Technologies For Smart Nation
(SmartTechCon), 2017 International Conference On
2017 Aug 17 (pp. 353-356). IEEE.

Paper Theme • The research paper mentioned is used to detect the


kidney stones by using Level set segmentation
method.
• The input images are pre-processed and region of
interest is segmented using level set segmentation.
• The segmented images are analyzed to detect the
presence of stone.

8
Author K.Viswanath, R. Gunasundari

Title of the Paper Design and analysis performance of Kidney Stone


Detection from Ultrasound Image by Level Set
Segmentation and ANN Classification
Year of Publication International Conference on Advances in Computing,
Communication and Informatics (ICACCI) Vol. 5, 2014

Paper Theme In this paper the authors have proposed kidney stone
detection using semi-automatic segmentation approach

Author Saman Ebrahimi and Vladimir Y. Mariano

Title of the Paper Image quality improvement in kidney stone detection on


computed tomography images
Year of Publication Journal of Image and Graphics,Vol. 3, no. 1,June 2015

Paper Theme It uses image processing techniques and geometry


principles to define the boundary which enhances
kidney stone detection.
9
Author Dr P.R. Tamilselvi and Dr.P. Thangaraj

Title of the Paper Computer Aided Diagnosis System for Stone


Detection and Early Detection of Kidney
Stones

Year of Publication Journal of Computer Science.Vol 5, no.


3,June 2015

Paper Theme It uses image diagnosis for stone and its early
detection based on improved seeded region
growing based segmentation and
classification of kidney images with stone
sizes.

10
Database Collection
 First step involves the collection of database of the kidneys.

 Ultrasound, MRI, computed tomography (CT) are different medical imaging


modalities used in detecting and diagnosing a disease.

 Ultrasound imaging uses high-frequency sound waves to view inside the body.

It can also show movement of the body's internal organs as well as blood
flowing through the blood vessels.

 Magnetic resonance imaging (MRI) uses strong magnetic fields and radio waves
to produce detailed images of the inside of the body.

 Computed Tomography(CT) scans are a specialized type of x-ray that uses a


combination of X-rays and a computer to create pictures of the organs, bones, and
other tissues.

11
 In our method it involves the collection of Computed Tomographic (CT) images of
the kidney in the database.

 Compared to medical imaging modalities such as ultrasound, MRI, etc.,


Computed Tomography (CT) imaging is the best way to detect and diagnose the
abnormalities of Kidney.

 The Computed Tomography of Abdomen gives a maximum detail about the


anatomy of kidney when compared to other modalities.

CT Scan image of kidney

12
Pre-Processing

 In order to improve the visual effects of the image for further image
recognition and for medical imaging purpose.

 CT image pre-processing is needed, mainly including colour image


greyscale, image smoothing and sharpening and so on.

 Image smoothing is to eliminate noise and improve image quality.

 Initially the CT images are converted into grey scale image.

 Further Grey scale images are converted into double.

 Next Dilation is performed on the image obtained.

 Image obtained is resized into 256*256 pixels

13
.

Fig2: Cropped
Fig1: Input image image

14
Clustering

 Process of partitioning or grouping a given sector into a number of clusters


such that similar patterns are assigned to a group, which is considered as a
cluster.

 There are two main approaches to clustering which are hard clustering and
soft clustering techniques.

 Fuzzy C means is a very important clustering technique based on fuzzy


logic also called Soft Clustering Technique.

 Also we have some hard clustering techniques available like K-means


among the popular ones.

15
Fuzzy C-means clustering:

 In case of soft clustering techniques, fuzzy sets are used to cluster data, so
that each point may belong to two or more clusters with different degrees of
membership

 Fuzzy c-means clustering is used to detect the region of interest.

 Algorithm works by assigning membership to each data point


corresponding to each cluster center on the basis of the distance between
the data point and the cluster center.

 Nearer the data to the cluster center, more is the membership towards the
particular center.

16
Algorithm
1 Set value for c, q and 𝜖.
2 initialize the fuzzy partition matrix 𝑈 = [𝑢𝑖𝑘 ]
3 Set the loop counter b = 0
𝑏
4 Calculate the c cluster centers {𝑣𝑖 }
with 𝑈 (𝑏) .
𝑏
(𝑏) σ𝑛
𝑘=1 𝑢𝑖𝑘 𝑥𝑘
𝑣𝑖 = (𝑏) ………..(1)
σ𝑛
𝑘=1(𝑢𝑖𝑘 )

5 Calculate the membership 𝑈 (𝑏+1) .For k=1 to n, compute the new membership value:

(a) If 𝐼𝑘 = ∅, then
(𝑏+1) 1
𝑢𝑖𝑘 = 2 …………(2)
𝑐 𝑑𝑖𝑘 (𝑞+1)
σ𝑗=1( )
𝑑𝑗𝑘
(𝑏+1) (𝑏+1)
(b) Else 𝑢𝑖𝑘 =0 for all 𝑖 ∈ 𝐼 and σ𝑖∈𝐼𝑘 𝑢𝑖𝑘 = 1; next k.

6 If ||𝑈 (𝑏) − 𝑈 (𝑏+1) || <є, stop; otherwise, b=b+1 and go to step 4

otherwise,
go to step 3.
17
(1) set the number of clusters, fuzzy parameter (a constant> 1),
And the stopping condition

(2) initialize the fuzzy partition matrix

(3) Set the loop counter k=0

(4) Calculate the cluster centroids, calculate the


objective value J

(5) For each object, for each cluster, compute the membership values in
the matrix

(6) If the value of J between consecutive iterations is less than the stopping
condition, then stop; otherwise, set k=k+1 and goto step 4

(7) Defuzzification and segmentation 18


Advantages

 It gives the flexibility to express that data points can belong to more than
one cluster.
 FCM gives best result for overlapped data set and comparatively better
then k-means algorithm.

Applications
 Clustering algorithms can be applied in many fields, for instance:
 Marketing: finding groups of customers with similar behavior given a large
database of customer data containing their properties and past buying
records;
 Biology: classification of plants and animals given their features;
 Libraries: book ordering;

19
Fig: FCM
Output where
white sections
are the region of
interest.

20
Extraction of Last Cluster
From the clusters, single cluster image is chosen depending on the area of renal stone
presence.

Here Fig 7. is the Extraction of Last Cluster while Fig 8. is the Masked output of this
image which enhances the region of interest.
21
Segmentation

 Image Segmentation is used to locate objects and boundaries in an image.

 It assigns label to every pixel in image such that pixels with same label have same
characteristics.

 Many of the PDEs used in image processing are based on moving curves and surfaces
with curvature-based velocities.

 In this area, the level set method is very influential and useful.

22
Level Set Segmentation
 Level Set Method is one of the emerging image segmentation techniques for
medical image segmentation.

 The level set method is a numerical technique for tracking interfaces and shapes.

 The basic idea is to represent the curves or surfaces as the zero level set of a higher

dimension, referred to as the level set function

 This technique not only provides more accurate numerical implementations but also
handle topological change very easily.

23
Cntd..

 The standard level set function is defined as


𝜕∅
+ 𝐹|𝛻∅| = 0…………(1)
𝜕𝑡

∅ 0, 𝑥, 𝑦 = ∅0 (𝑥, 𝑦)…………(2)

 ∅0 (𝑥, 𝑦)is defined as


−𝐶0 𝑖𝑓 𝑥, 𝑦 𝑖𝑠 𝑖𝑛𝑠𝑖𝑑𝑒 ∅0
 ∅0 𝑥, 𝑦 = ቐ 𝐶0 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒

24
Advantages

 Versatile, robust, and efficient techniques for a wide class of problems


in etching, deposition and photolithography development.
 Relatively fine resolution achieved.

Applications

 Combustion
 Trajectory planning
 Optimization
 Image processing

Fig 9. Operation of Level Set Segmentation

25
Detection
 In MATLAB we have a function “regionprops“ which is used to measure the
properties of objects in the image.

 The specific properties that are important in the object for our concern are the area
and also the centroid of the object.

 Next, based on the threshold value we select the smallest object and tag it’s centroid
with red marks

Fig 11. Superimposed stone on


Fig 10. Detection of Stone initial image
26
RESULTS

Abnormal kidney with stone detected 27


Normal kidney where stone not detected 28
CONCLUSION

 Image segmentation plays a dynamic role in the


field of medical image analysis.
 Fuzzy C means clustering and Level set are two
obvious methods used in Image segmentation.
 In this paper, both of these methods are
integrated and used to segment renal stones from
CT abdomen images.
 This proposed method segments the renal stone
from the CT abdomen images.
 The method is efficient and the time taken to
process the image is low.

29
FUTURE SCOPE

 Early detection of the abnormalities in the kidney is very


necessary to avoid serious damage.

 The region of the most probable occurrence for left and


right kidney in the abdominal , well aligned CT datasets
where segmentation is performed.

 The developed system is examined for different kidney


images from the database and results are effective in
classifying the stone presence successfully.

30
REFERENCES

 Prema T. Akkasaligar,Sunanda Biradar and Veena Kumbar,”Kidney stone


detection in computed tomography images” Published in: 2017
International Conference On Smart Technologies For Smart Nation
(SmartTechCon).

 Saman Ebrahimi and Vladimir Y. Mariano”Image quality improvement in


kidney stone detection on computed tomography images”,published in
Journal of Image and Graphics,Vol. 3, no. 1,June 2015.

 K. Viswanath and R. Gunasundari, “Kidney stone detection from


ultrasound images by Level Set Segmentation and ANN classification,” in
Proceedings of the International Conference on Communication and
Computing.

31
 T.anzila Rahman, Mohammad Shorif Uddin, "Speckle Noise Reduction and
Segmentation of Kidney Regions from Ultrasound Image", 978-14799-
0400-6113, 2013 IEEE.

 Wan Mahani Hafizah, "Feature Extraction of Kidney Ultrasound Images


based on Intensity Histogram and Gray Level Co-occurrence Matrix" 2012
sixth Asia Modeling Symposium, 978-0-7695-4730-5112, 2012 IEEE.

 V. P. Gladis Pushpa Rathi, "Detection and Characterization of Brain Tumor


Using Segmentation based on HSOM, Wavelet packet feature spaces and
ANN", 978-1-4244-8679- 3111, 2011 IEEE.

32

You might also like