Professional Documents
Culture Documents
ICSEC2020
ICSEC2020
ICSEC2020
Bangkok, Thailand
3 - 4 December 2020
Introduction
Related Works
Problem Statement
Objectives
Proposed Method
Experiment Results
Result Analysis
Conclusion
2
Introduction
Kidney Stones are small, hard deposits of mineral and acid salts on the inner surfaces
of the kidneys (drops along the urinary track).
http://www.pilotfriend.com/aeromed/medical/images2/25.jpg
3
Cont’d
10.0%
8.8%
9.0%
8.0%
7.0%
6.0% 5.2%
5.0%
4.0%
3.2%
3.0%
United States
2.0%
1.0%
0.0%
Fig: Prevalence of kidney stone disease from 1976 to 2010 [1] Fig: Recurrent rates of kidney stone forming after treatment [1]
4
Cont’d
5
Related Works
Related Research Advantages Disadvantages
Feature Extraction of Kidney Ultrasound delete and remove speedily the depend on the skill of
Images based on Intensity Histogram and complicated background because of human experts
Gray Level Co-occurrence Matrix [3] manual cropping of the region of
interest
Kidney stone detection in Computed detect the size and location of the remain some false-positive
Tomography Images [4] stone due to using manual noise in the image
removal to crop the kidney regions in
CT images based on X and Y direction
of the matrix
Image Quality Improvement in Kidney Stone detect and mark kidney stone with require a qualified medical
Detection on Computed Tomography 84.61% accuracy by using thresholding professional to validate its
Images [5] based on the optimum value of output result
contrast adjustment
Automated False Positive Reduction and reduce false positive with 70 % of less robust than other
Feature Extraction of Kidney Stone Object in overall accuracy by evaluating the two previous researches
3D CT Images [6] thresholding values based on volume
and surface-area of kidney stones
6
Problem Statement
7
Objectives
8
Proposed Method
Three stages:
Hypodense and isodense region removing
Intensity-based thresholding method
Unneeded hyperdense region removing
Size-based thresholding method
False-positive reducing
Volume and average HU values -based thresholding method
9
Intensity-based thresholding method
In step 1, Two constant threshold values, and (200 HU and 2800 HU values )
Step 2, foreground and background are segmented into two groups of pixels and ,
using and :
(1)
Step 3, finally, the new image is outputted with, after eliminating group.
10
Size-based thresholding method
Step 1, Calculate the size (longest diameter )of each object through the bounding
box calculation in [7].
Step 2, Two constant threshold values, and (3mm and 50 mm) [8].
Step 3, Foreground and background are segmented into two groups of pixels and
without using, and
(2)
Step 4, Finally, the new image is outputted with , after elimination group.
11
Volume and average HU values - based thresholding method
Step 1, Calculate the required threshold values ( and ) using following equation, and
regarded with 0.03 and 35 as consideration of the relationship between volume and HU of
kidney stone.
is the average HU for each object, n is the number of voxels in an object. µ is the HU value of each voxel in an object.
Step 2, Segment the image into two groups of pixels and without using, and
(6)
Step 3, Develop the new image with , after elimination group.
12
Experiment Results
13
Experiment Results
Result of after hyperdense region removing Result of after false positive removing
14
Results Analysis 1
After loading 500-600 slices of each patient, it is constructed the original image which consists of
over 500000 labels.
In the first process, the algorithm can remove all regions having hypodense and isodense
structure from the abdomen CT image and reduce the unwanted objects up to 4300 labels on
average from over 500000 labels.
In the second process, hyperdense region removing, the algorithm removes the bony skeleton,
bed mat, and some bones but the average of the remaining labels is about 112.
The final output is clear in visualization and remains only an average of 21 labels in an image.
So, it can remove most of unwanted objects.
Hypodense and isodense Hyperdense region False-positive reducing
region removing removing
4300 labels 112 labels 21 labels
15
Results Analysis 2
where, TP and FN are used for kidney stones which are correctly detected and not detected in system
TP FN Sensitivity
66 3 95.7%
Could carry out the unneeded regions removing with a significant stone detecting.
16
Conclusion
gives good support in kidney stone detection and classification, and correctly
remove the unneeded region
is simple and easy to understand for kidney stone classification because of using
thresholding methods based on the nature of kidney stones on CT
good and satisfactory result
95.7% in sensitivity
is effective and efficient with good sensitivity
17
References
[1] A. H. Milladipour and M. Rezaei Hemami, “Renal Function Assessment in Adults with Recurrent Calcium Kidney Stone Disease,”
J. Nephrol. Ther., vol. 2, no. 4, pp. 2–4, 2012.
[2] “SUPPORT LIFE EXPECTANCY RESEARCH!,” WORLD HEALTH ORGANIZATION 2017, 2017. [Online]. Available:
http://www.worldlifeexpectancy.com/myanmar-kidney-disease.
[3] W. M. Hafizah, E. Supriyanto, and J. Yunus, “Feature Extraction of Kidney Ultrasound Images based on Intensity Histogram and
Gray Level Co-occurrence Matrix,” pp. 115–120, 2012.
[4] P. Akkasaligar, S. Biradar, and V. Kumber, “Kidney stone detection in Computed Tomorgrahy Images,” in 2017 International
Conference On Smart Technologies For Smart Nation (SmartTechCon), 2017, vol. 6, no. 4, pp. 353–356.
[5] S. Ebrahimi and V. Y. Mariano, “Image Quality Improvement in Kidney Stone Detection on Computed Tomography Images,” J.
Image Graph., vol. 3, no. 1, pp. 40–46, 2015.
[6] N. Thein, “Automated False Positive Reduction and Feature Extraction of Kidney Stone Object in 3D CT Images,” Int. J. Intell.
Eng. Syst., vol. 12, no. 2, pp. 62–73, 2019.
[7] M. Vallières, C. R. Freeman, S. R. Skamene, and I. El Naqa, “A radiomics model from joint FDG-PET and MRI texture features for
the prediction of lung metastases in soft-tissue sarcomas of the extremities,” Phys. Med. Biol., vol. 60, no. 14, pp. 5471–5496,
Jul. 2015.
[8] C. D. Scales, A. C. Smith, J. M. Hanley, and C. S. Saigal, “Kidney Stones in Adults,” NIH Publ., vol. 13, no. 2495, pp. 160–165,
2012.
18
19