Associate Professor Charnchai Pluempitiwiriyawej, Ph.D.
รศ. ดร.ชาญชัย ปลื้มปิติวิริยะเวช
Education
- Ph.D. in Electrical and Computer Engineering, Carnegie Mellon University, USA.
- M.S. in Electrical and Computer Engineering, Carnegie Mellon University, USA.
- B.S. in Electrical Engineering (Summa Cum Laude), University of Maryland at College Park, USA.
Email: Charnchai.P@chula.ac.th
Research Interest
- Image Processing
- Medical Image Segmentation (CT Scan, MRI)
- 3D image Reconstruction and Modeling
- Face Recognition, White blood cell image classification
- Gait recognition.
Research Cluster
1.
Chaiwatanaphan, S; Pluempitiwiriyawej, C; Wangsiripitak, S
Printed Thai character recognition using shape classification in video sequence along a line Journal Article
In: Engineering Journal, vol. 21, no. 6 Special Issue, pp. 37-45, 2017, ISSN: 01258281, (cited By 1).
@article{Chaiwatanaphan2017,
title = {Printed Thai character recognition using shape classification in video sequence along a line},
author = {S Chaiwatanaphan and C Pluempitiwiriyawej and S Wangsiripitak},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85033604089&doi=10.4186%2fej.2017.21.6.37&partnerID=40&md5=8006cdcb2046578a098779d83e078714},
doi = {10.4186/ej.2017.21.6.37},
issn = {01258281},
year = {2017},
date = {2017-01-01},
journal = {Engineering Journal},
volume = {21},
number = {6 Special Issue},
pages = {37-45},
publisher = {Chulalongkorn University 1},
abstract = {This paper presents a novel method for recognition of 68 printed Thai characters in image sequences captured along a line of characters, based on their shape appearance such as the height and width, the top, bottom, and right edges, the numbers and positions of the circles (head of Thai characters) and the end points. Since each character appears in more than one frame of the image sequence that moves along the line, an algorithm to identify the arrangement of the characters in each line is necessary for accurate recognition results. We tested our system on image sequences with four different Thai fonts. The recognition rate is about 85.64% correct. © 2017, Chulalongkorn University 1. All Rights Reserved.},
note = {cited By 1},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
This paper presents a novel method for recognition of 68 printed Thai characters in image sequences captured along a line of characters, based on their shape appearance such as the height and width, the top, bottom, and right edges, the numbers and positions of the circles (head of Thai characters) and the end points. Since each character appears in more than one frame of the image sequence that moves along the line, an algorithm to identify the arrangement of the characters in each line is necessary for accurate recognition results. We tested our system on image sequences with four different Thai fonts. The recognition rate is about 85.64% correct. © 2017, Chulalongkorn University 1. All Rights Reserved.
2.
Butdee, C; Pluempitiwiriyawej, C; Tanpowpong, N
3D plane cuts and cubic Bézier curve for CT liver volume segmentation according to Couinaud’s classification Journal Article
In: Songklanakarin Journal of Science and Technology, vol. 39, no. 6, pp. 793-801, 2017, ISSN: 01253395, (cited By 2).
@article{Butdee2017,
title = {3D plane cuts and cubic Bézier curve for CT liver volume segmentation according to Couinaud’s classification},
author = {C Butdee and C Pluempitiwiriyawej and N Tanpowpong},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-85042237382&doi=10.14456%2fsjst-psu.2017.97&partnerID=40&md5=ee7bd8b9afbdca3d73502f3efa33bf3f},
doi = {10.14456/sjst-psu.2017.97},
issn = {01253395},
year = {2017},
date = {2017-01-01},
journal = {Songklanakarin Journal of Science and Technology},
volume = {39},
number = {6},
pages = {793-801},
publisher = {Prince of Songkla University},
abstract = {In pre-operative planning for partial liver transplantation, the total liver volume must be virtually segmented from a set of CT scanned images. The liver, consequently, is divided into eight segments according to Couinaud’s classification using hepatic and portal veins as clues. To facilitate the visualization of the segmented liver model, we propose a computerized process using four 3D plane cuts and cubic Bézier curve. In our experiments, fifteen liver volumes were used, and each of them was cut into eight segments using our program and their average percentage volumes were analyzed. The results were in agreement with the ground truth. Our program is semi-automatic. It requires minimal user interactions. As a result, the user can easily view the segmented liver model in both 2D and 3D perspectives. © 2017, Prince of Songkla University. All rights reserved.},
note = {cited By 2},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
In pre-operative planning for partial liver transplantation, the total liver volume must be virtually segmented from a set of CT scanned images. The liver, consequently, is divided into eight segments according to Couinaud’s classification using hepatic and portal veins as clues. To facilitate the visualization of the segmented liver model, we propose a computerized process using four 3D plane cuts and cubic Bézier curve. In our experiments, fifteen liver volumes were used, and each of them was cut into eight segments using our program and their average percentage volumes were analyzed. The results were in agreement with the ground truth. Our program is semi-automatic. It requires minimal user interactions. As a result, the user can easily view the segmented liver model in both 2D and 3D perspectives. © 2017, Prince of Songkla University. All rights reserved.
3.
Prinyakupt, J; Pluempitiwiriyawej, C
Segmentation of white blood cells and comparison of cell morphology by linear and naïve Bayes classifiers Journal Article
In: BioMedical Engineering Online, vol. 14, no. 1, 2015, ISSN: 1475925X, (cited By 75).
@article{Prinyakupt2015,
title = {Segmentation of white blood cells and comparison of cell morphology by linear and naïve Bayes classifiers},
author = {J Prinyakupt and C Pluempitiwiriyawej},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-84933531026&doi=10.1186%2fs12938-015-0037-1&partnerID=40&md5=c06267b65383801cf162fb3e67f4619c},
doi = {10.1186/s12938-015-0037-1},
issn = {1475925X},
year = {2015},
date = {2015-01-01},
journal = {BioMedical Engineering Online},
volume = {14},
number = {1},
publisher = {BioMed Central Ltd.},
abstract = {Background: Blood smear microscopic images are routinely investigated by haematologists to diagnose most blood diseases. However, the task is quite tedious and time consuming. An automatic detection and classification of white blood cells within such images can accelerate the process tremendously. In this paper we propose a system to locate white blood cells within microscopic blood smear images, segment them into nucleus and cytoplasm regions, extract suitable features and finally, classify them into five types: basophil, eosinophil, neutrophil, lymphocyte and monocyte. Dataset: Two sets of blood smear images were used in this study's experiments. Dataset 1, collected from Rangsit University, were normal peripheral blood slides under light microscope with 100× magnification; 555 images with 601 white blood cells were captured by a Nikon DS-Fi2 high-definition color camera and saved in JPG format of size 960 × 1,280 pixels at 15 pixels per 1 μm resolution. In dataset 2, 477 cropped white blood cell images were downloaded from CellaVision.com. They are in JPG format of size 360 × 363 pixels. The resolution is estimated to be 10 pixels per 1 μm. Methods: The proposed system comprises a pre-processing step, nucleus segmentation, cell segmentation, feature extraction, feature selection and classification. The main concept of the segmentation algorithm employed uses white blood cell's morphological properties and the calibrated size of a real cell relative to image resolution. The segmentation process combined thresholding, morphological operation and ellipse curve fitting. Consequently, several features were extracted from the segmented nucleus and cytoplasm regions. Prominent features were then chosen by a greedy search algorithm called sequential forward selection. Finally, with a set of selected prominent features, both linear and naïve Bayes classifiers were applied for performance comparison. This system was tested on normal peripheral blood smear slide images from two datasets. Results: Two sets of comparison were performed: segmentation and classification. The automatically segmented results were compared to the ones obtained manually by a haematologist. It was found that the proposed method is consistent and coherent in both datasets, with dice similarity of 98.9 and 91.6% for average segmented nucleus and cell regions, respectively. Furthermore, the overall correction rate in the classification phase is about 98 and 94% for linear and naïve Bayes models, respectively. Conclusions: The proposed system, based on normal white blood cell morphology and its characteristics, was applied to two different datasets. The results of the calibrated segmentation process on both datasets are fast, robust, efficient and coherent. Meanwhile, the classification of normal white blood cells into five types shows high sensitivity in both linear and naïve Bayes models, with slightly better results in the linear classifier. © 2015 Prinyakupt and Pluempitiwiriyawej.},
note = {cited By 75},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Background: Blood smear microscopic images are routinely investigated by haematologists to diagnose most blood diseases. However, the task is quite tedious and time consuming. An automatic detection and classification of white blood cells within such images can accelerate the process tremendously. In this paper we propose a system to locate white blood cells within microscopic blood smear images, segment them into nucleus and cytoplasm regions, extract suitable features and finally, classify them into five types: basophil, eosinophil, neutrophil, lymphocyte and monocyte. Dataset: Two sets of blood smear images were used in this study's experiments. Dataset 1, collected from Rangsit University, were normal peripheral blood slides under light microscope with 100× magnification; 555 images with 601 white blood cells were captured by a Nikon DS-Fi2 high-definition color camera and saved in JPG format of size 960 × 1,280 pixels at 15 pixels per 1 μm resolution. In dataset 2, 477 cropped white blood cell images were downloaded from CellaVision.com. They are in JPG format of size 360 × 363 pixels. The resolution is estimated to be 10 pixels per 1 μm. Methods: The proposed system comprises a pre-processing step, nucleus segmentation, cell segmentation, feature extraction, feature selection and classification. The main concept of the segmentation algorithm employed uses white blood cell's morphological properties and the calibrated size of a real cell relative to image resolution. The segmentation process combined thresholding, morphological operation and ellipse curve fitting. Consequently, several features were extracted from the segmented nucleus and cytoplasm regions. Prominent features were then chosen by a greedy search algorithm called sequential forward selection. Finally, with a set of selected prominent features, both linear and naïve Bayes classifiers were applied for performance comparison. This system was tested on normal peripheral blood smear slide images from two datasets. Results: Two sets of comparison were performed: segmentation and classification. The automatically segmented results were compared to the ones obtained manually by a haematologist. It was found that the proposed method is consistent and coherent in both datasets, with dice similarity of 98.9 and 91.6% for average segmented nucleus and cell regions, respectively. Furthermore, the overall correction rate in the classification phase is about 98 and 94% for linear and naïve Bayes models, respectively. Conclusions: The proposed system, based on normal white blood cell morphology and its characteristics, was applied to two different datasets. The results of the calibrated segmentation process on both datasets are fast, robust, efficient and coherent. Meanwhile, the classification of normal white blood cells into five types shows high sensitivity in both linear and naïve Bayes models, with slightly better results in the linear classifier. © 2015 Prinyakupt and Pluempitiwiriyawej.
4.
Meechai, T; Tepmongkol, S; Pluempitiwiriyawej, C
Partial-volume effect correction in positron emission tomography brain scan image using super-resolution image reconstruction Journal Article
In: British Journal of Radiology, vol. 88, no. 1046, 2015, ISSN: 00071285, (cited By 6).
@article{Meechai2015,
title = {Partial-volume effect correction in positron emission tomography brain scan image using super-resolution image reconstruction},
author = {T Meechai and S Tepmongkol and C Pluempitiwiriyawej},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-84921657237&doi=10.1259%2fbjr.20140119&partnerID=40&md5=fc4e195a05f92a922d21450e842901dd},
doi = {10.1259/bjr.20140119},
issn = {00071285},
year = {2015},
date = {2015-01-01},
journal = {British Journal of Radiology},
volume = {88},
number = {1046},
publisher = {British Institute of Radiology},
abstract = {Objective: The partial-volume effect (PVE) is a consequence of limited (i.e. finite) spatial resolution. PVE can lead to quantitative underestimation of activity concentrations in reconstructed images, which may result in misinterpretation of positron emission tomography (PET) scan images, especially in the brain. The PVE becomes significant when the dimensions of a source region are less than two to three times the full width at half maximum spatial resolution of the imaging system. In the present study, the ability of super-resolution (SR) image reconstruction to compensate for PVE in PET was characterized. Methods: The ability of SR image reconstruction technique to recover activity concentrations in small structures was evaluated by comparing images before and after image reconstruction in the NEMA/IEC phantom (Washington, DC), in the Hoffman brain phantom and in four human brain subjects (three normal subjects and one atrophic brain subject) in terms of apparent recovery coefficient (ARC) and percentage yield. Results: Both the ARC and percentage yield are improved after SR implementation in NEMA/IEC phantom and Hoffman brain phantom. When tested in normal subjects, SR implementation can improve the intensity and justify SR efficiency to correct PVE. Conclusion: SR algorithm can be used to effectively correct PVE in PET images. Advances in knowledge: The current research focused on brain PET scanning exclusively; future work will extend to whole-body imaging. © 2015 The Authors. Published by the British Institute of Radiology.},
note = {cited By 6},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Objective: The partial-volume effect (PVE) is a consequence of limited (i.e. finite) spatial resolution. PVE can lead to quantitative underestimation of activity concentrations in reconstructed images, which may result in misinterpretation of positron emission tomography (PET) scan images, especially in the brain. The PVE becomes significant when the dimensions of a source region are less than two to three times the full width at half maximum spatial resolution of the imaging system. In the present study, the ability of super-resolution (SR) image reconstruction to compensate for PVE in PET was characterized. Methods: The ability of SR image reconstruction technique to recover activity concentrations in small structures was evaluated by comparing images before and after image reconstruction in the NEMA/IEC phantom (Washington, DC), in the Hoffman brain phantom and in four human brain subjects (three normal subjects and one atrophic brain subject) in terms of apparent recovery coefficient (ARC) and percentage yield. Results: Both the ARC and percentage yield are improved after SR implementation in NEMA/IEC phantom and Hoffman brain phantom. When tested in normal subjects, SR implementation can improve the intensity and justify SR efficiency to correct PVE. Conclusion: SR algorithm can be used to effectively correct PVE in PET images. Advances in knowledge: The current research focused on brain PET scanning exclusively; future work will extend to whole-body imaging. © 2015 The Authors. Published by the British Institute of Radiology.
5.
Khamwan, K; Krisanachinda, A; Pluempitiwiriyawej, C
Automated tumour boundary delineation on 18F-FDG PET images using active contour coupled with shifted-optimal thresholding method Journal Article
In: Physics in Medicine and Biology, vol. 57, no. 19, pp. 5995-6005, 2012, ISSN: 00319155, (cited By 6).
@article{Khamwan2012,
title = {Automated tumour boundary delineation on 18F-FDG PET images using active contour coupled with shifted-optimal thresholding method},
author = {K Khamwan and A Krisanachinda and C Pluempitiwiriyawej},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-84866452627&doi=10.1088%2f0031-9155%2f57%2f19%2f5995&partnerID=40&md5=4ca7e43b44da0e2183b3855df7f722b4},
doi = {10.1088/0031-9155/57/19/5995},
issn = {00319155},
year = {2012},
date = {2012-01-01},
journal = {Physics in Medicine and Biology},
volume = {57},
number = {19},
pages = {5995-6005},
abstract = {This study presents an automatic method to trace the boundary of the tumour in positron emission tomography (PET) images. It has been discovered that Otsu's threshold value is biased when the within-class variances between the object and the background are significantly different. To solve the problem, a double-stage threshold search that minimizes the energy between the first Otsu's threshold and the maximum intensity value is introduced. Such shifted-optimal thresholding is embedded into a region-based active contour so that both algorithms are performed consecutively. The efficiency of the method is validated using six sphere inserts (0.52-26.53 cc volume) of the IEC/2001 torso phantom. Both spheres and phantom were filled with 18F solution with four source-to-background ratio (SBR) measurements of PET images. The results illustrate that the tumour volumes segmented by combined algorithm are of higher accuracy than the traditional active contour. The method had been clinically implemented in ten oesophageal cancer patients. The results are evaluated and compared with the manual tracing by an experienced radiation oncologist. The advantage of the algorithm is the reduced erroneous delineation that improves the precision and accuracy of PET tumour contouring. Moreover, the combined method is robust, independent of the SBR threshold-volume curves, and it does not require prior lesion size measurement. © 2012 Institute of Physics and Engineering in Medicine.},
note = {cited By 6},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
This study presents an automatic method to trace the boundary of the tumour in positron emission tomography (PET) images. It has been discovered that Otsu's threshold value is biased when the within-class variances between the object and the background are significantly different. To solve the problem, a double-stage threshold search that minimizes the energy between the first Otsu's threshold and the maximum intensity value is introduced. Such shifted-optimal thresholding is embedded into a region-based active contour so that both algorithms are performed consecutively. The efficiency of the method is validated using six sphere inserts (0.52-26.53 cc volume) of the IEC/2001 torso phantom. Both spheres and phantom were filled with 18F solution with four source-to-background ratio (SBR) measurements of PET images. The results illustrate that the tumour volumes segmented by combined algorithm are of higher accuracy than the traditional active contour. The method had been clinically implemented in ten oesophageal cancer patients. The results are evaluated and compared with the manual tracing by an experienced radiation oncologist. The advantage of the algorithm is the reduced erroneous delineation that improves the precision and accuracy of PET tumour contouring. Moreover, the combined method is robust, independent of the SBR threshold-volume curves, and it does not require prior lesion size measurement. © 2012 Institute of Physics and Engineering in Medicine.
6.
Phumeechanya, S; Pluempitiwiriyawej, C; Thongvigitmanee, S
Active contour using local regional information on extendable search lines (LRES) for image segmentation Journal Article
In: IEICE Transactions on Information and Systems, vol. E93-D, no. 6, pp. 1625-1635, 2010, ISSN: 09168532, (cited By 3).
@article{Phumeechanya2010a,
title = {Active contour using local regional information on extendable search lines (LRES) for image segmentation},
author = {S Phumeechanya and C Pluempitiwiriyawej and S Thongvigitmanee},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-77953006631&doi=10.1587%2ftransinf.E93.D.1625&partnerID=40&md5=6dd295b65edb05515b2846e48bd86b12},
doi = {10.1587/transinf.E93.D.1625},
issn = {09168532},
year = {2010},
date = {2010-01-01},
journal = {IEICE Transactions on Information and Systems},
volume = {E93-D},
number = {6},
pages = {1625-1635},
publisher = {Institute of Electronics, Information and Communication, Engineers, IEICE},
abstract = {In this paper, we propose a novel active contour method for image segmentation using a local regional information on extendable search line. We call it the LRES active contour. Our active contour uses the intensity values along a set of search lines that are perpendicular to the contour front. These search lines are used to inform the contour front toward which direction to move in order to find the object's boundary. Unlike other methods, none of these search lines have a predetermined length. Instead, their length increases gradually until a boundary of the object is found. We compare the performance of our LRES active contour to other existing active contours, both edge-based and region-based. The results show that our method provides more desirable segmentation outcomes, particularly on some images where other methods may fail. Not only is our method robust to noise and able to reach into a deep concave shape, it also has a large capture range and performs well in segmenting heterogeneous textured objects. Copyright © 2010 The Institute of Electronics, Information and Communication Engineers.},
note = {cited By 3},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
In this paper, we propose a novel active contour method for image segmentation using a local regional information on extendable search line. We call it the LRES active contour. Our active contour uses the intensity values along a set of search lines that are perpendicular to the contour front. These search lines are used to inform the contour front toward which direction to move in order to find the object's boundary. Unlike other methods, none of these search lines have a predetermined length. Instead, their length increases gradually until a boundary of the object is found. We compare the performance of our LRES active contour to other existing active contours, both edge-based and region-based. The results show that our method provides more desirable segmentation outcomes, particularly on some images where other methods may fail. Not only is our method robust to noise and able to reach into a deep concave shape, it also has a large capture range and performs well in segmenting heterogeneous textured objects. Copyright © 2010 The Institute of Electronics, Information and Communication Engineers.
7.
Pluempitiwiriyawej, C; Moura, J M F; Wu, Y -J L; Ho, C
STACS: New active contour scheme for cardiac MR image segmentation Journal Article
In: IEEE Transactions on Medical Imaging, vol. 24, no. 5, pp. 593-603, 2005, ISSN: 02780062, (cited By 181).
@article{Pluempitiwiriyawej2005a,
title = {STACS: New active contour scheme for cardiac MR image segmentation},
author = {C Pluempitiwiriyawej and J M F Moura and Y -J L Wu and C Ho},
url = {https://www.scopus.com/inward/record.uri?eid=2-s2.0-18844373951&doi=10.1109%2fTMI.2005.843740&partnerID=40&md5=190c41dfbfc44b0606fcaa604493f779},
doi = {10.1109/TMI.2005.843740},
issn = {02780062},
year = {2005},
date = {2005-01-01},
journal = {IEEE Transactions on Medical Imaging},
volume = {24},
number = {5},
pages = {593-603},
abstract = {The paper presents a novel stochastic active contour scheme (STACS) for automatic image segmentation designed to overcome some of the unique challenges in cardiac MR images such as problems with low contrast, papillary muscles, and turbulent blood flow. STACS minimizes an energy functional that combines stochastic region-based and edge-based information with shape priors of the heart and local properties of the contour. The minimization algorithm solves, by the level set method, the Euler-Lagrange equation that describes the contour evolution. STACS includes an annealing schedule that balances dynamically the weight of the different terms in the energy functional. Three particularly attractive features of STACS are: 1) ability to segment images with low texture contrast by modeling stochastically the image textures; 2) robustness to initial contour and noise because of the utilization of both edge and region-based information; 3) ability to segment the heart from the chest wall and the undesired papillary muscles due to inclusion of heart shape priors. Application of STACS to a set of 48 real cardiac MR images shows that it can successfully segment the heart from its surroundings such as the chest wall and the heart structures (the left and right ventricles and the epicardium.) We compare STAGS' automatically generated contours with manually-traced contours, or the "gold standard," using both area and edge similarity measures. This assessment demonstrates very good and consistent segmentation performance of STACS. © 2005 IEEE.},
note = {cited By 181},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
The paper presents a novel stochastic active contour scheme (STACS) for automatic image segmentation designed to overcome some of the unique challenges in cardiac MR images such as problems with low contrast, papillary muscles, and turbulent blood flow. STACS minimizes an energy functional that combines stochastic region-based and edge-based information with shape priors of the heart and local properties of the contour. The minimization algorithm solves, by the level set method, the Euler-Lagrange equation that describes the contour evolution. STACS includes an annealing schedule that balances dynamically the weight of the different terms in the energy functional. Three particularly attractive features of STACS are: 1) ability to segment images with low texture contrast by modeling stochastically the image textures; 2) robustness to initial contour and noise because of the utilization of both edge and region-based information; 3) ability to segment the heart from the chest wall and the undesired papillary muscles due to inclusion of heart shape priors. Application of STACS to a set of 48 real cardiac MR images shows that it can successfully segment the heart from its surroundings such as the chest wall and the heart structures (the left and right ventricles and the epicardium.) We compare STAGS' automatically generated contours with manually-traced contours, or the "gold standard," using both area and edge similarity measures. This assessment demonstrates very good and consistent segmentation performance of STACS. © 2005 IEEE.