TY - GEN
T1 - Speckle detection in ultrasonic images using unsupervised clustering techniques
AU - Azar, Arezou Akbarian
AU - Rivaz, Hasan
AU - Boctor, Emad
PY - 2011/12/26
Y1 - 2011/12/26
N2 - In ultrasonic images, identification of speckled regions helps to estimate probe movement as well as improve performance of algorithms for adaptive speckle suppression and the elevational separation of B-scans by speckle decorrelation. By tracking FDS patch displacements over time we can calculate strain and detect tumor location. Previous studies for speckle detection were based on classification techniques which estimated parameters of the statistical distribution which were based on observation data and ultrasound echo envelope signal. However, in this study, we proposed a new combination of statistical features which were extracted from the ultrasound images and explored their properties for the speckle detection. These features were used as inputs to the unsupervised clustering algorithms for the speckle classification. We used five different types of unsupervised techniques and compared their performance by feeding different combinations of the statistical features. In order to quantitatively compare statistical features and classification methods, as ground truth, we used simulations of cyst and fetus ultrasound images which were generated using Field II ultrasound simulation program[1]. Initial results showed that by combining two statistical models (K and Rayleigh distributions) we can get best speck detection signatures to feed unsupervised classifiers and maximize speckle detection performance.
AB - In ultrasonic images, identification of speckled regions helps to estimate probe movement as well as improve performance of algorithms for adaptive speckle suppression and the elevational separation of B-scans by speckle decorrelation. By tracking FDS patch displacements over time we can calculate strain and detect tumor location. Previous studies for speckle detection were based on classification techniques which estimated parameters of the statistical distribution which were based on observation data and ultrasound echo envelope signal. However, in this study, we proposed a new combination of statistical features which were extracted from the ultrasound images and explored their properties for the speckle detection. These features were used as inputs to the unsupervised clustering algorithms for the speckle classification. We used five different types of unsupervised techniques and compared their performance by feeding different combinations of the statistical features. In order to quantitatively compare statistical features and classification methods, as ground truth, we used simulations of cyst and fetus ultrasound images which were generated using Field II ultrasound simulation program[1]. Initial results showed that by combining two statistical models (K and Rayleigh distributions) we can get best speck detection signatures to feed unsupervised classifiers and maximize speckle detection performance.
KW - pattern classification
KW - segmentation
KW - Speckle detection
KW - speckle tracking
KW - Ultrasound
KW - unsupervised clustering
UR - http://www.scopus.com/inward/record.url?scp=84861676515&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84861676515&partnerID=8YFLogxK
U2 - 10.1109/IEMBS.2011.6091997
DO - 10.1109/IEMBS.2011.6091997
M3 - Conference contribution
C2 - 22256221
AN - SCOPUS:84861676515
SN - 9781424441211
T3 - Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS
SP - 8098
EP - 8101
BT - 33rd Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS 2011
T2 - 33rd Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS 2011
Y2 - 30 August 2011 through 3 September 2011
ER -