LV Contours in US Images


- Ultrasound imaging system is a cardiac scaning modality with the highest temporal resolution.
- We want to measure the left ventricualr (LV) volume at each time to provide the quantitative parameters of clinical evaluation of cardiac function. For exmaple, 'volume stroke' or 'ejection fraction'.
- Automateic real-time segmentatino of LV is needed because manual tracing is tedius, time-consuming and impractical.

- We also propose an energy functional using the pre-determined Rayleigh distribution parameter * = 1.
- We incorporate a few control points and shape constraints into the energy functional to prevent contour leakage though gaps by weak edges.
- We numerically implement using the steepest descent approach of the energy functional.

General Electrics


- We analyze the characteristics of speckle of blood and tissue in ultrasound images.
- We propose the decomposition method of Rayleigh distribution parameters 1, 2 from histogram data h of LV image u.
- By the decomposition method, non-linear and ill-posed problem is changed into linear least square problem.

Numerical Experiments

1.  Synthetic Images


2. Real Images 



For more detailed description, see "Fast segmentation of ultrasound images using robust Rayleigh distribution decomposition" by Ahn et al., (2012).

< Circular shape constraint >
< Triangular shape constraint >
< Some contour evolving steps for LV segmentation >
< Segmentation results in various ultrasound images >
< Some examples of LV segmentation >
Research Topics

Electrical Impedance Tomography
Magnetic Resonance EIT
Electrical Property Imaging
LV contours in Ultrasound
Dental CT : Metal Artifacts Reduction
Bioimpedance spectroscopy
Quantitative susceptibility mapping
Image processing
Micro EIT
Lecture Notes
last update 2012.10.26 by Chi Young Ahn