Diagnostic Training System Based on Ultrasonic Image Synthesis
Using View-dependent Model
Abstract
Since ultrasonic diagnosis requires substantial experiences for both acquiring and reading echography, an effective education for physicians is desirable.
This research proposes an echo imaging simulator for synthesizing and displaying echography, according to user’s probe manipulation.
The proposed method is a samplebased approach that stores pre-captured real echo images as a volume data set in which each voxel has multi-directional values.
Output echography is synthesized by interpolating voxel values, considering spatial arrangement and directivity of the voxel values.
Experimental results demonstrated spatially seamless echography synthesis and characteristic artifacts with various probe postures.
Result
Simulation of a tadpole tail sign
 |
 |
(a) Generation Cross Section | (b) Real Ultrasonic Image |
 |
 |
(c) Conventional Method | (d) Proposed Method |
Simulation of a multiplex reflection
 |
 |
(a) Generation Cross Section | (b) Real Ultrasonic Image |
 |
 |
(c) Conventional Method | (d) Proposed Method |
poster of research introduction