Objective: To segment a 3D ultrasound image data that comprises extraction of surface of interests, smoothing of segmented image, thereby to estimate the surface area and volume of segmented 3D objects (e.g. fetus).
Method: (a) Seeded Region Growing (SRG) together with connectivity and marching cubes algorithms are used to segment the three dimensional (3D) ultrasound image data (I) (b) Using VTK (Visualization Tool Kit) a c++ program was developed which relies on the Maximum Unit Normal Component (MUNC) used for surface area measurement and Divergence Theorem Algorithms (DTA) used for volume estimation. The agreement between the program and the formula was tested on (1) computer generated spheres and cube (2) cylindrical shaped phantoms scanned by ultrasound scanner System Five, GE Vingmed (Horten, Norway) using a 3D annular array of 7.5 MHz. (3) 3D ultrasound fetus using cronbach alpha
Results: The cylindrical shaped phantom (diameter 15.4 mm, length. 21.7 mm) with x and y voxel size 1.24921 mm and 0.613032 mm respectively and z voxel size 1.249221 mm yield a percentage error of 5.3% for the surface area and 2.6% for the volume. The volume and surface area of the fetus with x, y, z voxel size 0.465508mm, 0.529645 mm, 1.014201 mm respectively as estimated with the developed program was 3.758ml for volume and 1.937 mm2 fbr surface area, while the measured volume of fetus with EchoPAC-3D was 3.74 ml.
Conclusions: It was concluded that. The agreement between the formula and the program for different shaped objects indicate that the methodology can provide measurements of the volume and the surface area of different anatomical structures.