Three-Dimensional Reconstruction for Irregular Ultrasound Sampling Grids

A method for three-dimensional reconstruction from an irregular ultrasound sampling grid, the method comprising:

• acquiring ultrasound data representing sample locations in first and second planes associated with scanning from different apertures along a linear array twisted about a longitudinal axis of the linear array such that the sample locations of the first plane are offset from the sample locations of the second plane by a non-integer amount;
• determining a grid voxel of a three-dimensional grid, the grid voxel between the first and second planes;
• bilinearly interpolating, by a graphics processing unit, a first sample in the first plane from data of the sample locations of the first plane;
• bilinearly interpolating, by the graphic processing unit, a second sample in the second plane from data of the sample locations of the second plane;
• linearly interpolating a voxel value for the grid voxel from the first and second samples;
• and
• rendering an image from the three-dimensional grid, the image being a function of the voxel value.

In a non-transitory computer readable storage medium having stored therein data representing instructions executable by a programmed processor for hardware accelerated three-dimensional reconstruction, the storage medium comprising instructions for:

• acquiring a plurality of range-azimuth slices of ultrasound data, the ultrasound data of the range-azimuth slices representing locations spaced in a two-dimensional ultrasound scan format, the range-azimuth slices being non-aligned along elevation;
• bilinearly interpolating to first and second positions in first and second ones of the range-azimuth slices from the ultrasound data of the first and second range-azimuth slices;
• and
• linear interpolating from the first and second positions.

A method for three-dimensional reconstruction from an irregular ultrasound sampling grid, the method comprising:

• identifying a point in Cartesian space;
• finding two acquired ultrasound scan planes closest to the point, the finding being from a volume of ultrasound scan planes having samples in a scan format;
• interpolating, with hardware acceleration, data at locations on the two acquired ultrasound scan planes, the locations having same coordinates in two-dimensions as the point;
• and
• interpolating a value for the point from the data at the locations.