The Art of Interface

Article 16

3D wavelet compression of ultrasound volume images

Category. 3D medical digital image processing (DIP) software development.

Abstract. The article demonstrates 3D ultrasound voxel image compression.

3D wavelet compression of ultrasound volume images

For archiving or transferring via Internet connection of 3d volume data it is good idea to compress them, because image voxel arrays are really huge in this case. Such data could be got, for instance, as result of 3D tomography or ultrasound acquisition. 3D wavelet compression is an effective technique to pack those graphic data arrays to reduce required storage space or increase transfer speed.

3D wavelet compression resembles the one used currently in JPEG format, just JPEG images are 2D and images in our case are 3D.

The base of compression is 3D wavelet transform, which indeed is combination of 1D transforms performed in every dimension. Applying transform in repeating manner multilevel compression could be performed.

As well 3D transform could be utilized to compress 3D volume animations. And in this case compression ratio could be increased by compressing not the frames itself, but their differences.

In fig. 1a depicted original 3D ultrasound volume image, and in fig. 1b you can see the result of 3D wavelet transform — click Play button for animation.

a.
Original 3D volume image
b.
Result volume image processed by 3D wavelet transform
Fig. 1. Input 3D volume image (a) and result volume image processed by 3D wavelet transform (b).

For 3D volume image processing speed is a critical issue. For Pentium IV at 3.2 GHz the solution we have built has performance 46 Mb/sec for one-level 3D wavelet transform.