Null subtraction imaging for small aperture phased array transducers

Abstract

In medical ultrasound imaging phased array transducers are used for non-invasive imaging. For 3D intra-cardiac-echography technical borders are reached since blood vessel size limits the dimensions of the array. This dimensional limitation lowers lateral resolution. To solve part of the problem, non conventional ways of lateral resolution improvements are needed. For linear arrays R. Reeg [1] proposed null subtraction imaging (NSI), which resulted at best in 30 dB lower side lobes and a width reduction of up to 25 times for the main lobe. This is a non-linear image processing technique based on implementing multiple apodizations in post-processing. Three images are made from which one has a sharp drop to zero in the middle which can be exploited by subtracting it from the other images. This results in a improvement over a regular apodization scheme.
To verify the technique for 2D phased array imaging, simulations are done using Field-II [2, 3] while taking array dimension into account. Measurements take place on an artificial phantom with evenly spaced line scatter targets. For the experiment a P4-l phased array transducer was used in combination with a Verasonics Vantage 256TMwhere a sub-aperture of 30 elements is used. The experiment is done on a CIRS 040GSE Phantom, where evenly spaced line scatter targets and lateral resolution targets are looked at. To form an image a simple delay-and-sum (DAS) algorithm was used. The effect of NSI on the speckle in an image is also measured, where for normal developed speckle a signal to noise ratio (SNR) of 1.91 is expected. Results for simulation were as expected with an average beamwidth reduction of 22.5 times and on average higher main lobe to side lobe ratio (MSR) of 33.4 dB for NSI imaging opposed to using a rectangular apodization. Experimentally an average beamwidth reduction of 5 times and an on average higher MSR of 20.5 dB are realised. The beamwidth reduction is lower then expected, this is due to rounding errors in the DAS algorithm, where data can only be delayed by integer elements. For the lateral scatter targets no extra targets can be distinguished when using NSI imaging opposed to rectangular apodization imaging, only the targets that where already visible show reduction in beamwidth and side lobe levels. When using NSI speckle SNR is reduced from 1.78 to on average 0.50 and seems to scale with energy present in the image. An hybrid image between the NSI image and rectangular apodization image is made to solve this. It results in reduced beamwidth, restored speckle SNR but the MSR is reduced. It can be concluded that the method works for phased array imaging under perfect conditions, the simulation. To get the full potential of the technique more research is needed towards the effects of improper sampling. Which in this case resulted in less then ideal ’sharp’ drops, lowering the reduction in beamwidth. Also the reduced speckle needs further research to understand it well. Once this is all achieved the next step to 3D imaging can be made.