ULTRASOUND IMAGE RECONSTRUCTION USING FOCUS INVERSION
Publication Date: 2015-Mar-04
The IP.com Prior Art Database
The present invention provides a focus inversion reconstruction technique for generating an ultrasound image from element data. The element data results from a standard set of transmissions. The standard set of transmissions are fixed focus transmits that are equally spaced set of angles covering a sector. The technique utilizes a retrospective transmit beamforming (RTB) data set for generating synthetic focus images. In the disclosed technique, rather than using a matched filter, a least squares estimate of the transmitted pulse is computed at each location, using all of the available beamformed data. The technique results in sidelobe levels that are at least as low as those of the conventional RTB scheme but without losing axial resolution.
The present invention relates generally to ultrasound system and more particularly to a technique of reconstructing ultrasound image.
Generally, ultrasound data results from a transmission that is focused at certain fixed depth. Ultrasound data is typically dynamically focused on a receive. Due to degradation of image focus at depths which are different from transmit focal depth, several ultrasound systems transmit multiple times in similar look direction with a different focal depth on each transmit. This produces an image that is highly in-focus at all locations. However, use of multiple transmissions reduces frame rate or image update rate which is not desirable and at certain instances not allowable.
Conventional techniques include retrospective transmit beamforming (RTB) schemes that combine single samples from each of a set of transmits. Each sample is formed by a receive beamformer focused on a point that requires reconstruction.
For example, a conventional technique includes a RTB scheme in which multiple time samples are generated for each transmit (TX) beam and for each spatial sample to be reconstructed. The samples constitute a time waveform representing reflection from a point to be reconstructed. The samples are specified to be output of a fixed-focus receive beamformer. The samples are allowed to form multiple output time samples, starting at a minimum round-trip time from a probe to a location whose reflectivity requires reconstruction. These samples are matched-filtered using a model of an expected insonification waveform from each transmits. In general, in an unfocused portion of the TX beam, the waveform is longer than pulse waveform transmitted by a single element. For a given spatial location and a set of N distinct transmits,
Radiofrequency (RF) response matrix is composed of columns si,
S = [ s1 s2 … sN ]
Where the vector si is insonifying waveform at a given location sampled at the RF rate and every N waveform is started at a same base time which is less than or equal to the time of the earliest arrival of an ensemble. The observed echo from a given spatial location, over the N transmits, is modeled as a reflectivity α times the RF response matrix plus an interference matrix:
D = αS + N
In order to estimate alpha, Euclidian norm of N is minimized as:
αopt = minα |N|2 = minα|D – αS|2
Where D-αS has minimum norm when < (D-αS), αS > = 0
Matched filter provides a solution as given by equation below:
Figure 1 depicts comparison of image from matched filter reconstruction and total focusing method (TFM) using simulated data.
As illustrated in above Figure 1, TFM has better transmit/receive focus than matched filter reconstruction.
While the above mentioned conventional technique provides somewhat lower side lobes than a standard RTB approach, uses of the matched filter results in an objectionable loss of axial resolution.
It would be de...