This invention relates generally to the ultrasound field, and more specifically to a new and useful method of image processing in the ultrasound field.
Conventional ultrasound based tissue tracking systems produce two types of image products. The first type includes tissue displacement image products that describe tissue mechanical properties and that include displacement (axial and lateral), tissue velocity, strain (all components), strain magnitude, strain rate (all components), stain magnitude rate, correlation magnitude. The second type includes traditional image products that describe anatomical and functional characteristics, and that include B-mode, color flow (CF), M-mode, and Doppler. There is a need in the medical field to create a new and useful method to process these spatial-temporal data cubes. This invention provides such new and useful processing method.
The following description of the preferred embodiments of the invention is not intended to limit the invention to these preferred embodiments, but rather to enable any person skilled in the art to make and use this invention.
As shown in
As shown in
As shown in
Step S208 functions to evaluate the quality of the spatial-temporal data 205 such that the contribution of each sample on the model fit may be weighted based on data quality metrics (DQM). Each sample is preferably evaluated based on these data quality metrics, which may be used to identify poor samples in a spatial-temporal data product set. Identification can be a binary indicator (e.g., thresholding), weighting based on the sample DQM or combination. Poor samples may be culled and eliminated prior to spatial-temporal processing based DQM assessment, and may be replaced with a value determined by surrounding valid data (e.g., interpolation). Data quality weighting can be used to adjust the impact of samples on filter output. Many of the filtering techniques described below (e.g., Kalman, parametric modeling) may accommodate data quality weighting. Data quality metrics are preferably calculated for each sample or sub-set of samples of image region, forming DQM map. Preferably DQM's components include: Peak correlation, temporal and spatial variation (e.g., derivatives and variance) of tissue displacement, and spatial and temporal variation of correlation magnitude. Operational DQM may be individual or combination of preferable DQM components.
Step S212 functions to filter the spatial-temporal data 205. There are two preferred methods of temporal filtering data, but any method of temporal filtering may be used. Temporal finite impulse response filtering (FIR) is described by the following equation:
where pn is the data product for the nth image pixel and ck is the sample weighting across temporal window of size T. The temporally filtered result is given by pfn. Temporal infinite impulse response filtering (IIR) is described by the following equation:
This expression is similar to the FIR filter, with the addition of a weighted sum of previous outputs. Both may be spatially variant or invariant (e.g., different weightings given by c & b for each pixel). Temporal filtering is typically done to improve image quality (e.g. reduce noise), but may have other advantages.
Step S212 may also include space-time filtering. Space-time filtering is an extension of temporal FIR processing. The spatial-temporal data product cube is preferably convolved with a 3-D kernel, and can be equivalently done using 3D Fourier transform multiply. The filtering provides control of spatial and temporal characteristics simultaneously. For example, mechanical waves of tissue motion can be reduced or emphasized using space-time filtering.
Step S212 may also include recursive (Kalman) filtering. The Kalman filter is an efficient recursive filter that estimates the state of a dynamic system from a series of incomplete and noisy measurements. The dynamic system in this case is tissue mechanical properties (e.g., tissue displacement products). The weighting of each sample in the recursive filter may be based on a data quality metrics and acquisition time (time history).
As shown in
Step S310 functions to calculate model parameters from the spatial-temporal data 304. The model parameters calculated are preferably amplitude, phase, and error. The model parameters may, however, be any suitable parameters that could be used in a parametric model. The model parameter(s) are preferably calculated based on the product data cube. For example, least square error (LSE) can be calculated from the data to determine model parameters.
Step S310 preferably also functions to fit the model parameters to at least one displacement model. A parametric model or assumed form for tissue displacement products is preferably formulated. The parametric tissue model and estimated parameters are used to determine tissue displacement product at desired times and locations. As shown in
Step S320 functions to calculate new spatial-temporal data based on the model, to replace the original noisy data cube with a new processed data cube 325 calculated from the new parametric model. This new spatial temporal data is preferably calculated to reduce noise, but may also have other advantages.
As shown in
As a person skilled in the art will recognize from the previous detailed description and from the figures and claims, modifications and changes can be made to the preferred embodiments of the invention without departing from the scope of this invention defined in the following claims.
This application claims the benefit of U.S. Provisional Application No. 60/807,881 filed on 20 Jul. 2006 and entitled “Temporal Processing”, which is incorporated in its entirety by this reference.
Number | Date | Country | |
---|---|---|---|
60807881 | Jul 2006 | US |