Schematic of a 2D cross-correlation displacement estimation showing (a) the reference image and (b) the search image. The algorithm finds the pattern in the search image that most closely corresponds to a unique pattern in the reference image. The distance between the two then provides an estimate of the local displacement . For computational efficiency, the algorithm only searches for the best match within a pre-defined search area.
(a) -mode ultrasound image of an ultrasound tissue-equivalent phantom, acquired using a curved array transducer, with the SL25 off. The imaging depth was . (b) -mode ultrasound image acquired immediately after (a), but with the SL25 beam switched on. Note the presence of “noise spots” in the image caused by the SL25. (c) The result of subtracting image (a) from image (b) to enhance the spatially periodic, localized noise that occurs while the SL25 is on. The noise is expressed as the percent change in the greyscale value relative to image (a).
-mode images in air acquired with a linear probe while the SL25 is irradiating, operating at (a) and (b) . The images were reconstructed from captured rf images. The gain on the scanner was set to maximum to enhance the visibility of the spatially periodic noise.
(a) -mode image of a normal prostate (outlined in solid black) acquired using a linear probe. The gain and other display settings were chosen so as to be appropriate for prostate imaging. (b) -mode image in air acquired with a linear probe while the SL25 was irradiating, operating at . The imaging depth was . The scanner settings were identical to those used in (a). (c) Vertical profile through the -mode image at the position of the second deepest “noise spot” visible in (b). (d) Amplitude histogram (open circles) of the prostate outlined in (c). The magnitude of the peak of the vertical profile through the noise spot has been shown for comparison (solid line).
A plot of the spatial mean displacement calculated from consecutive frames taken at half-second intervals for both the SL25 on (circles and solid line) and off (squares and dashed line). The phantom was moving at . The average displacement between frames was found to be with both the SL25 on and off.
Plots showing the calculated displacements for two different reference window sizes, (a) and (b) . The phantom was moving at . The solid horizontal line at indicates the interframe displacement that would be expected for smooth operation of the motors. The calculated displacements for the SL25 off case are shown with solid squares and dashed lines, while the results for the SL25 on case are shown with solid circles and solid lines. The uncertainty bar for each point is the standard deviation on the mean displacement value within the studied region, where for (a) and for (b).
Spatial mean value of the maximum cross-correlation coefficient as a function of the area of the reference window, while the SL25 was on.
(a) Isodose distribution in the Rando phantom derived from the ADAC plan calculated for the three-field prostate treatment. Two sets of numbers have been superimposed on the image. The numbers in parentheses show the positions of the 27 TLD chips loaded into the phantom. The positive and negative values indicate the difference in dose (in cGy) recorded by the TLD chips with the transducer present and absent during irradiation. A positive value indicates that the dose was higher when the transducer was present. (b) Dose as a function of TLD chip position according to the ADAC plan (dashed line), the TLD readings with the transducer present (closed triangle and dashed line) and the TLD readings with the transducer absent (closed squares and line).
The mean and standard deviation of the displacement estimates and cross-correlation values for different sizes of the reference window.
Article metrics loading...
Full text loading...