http://www.k-wave.org/forum/topic/cart2grid-issue-with-linear-arrays Support for the k-Wave MATLAB toolbox en-US Wed, 13 May 2026 00:21:08 +0000 http://bbpress.org/?v=1.0.2 q http://www.k-wave.org/forum/search.php http://www.k-wave.org/forum/topic/cart2grid-issue-with-linear-arrays#post-434 Tue, 10 Apr 2012 02:54:37 +0000 Bradley Treeby 434@http://www.k-wave.org/forum/ <p>Hi Dan,</p> <p>Thanks for the report and your detective work. We actually check for this error inside <code>private/kspaceFirstOrder_createStorageVariables.m</code> for the specific case of a Cartesian sensor mask used with <code>&#39;CartInterp&#39;</code> set to <code>&#39;nearest&#39;</code>. However, it would seem that this code would be better off inside <code>cart2grid</code> to give a warning for other usage scenarios. We'll migrate this across for the next release.</p> <p>In your case, one option would be to use a binary sensor mask with dx set exactly to the element spacing. This way you won't need interpolation, and you will also be forced to use a grid spacing small enough to capture the same range of spatial frequencies that would be physically sampled by your array.</p> <p>Brad </p> http://www.k-wave.org/forum/topic/cart2grid-issue-with-linear-arrays#post-431 Thu, 05 Apr 2012 21:02:30 +0000 DanR 431@http://www.k-wave.org/forum/ <p>Brad and Ben,<br /> A small issue has popped up as I have done a little modeling of linear arrays. Depending on my selected grid interval and size of the sensor elements, I get a fatal error:<br /> <code><br /> Subscripted assignment dimension mismatch.<br /> Error in kspaceFirstOrder_inputChecking (line 227)<br /> sensor.time_reversal_boundary_data(:, new_col_pos) =<br /> order_index;<br /> Error in kspaceFirstOrder2D (line 427)<br /> kspaceFirstOrder_inputChecking; ...<br /> </code></p> <p>I modified makeCartCircle to generate the Cartesian coordinates of a line of sensor elements. Generally, the elements are quite small. If I choose a dx of 0.044 mm and 9 points per sensor (with two points as spaces), I do not get an error. If I choose dx to be 0.067 mm and 6 points per sensor, I do get this error.</p> <p>I have traced the problem to <code>Cart2grid.m</code>. Looking at the array <code>data_x</code> which holds the Cartesian x-value, I see that it can contain 1 or more duplicate values as a result of the rounding function. Since <code>data_y</code> values are constant (the array was oriented vertically), this put duplicate points into the same position in <code>grid_data</code>, losing the first point. Now the arrays become mismatched. In the past, I have used this code for circular arrays without a problem, but the probability of both the x- and y-coordinates rounding to the same grid point as a previous point is much smaller.</p> <p>Relevant code in <code>Cart2grid</code>:<br /> <code><br /> % scale position values to grid centered pixel coordinates using nearest<br /> % neighbour interpolation<br /> data_x = round(data_x./kgrid.dx);<br /> data_y = round(data_y./kgrid.dy);</p> <p>% shift pixel coordinates to coincide with matrix indexing<br /> data_x = data_x + floor(kgrid.Nx/2) + 1;<br /> data_y = data_y + floor(kgrid.Ny/2) + 1;</p> <p>% map values<br /> for data_index = 1:length(data_x)<br /> grid_data(data_x(data_index), data_y(data_index)) = point_index(data_index);<br /> end<br /> </code></p> <p>I can work around this by judiciously choosing my grid size. Or I can probably reduce the chances of it happening by putting my sensors at an angle. So it is not big issue.<br /> Regards,<br /> Dan </p>