Fig 1.
Primary graphical outputs from SMART_FUS.
A) SMART_FUS estimates attenuation and refraction of specific inputs through linear interpolation and also provides a depiction of the simulation with parameters nearest the inputs. Here, a simulation conducted with a bone thickness of 2mm, target depth of 60mm, angle of entry of 90 degrees, a fundamental frequency of 700kHz, and a transducer diameter of 80mm is shown both when performed through a medium containing bone and when performed through a medium containing only water. The -3dB focal region (equivalent to the region exposed to 50% or more of the maximum intensity) is depicted to the right of the full simulations. The point of maximum intensity is highlighted. Finally, against a black background, these -3dB focal regions are overlaid and their points of maximum intensity highlighted again to clearly demonstrate the impact of bone on focal properties. It is important to note that this output would be generated if a close but not identical parameter set were inputted (e.g., with a bone thickness of 2.5mm and all other parameters remaining constant). However, the estimated attenuation and refraction with this unique parameter set would be generated through interpolation and output in the MATLAB terminal. B) A sample output from the script SMARTF_FUS_vis2d.m. Here, the user selected to visualize the impact of bone thickness and carrier-wave frequency at specific values for the remaining three free parameters. This 2-dimensional space is shown for both attenuation and refraction.
Fig 2.
Workflow for the creation of the dataset provided in SMART_FUS.
As described in more detail in the Methods section, a 5-dimensional parameter space was first defined and discretized. From this, 6,048 unique simulation environments were constructed (shown in A). Simulations with bone and water as well as through only water were run for each of these environments (resulting in 12,096 total simulations). The value and position of the peak intensity were located for bone and water simulations at each of the 6,048 points in the space defined. The distance between these points defined the observed “refraction” and the difference in intensity between these points defined the observed “attenuation” at each point in the space. Thus, one 5D “refraction” tensor (an array of 3D matrices) and one 5D “attenuation” tensor were created. These tensors are linearly interpolated according to the values input into SMART_FUS and are used to predict attenuation/refraction at any point within the described parameter space.