High Intensity Focused Ultrasound (HIFU) is a noninvasive thermo-therapeutic modality for the localized treatment of disease. In addition, Magnetic Resonance Acoustic Radiation Force Imaging (MR-ARFI) provides quantification of the displacement proportion to the intensity at the ultrasound focus. This study evaluates the combination of two algorithms^1,2 which maximize the intensity at the focal point to compensate for tissue related aberrations in clinically acceptable conditions. The proposed method was evaluated in a phantom using weak, medium and strong aberrators with acquisition times of 10 and 25 minutes at 3T. The Sonalleve V1 MR-HIFU system in a 3T Achieva MRI (Philips Healthcare, Best, The Netherlands) was used to apply ultrasound pulses up to 300 Wac at 1.2 MHz synchronized with an MR-ARFI sequence based on an EPI gradient echo sequence providing low distortion³ with simultaneous measurement of the displacement and the temperature⁴.
The set of optimal phases Φopt were characterized for 64 groups of 4 elements of the 256 elements phased array transducer. To maximize the intensity at the focal point, sonication patterns composed of an Hadamard combination of these groups of elements² were applied using 6 phase steps to fit a cosine function to measured displacements¹. These 384 sonication patterns were acquired in 10 minutes; 5 minutes to modify the sonication patterns and 5 minutes to collect MR-ARFI data. This algorithm was repeated 4 times to assess the reproducibility and compare the results with an acquisition time equivalent to 25 minutes (5 minutes transition and 5 minutes ×4 collection of MR-ARFI data).
Three sets of phases Φ^ab, acted as virtual aberrators, which represented a neonatal skull, an 8-year-old skull and an artificial aberrator (arranged as a checkered pattern of ±60°) were added to each sonication pattern at the software level. This method allowed to quantify the Error = std(Φ^opt- Φ^ab) of the measured phase. Four additional MR-ARFI acquisitions were used to quantify the relative intensity at the focal point with and without refocusing the beam using Φ^opt. Figure 1 shows the relative intensity at the focal point for the three virtual aberrators with and without beam refocusing. When using no aberrator the refocusing algorithm provides an increase in the intensity of 2.9±4.7 % and 5.7±2.3 % with a 10 and 25 minute acquisition time respectively. The three virtual aberrators induced a relative intensity of 95±3.6 %, 67.4±2.7 % and 25.3±1.8 %. The refocusing of the beam based on the 10 minute acquisition time achieved a relative intensity of 101.6±4.5 %, 91.3±1.9 % and 93.3±3.5 %. The usage of a 25 minute acquisition time provided an improvement of the relative intensity up to 103.9±3.7 %, 94.3±6.2 % and 101±3.9 %.
Figure 2 indicates that the optimal phase Φ^opt matched the three aberrator phases Φ^ab with an error of 12.6°, 22.1° and 18.3° for the 10 minute acquisition time. A smaller error of 10.6°, 22° and 11.4° were obtained with the 25 minute acquisition time. The average temperature increase measured during each collection of MR-ARFI data was 4.2±1 °C. Since more than 100 % of the relative intensity (obtained when no phase shift is applied to the transducer elements) was observed after refocusing the beam suggests that part of imperfections inherent in the system were also compensated. These experimental imperfections might be due to irregular transducer curvature or dephasing associated with the oil tank.
While a 10 minute acquisition time is sufficient for weak aberrators such as the neonatal skull, a longer acquisition duration is preferable for stronger aberrators such as the artificial aberrator. Additional acquisition time was not providing significant gain in refocusing the beam for the 8-year-old skull aberrator because the distribution of the Φ^ab had strong spatial irregularities which didn’t match the decomposition in 64 groups of 4 elements. For this aberrator the quantification of the optimal phase for the 256 elements individually would be necessary to achieve a better refocusing.
Three virtual aberrators were successfully refocused using an MR-ARFI acquisition time of 10 minutes. Refocusing could be further improved for stronger aberrators using a 25 minute acquisition time. This technique was applied with an acceptable temperature increase.