In transcranial magnetic resonance imaging–guided focused ultrasound (TcMRgFUS), a high skull density ratio (SDR) is advantageous to achieve a sufficiently high temperature at the target. However, it is not easy to estimate the temperature rise because the SDR shows different values depending on the reconstruction filter used. The resolution characteristic of a computed tomography (CT) image depends on a modulation transfer function (MTF) defined by the reconstruction filter. Differences in MTF induce unstable SDRs. The purpose of this study was both to standardize SDR by developing a method to correct the MTF and to enable effective patient screening prior to TcMRgFUS treatment and more accurate predictions of focal temperature.
CT images of a skull phantom and five subjects were obtained using eight different reconstruction filters. A frequency filter (FF) was calculated using the MTF of each reconstruction filter, and the validity of SDR standardization was evaluated by comparing the variation in SDR before and after FF correction. Subsequently, FF processing was similarly performed using the CT images of 18 patients who had undergone TcMRgFUS, and statistical analyses were performed comparing the relationship between the SDRs before and after correction and the maximum temperature in the target during TcMRgFUS treatment.
The FF was calculated for each reconstruction filter based on one manufacturer's BONE filter. In the CT images of the skull phantom, the SDR before FF correction with five of the other seven reconstruction filters was significantly smaller than that with the BONE filter (p < 0.01). After FF correction, however, a significant difference was recognized under only one condition. In the CT images of the five subjects, variation of the SDR due to imaging conditions was significantly improved after the FF correction. In 18 cases treated with TcMRgFUS, there was no correlation between SDR before FF correction and maximum temperature (rs = 0.31, p > 0.05); however, a strong positive correlation was observed after FF correction (rs = 0.71, p < 0.01).
After FF correction, the difference in SDR due to the reconstruction filter used is smaller, and the correlation with temperature is stronger. Therefore, the SDR can be standardized by applying the FF, and the maximum temperature during treatment may be predicted more accurately.