M. Lam

Quality of MR thermometry during palliative MR-guided high-intensity focused ultrasound (MR-HIFU) treatment of bone metastases

BackgroundMagnetic resonance (MR)-guided high-intensity focused ultrasound has emerged as a clinical option for palliative treatment of painful bone metastases, with MR thermometry (MRT) used for treatment monitoring. In this study, the general image quality of the MRT was assessed in terms of signal-to-noise ratio (SNR) and apparent temperature variation. Also, MRT artifacts were scored for their occurrence and hampering of the treatment monitoring.MethodsAnalyses were performed on 224 MRT datasets retrieved from 13 treatments. The SNR was measured per voxel over time in magnitude images, in the target lesion and surrounding muscle, and was averaged per treatment. The standard deviation over time of the measured temperature per voxel in MRT images, in the muscle outside the heated region, was defined as the apparent temperature variation and was averaged per treatment. The scored MRT artifacts originated from the following sources: respiratory and non-respiratory time-varying field inhomogeneities, arterial ghosting, and patient motion by muscle contraction and by gross body movement. Distinction was made between lesion type, location, and procedural sedation and analgesic (PSA).ResultsThe average SNR was highest in and around osteolytic lesions (21 in lesions, 27 in surrounding muscle, n = 4) and lowest in the upper body (9 in lesions, 16 in surrounding muscle, n = 4). The average apparent temperature variation was lowest in osteolytic lesions (1.2°C, n = 4) and the highest in the upper body (1.7°C, n = 4). Respiratory time-varying field inhomogeneity MRT artifacts occurred in 85% of the datasets and hampered treatment monitoring in 81%. Non-respiratory time-varying field inhomogeneities and arterial ghosting MRT artifacts were most frequent (94% and 95%) but occurred only locally. Patient motion artifacts were highly variable and occurred less in treatments of osteolytic lesions and using propofol and esketamine as PSA.ConclusionsIn this study, the general image quality of MRT was observed to be higher in osteolytic lesions and lower in the upper body. Respiratory time-varying field inhomogeneity was the most prominent MRT artifact. Patient motion occurrence varied between treatments and seemed to be related to lesion type and type of PSA. Clinicians should be aware of these observed characteristics when interpreting MRT images.

Short and long time MR signal behavior of randomly distributed water and fat-numerical simulations.

The MR time‐signal behavior of water has been reported to be different on short and long time scales for systems of randomly distributed perturbers in water in the static dephasing regime. Up to now, the signal of the perturbers in such systems has not been taken into consideration. Water–fat emulsions are macroscopically homogeneous systems and can be considered as microscopically randomly distributed perturbing fat spheres embedded in water. In such water–fat systems, the signal of the perturber, fat, cannot be ignored. Since water and fat are within the same system, the fat signal behavior may show similarities with water, with differences in short and long time scales. This could complicate fat‐referenced MR thermometry (MRT) methods such as multi‐gradient echo‐based (MGE) MRT. Simulations were performed using a numerical phantom comprising spherical fat objects embedded in a spherical water medium. To characterize the fat signal, the theoretical signal description of water was fitted to the simulated fat signal. The simulated signals were sampled as an MGE signal and MGE MRT was used to calculate temperatures. The sampling was done with and without delay, to investigate the effect on the temperature error of the time ranges in which the signal was sampled. It was confirmed that the fat signal behavior was similar to that of water and consisted of two regimes. The separation between the short and long time scales was approximately at 55 ms for fat, as compared with 8.9 ms for water. Without delayed signal sampling, the MGE MRT temperature error was about 2.5°C. With delayed sampling such that both the water and the fat signals were either in the short or in the long time scale the error was reduced to 0.2°C.