TL;DR
This paper investigates using T1-weighted MRI as a replacement for CT scans in guiding transcranial focused ultrasound (TcMRgFUS) treatments. They found that T1-weighted MRI can accurately estimate skull properties needed for refocusing ultrasound beams, offering a radiation-free alternative that could streamline treatment planning.
What Problem Does This Paper Address?
Transcranial focused ultrasound (TcMRgFUS) is a promising non-invasive therapy, but its effectiveness hinges on precisely focusing ultrasound beams through the skull. The skull's complex shape and varying density distort these beams, requiring accurate correction. Traditionally, CT scans are used to create a model of the skull and calculate the necessary phase shifts to compensate for these distortions. However, CT scans add time to the treatment process and expose patients to ionizing radiation. This study aims to determine if MRI, specifically T1-weighted sequences, can serve as a viable, radiation-free alternative to CT for this crucial skull aberration correction step in TcMRgFUS.
Methods
The researchers compared three MRI pulse sequences (T1-weighted, T2-weighted, and proton density) against CT for their ability to estimate skull properties relevant to ultrasound focusing. They used data from ten adult patients who had both CT and MRI scans.
- Skull Segmentation and Thickness Estimation: MRI data was used to create template masks. Bayesian segmentation was then applied to estimate the thickness of the entire skull and its individual layers (inner table, diploe, outer table).
- Virtual CT Density Calculation: An average CT density was derived from the average MRI intensity.
- Regression Model Development: A training dataset was used to build regression models predicting CT skull thickness from MRI measurements. These models were then validated on a separate dataset.
- Phase Shift Calculation: Standard CT data and the "virtual CT" data derived from MRI were imported into the ExAblate© Neuro system (720kHz) to calculate phase shifts.
- Phantom Study: A degassed human skull embedded in a mannequin head was used to validate the MRI-based phase correction. The skull was filled with hydrogel, and the system was used to induce thermal rise at a focal point using both standard CT and MRI-derived CT data for phase correction.
Datasets
- In Vivo Data: Data from ten adult patients who had undergone both CT and MRI scans for intact skulls. This dataset was split into training and validation sets for developing and testing the regression models.
- In Vitro Data: A degassed human skull embedded in a mannequin head filled with hydrogel, used as a phantom for experimental validation.
Key Insights and Ideas
- MRI as a Radiation-Free Alternative: The core idea is to leverage the widespread availability and non-ionizing nature of MRI to replace CT in a critical step of TcMRgFUS, thereby improving patient safety and potentially streamlining workflows.
- T1-weighted MRI's Superiority: The study highlights that not all MRI sequences are equally effective. T1-weighted MRI emerged as the most promising for accurately capturing the skull's acoustic properties.
- Systematic Over/Underestimation: The finding that T1-MRI systematically overestimates CT thickness for thicker skulls and underestimates for thinner ones is crucial. This suggests that while T1-MRI is a good proxy, its predictions might require calibration or further refinement for extreme skull variations.
- Phantom Validation: The use of a physical skull phantom is a strong point, demonstrating the practical feasibility of using MRI-derived data for actual ultrasound focusing and thermal deposition.
Key Results
- Total Skull Thickness: The T1-weighted MRI sequence showed the closest results to CT for total skull thickness, with a mean discrepancy of 0.025 ± 0.112 mm (p = 0.825).
- Layer Thickness Accuracy: The accuracy of predicting individual skull layer thicknesses (inner, middle, outer) ranged from 0.42 to 0.57 mm.
- Phantom Study Results: For the phantom data, the mean absolute difference between standard CT and MRI-derived CT was 0.8 ± 0.6 radians. This translated to a mean temperature elevation drop of 7 ± 4% when using MRI-based correction compared to CT-based correction.
Limitations and Future Work
- Systematic Discrepancies: The paper acknowledges that T1-MRI systematically overestimates CT skull thickness for thicker skulls and underestimates for thinner ones. This suggests that the regression models might need further refinement or that a direct mapping might not be perfectly linear across all skull types.
- Limited Patient Cohort: The study involved only ten patients, which is a relatively small sample size. Larger, more diverse cohorts would be beneficial for robust validation.
- Future Work: The authors suggest that MRI-based phase correction in TcMRgFUS is a viable alternative to CT-based refocusing, implying that further clinical implementation and validation are warranted. Future work could focus on optimizing MRI sequences, improving segmentation algorithms, and exploring more advanced modeling techniques to further reduce discrepancies.
Summary updated