Purpose: To optimize an MRI-based proton therapy treatment planning simulation protocol in patients with ocular melanoma using a dedicated 70-mm loop coil.
Methods: An SNR comparison was performed by acquiring T1 TSE images of a phantom with a 20-channel head coil or a 70-mm loop coil using identical parameters at 1.5T. Three normal volunteers underwent MRI of the eye for sequence development. An iterative optimization process was performed on several 2D and 3D spin echo and gradient echo sequences, maximizing spatial resolution, minimizing artifacts, and maintaining an SNR of at least 2.5 for 2Ds and 25 for 3Ds. Then, under IRB approval, patients with ocular melanoma were recruited to partake in a treatment planning study after their initial consultation and simulation CT, using the non-contrast MRI protocol as outlined in Figure 1.
Results: Higher SNR was measured with the loop coil (735) than the 20-channel head coil (53). Volunteer scanning confirmed the increased SNR from the loop coil allowed for higher-resolution imaging than the head coil. Spin echoes (2D TSE and 3D SPACE) demonstrated fewer susceptibility artifacts around the orbits and sinuses than gradient echoes (2D GRE and 3D CISS). The use of oversampling on 2D sequences and saturation bands on 3D sequences was critical to avoid aliasing. Tumors were well-delineated in Patients 1 and 2 as positive contrast on T1 images and negative contrast on T2 images. However, Patient 3 had a very thin tumor (less than 1 mm) near the cornea, which was not well visualized due to volume averaging.
Conclusion: This work demonstrated the optimization of an MRI treatment planning simulation protocol for proton therapy in patients with ocular melanoma. The use of a loop coil allows for high-resolution imaging of tumors greater than 1-mm thick at 1.5T with scan times less than 5 minutes per sequence.
MRI, Treatment Planning, Ophthalmic Applicators