GPP03 Presentation Time: 10:48 AM

Purpose

The gold standard Brachytherapy (BT) treatment for cervical cancer uses MRI guidance. Some employ MRI-only BT planning, eliminating registration errors between MRI and CT. Challenges in MRI-only BT include accurately reconstructing applicators and catheters due to geometric distortion. Distortions should be < 2 mm for MRI-only BT planning. The EMBRACE II protocol recommends acquiring T2-weighted sequences for contouring and T1-weighted or proton-density weighted sequences for applicator reconstruction. Various sequences have been assessed for MRI-only BT. The commercially-available General Electric (GE) PROPELLER (Periodically Rotated Overlapping ParallEL Lines with Enhanced Reconstruction) sequence has not yet been evaluated for this use. This T2-weighted sequence yields high contrast resolution and reduces motion artifact. This work evaluates the use of the PROPELLER sequence in MRI-only BT planning. The use of another commercially available sequence: 3D LAVA-FLEX (liver acceleration volume acquisition) is also assessed. Utilizing 3D LAVA FLEX enables acquiring a high resolution, high signal, T1-weighted image in a relatively short time.

Materials and Methods

A cylindrical water-equivalent gel phantom was created, incorporating a Venezia (Elekta) applicator with one straight and one oblique needle. Three plastic spheres were glued on the phantom as external coordinate markers. The phantom was scanned using the PROPELLER and 3D-LAVA Flex pulse sequences with a 1.5 Tesla GE wide-bore MRI scanner and body coil. Oblique axial slices oriented in the tandem plane were acquired. The phantom was also scanned on a Philips Big Bore RT CT scanner. The slice thickness was 4 mm for PROPELLER, 2 mm for 3D-LAVA FLEX, and 1.5 mm for CT. In-plane resolution was 1 mm for the MRI images and 0.6 mm for CT images. Images were imported into the Oncentra treatment planning system (Elekta, version 4.6.2). Library modelling was used to reconstruct the Venezia applicator; catheters were manually reconstructed. Physical dimensions measured included the distance between the tandem, needles, and external markers. Measurements acquired from the MRI images were compared to those acquired from CT; distortions were calculated as the absolute difference and a maximum was computed.

Results

Maximum distortions of five total geometric measurements in phantom were 1.3 mm for PROPELLER and 0.7 mm for 3D LAVA-FLEX. Figure 1 reveals axial and reconstructed sagittal images using the investigated sequences and CT for the phantom.

Conclusion

Given that distortions are within 2 mm and image quality and contrast is suitable for reconstruction and contouring, PROPELLER and 3D LAVA-FLEX are appropriate pulse sequences for MR-only BT planning, and their corresponding advantages can thus be exploited.