Multi-Institutional Comparative Study Of Mri Technique In Cervical Cancer Image-Based Brachytherapy (Igbt): 3d Mri With High Sampling Efficiency Versus Conventional 2d Multiplanar Mri

In order to elucidate the role of 3D MRI in image guided brachytherapy (IGBT) for cervical cancer, a multicenter study to compare the contrast ratio (CR) and contrast-to-noise ratio (CNR) against the gold-standard 2D approach was conducted. In addition, delineation variability assessment to clarify how different sequences affect the contouring was also performed. A composite phantom consisting of six syringes filled with different T2/T1 relaxation, which comprise the range of relaxation values for representative uterus normal tissues (cervix, myometrium and endometrium) and tumor of three different types of relaxation at 1.5 Tesla, was developed. MRI was performed on a Signa 1.5 T system (General Electric Medical Systems, Milwaukee, WI), an Ingenia 1.5 T system (Pinnacle-Treatment planning system, Best, The Netherlands) and a magnetic resonance diagnostic device 1.5 T system. The imaging protocol consisted of two sequences; a 2D T2 weighted FSE sequence (slice thickness 3 mm) in the axial plane, and an isotropic 3D T2 volumetric sequence (isotropic voxel size of 1.4 mm, acquisition time <6 minutes) acquired in the sagittal plane, which were recommended by manufacturers and considered to be widely available in current clinical practice. From the 3D data-set, axial images were resampled with 1.4 mm slice thickness. The signal intensity in order to calculate the CR and CNR referenced to the cervix syringe was measured. Four radiation oncologists contoured high risk CTV (HR CTV) and GTV of the FIGO IIB case on the post-insertion MR images. MR images included two sets; (A) 3 mm slice thickness 2D T2w axial images, (B) 1.4 mm slice thickness axial images resampled from isotropic 3D T2 volumetric sequence. The values compared between 2D and 3D included volume, DICE coefficient, D90 and V100 to HR CTV. The median CR of 3D and 2D sequence was 0.20 (range, 0.058 - 0.36) and 0.20 (range, 0.062 - 0.35), respectively. CR of 3D sequence was not inferior to that of 2D sequence. The median CNR of 3D and 2D sequence was 51.6 (range, 11.5 - 126.3) and 26.4 (range, 8.3 - 52.7), respectively. CNR of 3D sequence was significantly higher (P<0.001) than that of 2D sequence. The median volume of HR CTV/GTV in 3D and 2D sequences was 18.3 ml/5.6 ml and 17.9 ml/7.3 ml, respectively. The median DICE coefficients were 79.5% for the HR CTV and 67.5% for GTV. The median D90 of HR CTV in each sequence was 8.4 Gy and 8.8 Gy, respectively. The median V100 of HR-CTV in each sequence was 97.9 ml and 100.2 ml, respectively. To the best of our knowledge, this is the first study that quantitatively evaluated the contrast of 3D MRI compared with 2D T2w FSE. Results support the clinical feasibility of our new 3D MRI technique in the IGBT for cervical cancer.