A Tri-Modality Image Fusion Method for Target Delineation of Brain Tumors in Radiotherapy

PURPOSE: To develop a tri-modality image fusion method for better target delineation in image-guided radiotherapy for patients with brain tumors. METHODS: A new method of tri-modality image fusion was developed, which can fuse and display all image sets in one panel and one operation. And a feasibility study in gross tumor volume (GTV) delineation using data from three patients with brain tumors was conducted, which included images of simulation CT, MRI, and 18F-fluorodeoxyglucose positron emission tomography (18F-FDG PET) examinations before radiotherapy. Tri-modality image fusion was implemented after image registrations of CT+PET and CT+MRI, and the transparency weight of each modality could be adjusted and set by users. Three radiation oncologists delineated GTVs for all patients using dual-modality (MRI/CT) and tri-modality (MRI/CT/PET) image fusion respectively. Inter-observer variation was assessed by the coefficient of variation (COV), the average distance between surface and centroid (ADSC), and the local standard deviation (SDlocal). Analysis of COV was also performed to evaluate intra-observer volume variation. RESULTS: The inter-observer variation analysis showed that, the mean COV was 0.14(± 0.09) and 0.07(± 0.01) for dual-modality and tri-modality respectively; the standard deviation of ADSC was significantly reduced (p<0.05) with tri-modality; SDlocal averaged over median GTV surface was reduced in patient 2 (from 0.57 cm to 0.39 cm) and patient 3 (from 0.42 cm to 0.36 cm) with the new method. The intra-observer volume variation was also significantly reduced (p = 0.00) with the tri-modality method as compared with using the dual-modality method. CONCLUSION: With the new tri-modality image fusion method smaller inter- and intra-observer variation in GTV definition for the brain tumors can be achieved, which improves the consistency and accuracy for target delineation in individualized radiotherapy.