A fuzzy feature fusion method for auto-segmentation of gliomas with multi-modality diffusion and perfusion magnetic resonance images in radiotherapy

Abstract

The diffusion and perfusion magnetic resonance (MR) images can provide functional information about tumour and enable more sensitive detection of the tumour extent. We aimed to develop a fuzzy feature fusion method for auto-segmentation of gliomas in radiotherapy planning using multi-parametric functional MR images including apparent diffusion coefficient (ADC), fractional anisotropy (FA) and relative cerebral blood volume (rCBV). For each functional modality, one histogram-based fuzzy model was created to transform image volume into a fuzzy feature space. Based on the fuzzy fusion result of the three fuzzy feature spaces, regions with high possibility belonging to tumour were generated automatically. The auto-segmentations of tumour in structural MR images were added in final auto-segmented gross tumour volume (GTV). For evaluation, one radiation oncologist delineated GTVs for nine patients with all modalities. Comparisons between manually delineated and auto-segmented GTVs showed that, the mean volume difference was 8.69% (±5.62%); the mean Dice’s similarity coefficient (DSC) was 0.88 (±0.02); the mean sensitivity and specificity of auto-segmentation was 0.87 (±0.04) and 0.98 (±0.01) respectively. High accuracy and efficiency can be achieved with the new method, which shows potential of utilizing functional multi-parametric MR images for target definition in precision radiation treatment planning for patients with gliomas.

Evaluation of the Impact of Magnetic Resonance Imaging (MRI) on Gross Tumor Volume (GTV) Definition for Radiation Treatment Planning (RTP) of Inoperable High Grade Gliomas (HGGs)

“…Several studies have also used the ground truth target volume defined by experienced physicians and experts as a reference for comparison [15][16][17]. Ground truth GTV was used for RT of patients and for validation of the results in our study, consistent with the methodology of aforementioned studies [15][16][17]. Given the negligible and statistically insignificant difference between GTV defined by incorporation of MRI and ground truth GTV, we suggest that incorporation of MRI into RTP of inoperable HGGs improves target definition.…”

“…This ground truth GTV has also served as a reference for comparison purposes, since interobserver variations may lead to substantial diversity in definition of target volumes for HGGs. Several studies have also used the ground truth target volume defined by experienced physicians and experts as a reference for comparison [15][16][17]. Ground truth GTV was used for RT of patients and for validation of the results in our study, consistent with the methodology of aforementioned studies [15][16][17].…”

“…In this context, detailed multimodality imaging information may add to the accuracy of target definition in HGGs. The usefulness of neuroimaging with Magnetic Resonance Imaging (MRI) for several purposes including diagnosis, lesion characterization, presurgical planning and navigation for preservation of eloquent brain regions and fiber tracts, brain tumor segmentation, disease monitoring, and discrimination of tumor progression from adverse effects of irradiation has been widely addressed in the literature [13][14][15][16][17][18]. Nevertheless, there is paucity of data on its utility for RT target definition for the distinct group of patients with inoperable HGGs.…”

Advanced Imaging Techniques for Radiotherapy Planning of Gliomas

“…Furthermore, automatic segmentation methods integrating advanced multimodal MRI have been recently proposed to reduce the intrinsic intra- and inter-observer variability of target delineation and to tackle the intrinsic complexity of data analysis [ 104 , 105 ]. Guo et al recently proposed a fusion method for auto-segmentation of gliomas in RT planning using multiparametric dMRI derived ADC, DTI-derived FA, and DSC-derived rCBV.…”

“…The method was preliminarily applied on four clinical image datasets (two low-grade and two high-grade astrocytomas). When compared with manually delineated GTVs, high accuracy and efficiency of the automatic segmentation methods were achieved, suggesting the potential of utilizing functional multiparametric images for GTV definition in precision RT planning of gliomas [ 105 ]. The limitations of multimodality approaches for target delineation are related to the blurred tumor border on advanced MRI maps, due to the relatively low resolution with respect to conventional MRI, as well as to the partial volume effects and the inherent noise in image acquisition that could produce a negative influence on the segmentation results [ 104 , 105 ].…”