A Multi-Atlas Based Method for Automated Anatomical Rat Brain MRI Segmentation and Extraction of PET Activity

Abstract

IntroductionPreclinical in vivo imaging requires precise and reproducible delineation of brain structures. Manual segmentation is time consuming and operator dependent. Automated segmentation as usually performed via single atlas registration fails to account for anatomo-physiological variability. We present, evaluate, and make available a multi-atlas approach for automatically segmenting rat brain MRI and extracting PET activies.MethodsHigh-resolution 7T 2DT2 MR images of 12 Sprague-Dawley rat brains were manually segmented into 27-VOI label volumes using detailed protocols. Automated methods were developed with 7/12 atlas datasets, i.e. the MRIs and their associated label volumes. MRIs were registered to a common space, where an MRI template and a maximum probability atlas were created. Three automated methods were tested: 1/registering individual MRIs to the template, and using a single atlas (SA), 2/using the maximum probability atlas (MP), and 3/registering the MRIs from the multi-atlas dataset to an individual MRI, propagating the label volumes and fusing them in individual MRI space (propagation & fusion, PF). Evaluation was performed on the five remaining rats which additionally underwent [18F]FDG PET. Automated and manual segmentations were compared for morphometric performance (assessed by comparing volume bias and Dice overlap index) and functional performance (evaluated by comparing extracted PET measures).ResultsOnly the SA method showed volume bias. Dice indices were significantly different between methods (PF>MP>SA). PET regional measures were more accurate with multi-atlas methods than with SA method.ConclusionsMulti-atlas methods outperform SA for automated anatomical brain segmentation and PET measure’s extraction. They perform comparably to manual segmentation for FDG-PET quantification. Multi-atlas methods are suitable for rapid reproducible VOI analyses.

Multi-atlas segmentation of biomedical images: A survey

“…MAS has also been employed for the segmentation of cortical and subcortical structures in MRI data from fetuses, neonates, and infants too (Gholipour et al, 2012; Gousias et al, 2008, 2010, 2013; Shi et al, 2010; Wang et al, 2014e; Li et al, 2014; Koch et al, 2014; Makropoulos et al, 2014; Wang et al, 2014d), in which the contrast inversion due to ongoing myelination complicates the segmentation. Another area of application of MAS has been the segmentation of brain MRI in animal studies, e.g., mice (Da et al, 2012; Ma et al, 2014; Nie and Shen, 2013; Lee et al, 2014a; Khan et al, 2014), rats (Lancelot et al, 2014) and non-human primates (Ballanger et al, 2013). Finally, there are also studies that have applied MAS to the analysis of diffusion brain MRI data of humans (Jin et al, 2012; Tang et al, 2014; Traynor et al, 2010), which requires specific strategies for the registration, atlas selection and label fusion steps, due to the nature of the data, which are typically described by directional functions defined on the sphere at each voxel.…”

A Multicenter Preclinical MRI Study: Definition of Rat Brain Relaxometry Reference Maps

“…The individual deformation field was then applied to the corresponding individual anatomical and relaxometry images, and all these images were averaged out to compute anatomical and relaxometry mean templates. The anatomical template was segmented using two multi-atlas approaches (sC3 and sC4) following a similar scheme as depicted in (Lancelot et al, 2014).…”

“…Multi-atlas techniques outperform single atlas approaches for accounting individual structural variability (Wang et al, 2014). Here, we considered two implementations on two different executing platforms, BrainVISA and VIP, of the multi-atlas approach proposed in (Lancelot et al, 2014) for rats. As mentioned in Table 1, some preprocessing steps differ: registration (ANTS (Avants et al, 2010) vs block-matching combined to Free Form Deformation (Lebenberg et al, 2010)) and the number of atlas used (11 vs 12).…”

“…Rat brain parcellation was performed using two multi-atlas approaches similar as the one proposed by Lancelot et al (Lancelot et al, 2014), for a precise and reproducible delineation of brain structures in preclinical in vivo imaging. A maximum probability automatic delineation was obtained by the fusion of several manually delineated images placed in a common space and constituting the multi-atlas dataset.…”

An MRI-Derived Neuroanatomical Atlas of the Fischer 344 Rat Brain

“…This permits researchers to automatically record morphological information, such as the volume of brain regions, at the level of individual structures. 4,5 A number of digital brain atlases for different rat strains have previously been published and disseminated, 6,7,8,9 as shown in Table 1. The current study is motivated by the fact that there is currently no digital atlas for the Fischer 344 rat strain published in the literature or available online.…”