Faculty Bibliography

Objectives: MicroPET imaging has been increasingly performed on mouse models for a variety of human CNS disorders. Despite high demand, digital mouse brain atlases based on PET are still lacking. Further, most microPET systems do not provide means of mapping mouse brain with atlas. For quantitative data analysis and accurate anatomical localization, the development and evaluation of an automated atlas-based data analysis on microPET mouse brain studies is presented. Methods: MicroPET imaging studies were performed after injection of F-18 labeled Amyvid (a tracer for imaging amyloid (Aa) plaques) in isoflurane-anesthetized adult mice using Inveon PET/CT (Siemens). The list mode dynamic PET data were collected for 30-60 min and rebinned using a Fourier rebinning algorithm. A CT scan was also performed for attenuation correction and anatomical co-registration. A 3D digital magnetic resonance microscopy (MRM)-based volume of interest (VOI) atlas generated from live C57BL/6J adult mouse brain was used for brain mapping (Ma et al., 2008). Landmarks, including left and right centroids of midears and eyes (4 landmarks), were generated on atlas template and individual mouse CT images. Co-registration of atlas, CT and PET was performed using Firevoxel (FVX) (https://urldefense.proofpoint.com/v2/url?u=https- 3A__wp.nyu.edu_Firevoxel&d=DgIBAg&c=j5oPpO0eBH1iio48DtsedbOBGmuw5jHLjgvtN2r4ehE&r=KRXe NoRy5_8lkSwAJG5vjS1yT0aFSItfe494dmkdSVs&m=B4bFtJccWjUzJ- dbK1qURkxJmihDqjf87yIgZlYKTMk&s=soyp2V3_QGPs--q8qXcfkDHjv7kMngxeekpEknOQoi8&e= ) and time-activity curves (TAC) for 20 specific 3D brain regions were generated. For comparison, an expert in mouse neuroanatomy manually drew corresponding VOIs on PET-CT co-registered images derived from IRW (Inveon data analysis software without atlas). The TACs thus generated via both methods were compared. For further evaluation, the tracer uptake and kinetics in both tau and Aa transgenic mouse models were also compared. Results: Using FVX, single step co-registration of atlas, CT and PET was accomplished in seconds (by one-button pressing) and the TACs for specific ROIs of mouse brain were automatically generated after co-registration. In contrast, it took an average of 15 min to manually draw a single VOI (total 5 hours/mouse for 20 VOIs) directly on CT images using Inveon IRW without an atlas, a process that required an expert in mouse neuroanatomy. Overall, the TACs for the corresponding VOIs derived from IRW and FVX were similar in counts and shapes. Most importantly, this VOI atlas-based method can provide unbiased measures of radioactivity concentration from PET studies. The results from studies of tau vs. Aa transgenic mouse models after injection of Amyvid showed an apparent difference in the tracer uptake and kinetics (Fig. 1). Conclusions: We have demonstrated the feasibility to map mouse brain with an automated atlas-based co-registration for data analysis of microPET brain studies using FVX. This novel time-saving data analysis methodology, unachievable with current microPET imaging systems, will facilitate accurate assessment and spatial localization of brain signals in mouse model studies for a variety of human CNS disorders

In vivo imaging modalities provide powerful tools for the noninvasive longitudinal characterization of preclinical cancer models. Medulloblastoma (MB) is the most common malignant brain tumor in children, and the subject of intense research, much of which involves mouse models. Manganese-enhanced magnetic resonance imaging (MEMRI) produces unparalleled images of the cerebellum, the site of most MBs [1,2]. For this reason, longitudinal MEMRI of preclinical medulloblastoma models enables analysis of the region of origin, monitoring of tumor progression, and treatment response evaluation. In this study, we present the initial MEMRI characterization of a novel mouse medulloblastoma model with an activating mutation in the Smo gene, which exhibit different growth characteristics than those observed in previous studies of Ptch1 knockout mice [1]. SmoM2 mice were engineered by crossing Atoh1-CreER [3] male mice with homozygous R26-floxedSTOP-SmoM2 females [4]. The SmoM2 mutation was induced by subcutaneous injection of low dose (1mug/g) Tamoxifen (TMX) at postnatal day P2. Biweekly imaging sessions using 7-Tesla MRI (Bruker) began at postnatal day P21. MnCl2 (50-60 mg/kg) was injected intraperitoneally 24 hours before imaging. Scan protocol: 1 min low-resolution pilot, 20 min 150mum resolution T1-weighted GE sequence (TE/TR = 4/30 ms; FA = 20degree; FOV = 19.2 mm x 19.2 mm x 12 mm; Matrix = 128 x 128 x 80). Images were analyzed in 3-space using Amira and Fiji. Morphological characterization was corroborated with histology as shown in Fig1. Longitudinal MEMRI results are summarized in Fig2. Based on our preliminary results, all SmoM2 mice had preneoplastic lesions, while approximately half developed into full tumor morphology (n=21). Of the mice with tumors, approximately 72% developed bilateral tumors and the remaining developed tumors in either the right or left hemisphere. Approximately 50% of animals with bilateral tumors exhibited regression in one lateral tumor and progression in the other, or progression in both tumors (n=8). General disease progression is as follows: at approximately postnatal week W3, small lesions are apparent in the majority of interlobule spaces including the mid vermis; at ~W7, regions of proliferative lesion thickening are apparent and smaller lesions regress; at ~W13 significant tumor encroachment into the forebrain as well as expansion of the third and fourth ventricles are apparent. Tumors were observed to originate in the posterior hemispheres, shift and compress the normal appearing cerebellum as they progress, and finally encroach into the forebrain. Estimated tumor volume doubling time is approximately 4.5 days at early timepoints (W11.5). Noticeable symptoms - including delayed tail-pull reflex, ataxia, and hydrocephalus - in SmoM2 mice were apparent as early as W10. In addition to qualitative understanding of tumor progression, we have manually segmented and quantified tumor volume at these key timepoints in an effort to produce a unified growth model. Current efforts in automated segmentation and hierarchical clustering-based classification of tumors will guide upcoming preclinical trials of anticancer therapeutics

Choosing one's preferred hypothesis requires multiple brain regions to work in concert as a functionally connected network. We predicted that a stronger network signal would underlie cognitive coherence between a hypothesis and the available evidence. In order to identify such functionally connected networks in magnetoencephalography (MEG) data, we first localized the generators of changes in oscillatory power within three frequency bands, namely alpha (7-13 Hz), beta (18-24 Hz), and theta (3-7 Hz), with a spatial resolution of 5mm and temporal resolution of 50 ms. We then used principal component analysis (PCA) to identify functionally connected networks reflecting co-varying post-stimulus changes in power. As predicted, PCA revealed a functionally connected network with a stronger signal when the evidence supported accepting the hypothesis being judged. This difference was driven by beta-band power decreases in the left dorsolateral prefrontal cortex (DLPFC), ventromedial prefrontal cortex (VMPFC), posterior cingulate cortex (PCC), and midline occipital cortex.