Automatic segmentation of the rat brain hippocampus in MRI after traumatic brain injury

Registration-based methods are commonly used in the anatomical segmentation of magnetic resonance (MR) brain images. However, they are sensitive to the presence of deforming brain pathologies that may interfere with the alignment of the atlas image with the target image. Our goal was to develop an algorithm for automated segmentation of the normal and injured rat hippocampus. We implemented automated segmentation using a U-Net-like Convolutional Neural Network (CNN). of sham-operated experimental controls and rats with lateral-fluid-percussion induced traumatic brain injury (TBI) on MR images and trained ensembles of CNNs. Their performance was compared to three registration-based methods: single-atlas, multi-atlas based on majority voting and Similarity and Truth Estimation for Propagated Segmentations (STEPS). Then, the automatic segmentations were quantitatively evaluated using six metrics: Dice score, Hausdorff distance, precision, recall, volume similarity and compactness using cross-validation. Our CNN and multi-atlas -based segmentations provided excellent results (Dice scores > 0.90) despite the presence of brain lesions, atrophy and ventricular enlargement. In contrast, the performance of singe-atlas registration was poor (Dice scores < 0.85). Unlike registration-based methods, which performed better in segmenting the contralateral than the ipsilateral hippocampus, our CNN-based method performed equally well bilaterally. Finally, we assessed the progression of hippocampal damage after TBI by applying our automated segmentation tool. Our data show that the presence of TBI, time after TBI, and whether the location of the hippocampus was ipsilateral or contralateral to the injury explained hippocampal volume (p=0.029, p< 0.001, and p< 0.001 respectively).

NIMG-69. PERSONALIZED CONNECTOMIC SIGNATURES: BRIDGING THE GAP BETWEEN NEURO-ONCOLOGY AND CONNECTOMICS

PURPOSE Connectomics has led to significant neuroscientific findings within the last two decades, eventually making impact in the clinics. Neuro-oncology can benefit immensely from connectomics in evaluating structural connectivity of brains with tumor for pre- and post-treatment planning, as a tumor connectome along with derived network measures will make it possible to determine the cognitive effects of treatment and quantify the effect of surgery on quality of life. However, generating connectomes in the presence of tumor is a challenging task. Specifically, registration of an atlas to the brain, which is essential in parcellating the brain into regions of interest, fails around the tumor due to mass effect and infiltration related distortions which are not present in the atlas that comes from a healthy brain. We aim to tackle this problem by introducing a novel atlas registration method. METHOD Although tumor deforms the geometrical shape of its surrounding regions, it does not violate the connectivity of displaced cortical voxels to the rest of the brain. Leveraging this fact, we represent the brain as an annotated graph with nodes representing ROIs encoding geometric features of regions and weighted edges representing the connectivity between regions. In encoding the surroundings of the tumor into the graph, we subsample the region into smaller patches to represent the area with multiple nodes. We then calculate many-to-one graph matching between the graphs of a tumor patient and a healthy control to associate surroundings of tumor with healthy ROIs. OUTCOME A tumor connectome showing how the connectivity is morphed around the tumor, which can further be extended to creating connectomes of recurrence. CLINICAL IMPLICATIONS Use of connectomes can revolutionize neuro-oncology by helping surgeons in estimating structural, functional, and behavioral outcomes of resection prior to surgery and in predicting recovery after the surgery, potentially suggesting subject specific treatment plans.

The NanoZoomer Connectomics Pipeline for Tracer Injection Studies of the Marmoset Brain

We describe our connectomics pipeline for processing tracer injection data for the brain of the common marmoset (Callithrix jacchus). Brain sections were imaged using a batch slide scanner (NanoZoomer 2.0-HT) and we used artificial intelligence to precisely segment the anterograde tracer signal from the background in the fluorescence images. The shape of each brain was reconstructed by reference to a block-face and all data was mapped into a common 3D brain space with atlas and 2D cortical flat map. To overcome the effect of using a single template atlas to specify cortical boundaries, each brain was cytoarchitectonically annotated and used for making an individual 3D atlas. Registration between the individual and common brain cortical boundaries in the flat map space was done to absorb the variation of each brain and precisely map all tracer injection data into one cortical brain space. We describe the methodology of our pipeline and analyze tracer segmentation and brain registration accuracy. Results show our pipeline can successfully process and normalize tracer injection experiments into a common space, making it suitable for large-scale connectomics studies with a focus on the cerebral cortex.