Search for: All records

Abstract Whole-head segmentation from Magnetic Resonance Images (MRI) establishes the foundation for individualized computational models using finite element method (FEM). This foundation paves the path for computer-aided solutions in fields such as non-invasive brain stimulation. Most current automatic head segmentation tools are developed using healthy young adults. Thus, they may neglect the older population that is more prone to age-related structural decline such as brain atrophy. In this work, we present a new deep learning method called GRACE, which stands for General, Rapid, And Comprehensive whole-hEad tissue segmentation. GRACE is trained and validated on a novel dataset that consists of 177 manually corrected MR-derived reference segmentations that have undergone meticulous manual review. Each T1-weighted MRI volume is segmented into 11 tissue types, including white matter, grey matter, eyes, cerebrospinal fluid, air, blood vessel, cancellous bone, cortical bone, skin, fat, and muscle. To the best of our knowledge, this work contains the largest manually corrected dataset to date in terms of number of MRIs and segmented tissues. GRACE outperforms five freely available software tools and a traditional 3D U-Net on a five-tissue segmentation task. On this task, GRACE achieves an average Hausdorff Distance of 0.21, which exceeds the runner-up at an average Hausdorff Distance of 0.36. GRACE can segment a whole-head MRI in about 3 seconds, while the fastest software tool takes about 3 minutes. In summary, GRACE segments a spectrum of tissue types from older adults’ T1-MRI scans at favorable accuracy and speed. The trained GRACE model is optimized on older adult heads to enable high-precision modeling in age-related brain disorders. To support open science, the GRACE code and trained weights are made available online and open to the research community at https://github.com/lab-smile/GRACE. 

Synopsis Animal communication is inherently spatial. Both signal transmission and signal reception have spatial biases—involving direction, distance, and position—that interact to determine signaling efficacy. Signals, be they visual, acoustic, or chemical, are often highly directional. Likewise, receivers may only be able to detect signals if they arrive from certain directions. Alignment between these directional biases is therefore critical for effective communication, with even slight misalignments disrupting perception of signaled information. In addition, signals often degrade as they travel from signaler to receiver, and environmental conditions that impact transmission can vary over even small spatiotemporal scales. Thus, how animals position themselves during communication is likely to be under strong selection. Despite this, our knowledge regarding the spatial arrangements of signalers and receivers during communication remains surprisingly coarse for most systems. We know even less about how signaler and receiver behaviors contribute to effective signaling alignment over time, or how signals themselves may have evolved to influence and/or respond to these aspects of animal communication. Here, we first describe why researchers should adopt a more explicitly geometric view of animal signaling, including issues of location, direction, and distance. We then describe how environmental and social influences introduce further complexities to the geometry of signaling. We discuss how multimodality offers new challenges and opportunities for signalers and receivers. We conclude with recommendations and future directions made visible by attention to the geometry of signaling. 

The geosciences have to solve increasingly complex problems relating to earth and society, as resources become limited, natural hazards and changes in climate impact larger communities, and as people interacting with Earth become more interconnected. However, the profession has dismally low representation from geoscientists who are from diverse racial, ethnic, or socioeconomic backgrounds, as well as women in leadership roles. This underrepresentation also includes individuals whose gender identity/expression is non-binary or gender-conforming, or those who have physical, cognitive, or emotional disabilities. This lack of diversity ultimately impacts our profession’s ability to produce our best science and work with the communities that we strive to protect and serve as stewards of the earth. As part of the NSF GOLD solicitation, we developed a project (Geoscience Diversity Experiential Simulations) to train 30 faculty and administrators to be “champions for diversity” and combat the hostile climates in geoscience departments. We hosted a 3-day workshop in November that used virtual simulations to give participants experience in building the skills to react to situations regarding bias, discrimination, microaggressions, or bullying often cited in geoscience culture. Participants interacted with avatars on screen, who responded to participants’ actions and choices, given certain scenarios. The scenarios are framed within a geoscience perspective; we integrated qualitative interview data from informants who experienced inequitable judgement, bias, discrimination, or harassment during their geoscience careers. The simulations gave learners a safe environment to practice and build self-efficacy in how to professionally and productively engage peers in difficult conversations. In addition, we obtained pre-workshop survey data about participants’ understanding regarding Diversity, Equity, and Inclusion practices, as well as observation data of participants’ responses during the simulations. Follow-up activities include monthly online meetings to engage problem solving and strategy-building skills for catalyzing institutional culture change within departments. This talk will specifically focus on workshop observations and preliminary reactions to the training.