48440 research outputs found
Sort by
Perceptions of Microaggressions among Black Students at WPI: A Semi-Structured Interview Study
No full textThe purpose of the project was to understand the experiences of Black WPI students, specifically if they had encountered racial microaggressions during their time at WPI. Data from 18 Black WPI students was acquired via semi-structured one-on-one interviews. The results of this project showed that while microaggressions were a significant issue for Black WPI students, Black student groups such as the National Society of Black Engineers (NSBE) and the Black Student Union (BSU) provided safe havens of Black pride and unity for these students. The data gathered in this project acts as evidence for why WPI needs to be more engaged with the experiences and voices of its Black student organizations and population. Recommendations for WPI include a greater focus on continual support for Black student groups and an effort to increase the number of Black staff members and students. It is recommended that research on racial microaggressions at WPI continue using a larger and more representative sample
Hardware Accelerated AI Super Resolution
No full textThis project brings machine learning and hardware development together to implement real-time video upscaling algorithms on FPGAs. Video frames from a Nintendo Wii are analyzed for variance and routed to either bilinear interpolation or a convolutional neural network for upscaling. Hardware was developed and verified in Vitis HLS C++. The two development boards used for testing and implementing the design were the KV260 and ZCU102. This report outlines the team’s successes and identifies areas for improvement in future designs
Improving Automated Coronal Hole Detection Algorithms through Coronal Hole Annotation using Semi-automatic Methods
No full textCoronal holes are characterized by their low density, open magnetic fields, and low solar activity which appear as dark regions in extreme ultraviolet imaging of the Sun (Cranmer, 2009). To provide forecasters with a snapshot of the solar environment at a single point in time (Horan, Karen et al.), researchers at the National Oceanic and Atmospheric Administration (NOAA) Space Weather Prediction Center (SWPC) have hand-drawn synoptic maps of the visible solar disk, including coronal holes, every day since 1972 (Solar Synoptic Map | NOAA / NWS Space Weather Prediction Center), leveraging data from satellite and ground-based imagery of the Sun. While various automated coronal hole detection techniques have been developed since 2007 (Garton et al., 2017; Henney & Harvey, 2007; Illarionov & Tlatov, 2018; Jarolim et al., 2021; Verbeeck et al., 2013), these SWPC drawings have primarily been used to visually validate the output of models but have not been used to train them (Garton et al., 2017; Hughes et al., 2019; Illarionov & Tlatov, 2018). This paper introduces CHASM (Coronal Hole Annotation using Semi-automatic Methods), a novel tool based on the Segment Anything Model (Kirillov et al., 2023) that enables users to quickly digitize SWPC maps into binary segmentation masks. Using CHASM, we labeled a dataset of 1,111 pre-segmented SWPC masks, making this the first available digitization pipeline for these historically accurate drawings (Garton et al., 2017). To demonstrate the utility of this dataset, we compared the performance of the pretrained CHRONNOS model to versions trained on the CHASM masks. Evaluated on unseen test CHASM masks, our leading model achieved an accuracy of 0.9805, a True Skill Statistic (TSS) of 0.8253, and an intersection-over-union (IoU) of 0.5810, whereas the pretrained CHRONNOS model achieved an accuracy of 0.9767, a TSS of 0.7000, and an IoU of 0.5022 on this test set. Bringing this data resource into modern coronal hole detection approaches provides an untapped opportunity to advance automated solar feature detection
3D Printed Custom Knee Brace and Test Fixture
No full textThe purpose of the “3D Printed Knee Brace” project was to design and test a custom athletic knee brace at a significantly lower cost than what is currently available. The brace was designed and printed using the available tools and printers at WPI to ensure WPI could replicate this process for these braces for their current student-athletes. Another goal of this project was to produce a standardized test for knee braces to ensure they reduce the forces acting on the knee during sports
Characterization of Single-Cell Anisotropy via Indentation and F-Actin Visualization
No full textA cell’s mechanical response to its environment, particularly in modulating its stiffness, is a key aspect of its role in tissue development, maintenance, and adaptation. Changes in cellular stiffness can help predict disease progression, including cancer. One of the primary contributors to stiffness is filamentous actin (F-actin), a cytoskeletal protein that forms interconnected networks throughout the cytoplasm to support cell structure and enable force transmission and mechanotransduction. Many adherent cells display anisotropy, meaning they have direction-dependent mechanical properties. This phenomenon is believed to stem from the internal alignment and tension of cytoskeletal filaments like F-actin. Understanding and accurately measuring anisotropy is critical for gaining insights into both normal cellular functions and pathological conditions. However, current techniques for assessing cell anisotropic stiffness — such as confocal microcopy concurrent with indentation by an atomic force microscope (AFM), magnetic twisting of microscopic beads attached to the cell surface, stretching while measuring traction force, and indenting in multiple directions using a toroidal probe — are complex and require specialized equipment. We hypothesize that a combination of spherical nanoindentation with F-actin fluorescence imaging can be used to predict cellular mechanical anisotropy. The prediction model was found to be sufficient in capturing the general anisotropic nature of the VICs, therefore allowing anisotropy to be characterized with simpler, faster techniques than those previously used
AI Equipped Ultrasound for In-Field Identification of Traumatic Bleeding
No full textInternal bleeding is a leading cause of preventable death in trauma patients, particularly in environments where rapid diagnosis is essential. Medical professionals in military, prehospital, out-of-hospital, and low-resource clinical settings across the U.S. face critical challenges in identifying internal hemorrhaging quickly and accurately. This project addresses that need by integrating artificial intelligence (AI) with diagnostic technology to localize and assess the severity of internal bleeding in real time. Our AI-enhanced tool accelerates trauma detection, enabling clinicians to make faster, more informed decisions that can ultimately save lives. By reducing diagnostic time and improving accuracy, this solution has the potential to transform emergency care. While the project lays the groundwork for continued research and future biomedical engineering careers, its core mission is the advancement of life-saving medical technologies
Decellularization of Adipose Tissue for Breast Tissue Regeneration
No full textThis project aims to design a process to decellularize adipose tissue and concentrate the regenerative proteins for collection. The concentrated protein paste will be used for regeneration of breast tissue for reconstruction after a lumpectomy procedure. Our method is intended to be used as an alternative to silicone implants and fat grafting procedures as it provides a natural option for women. The method had to be accomplished only using additive components that are approved for the operating room, and no harsh chemicals, guaranteeing safe reinsertion of the proteins into the body. The team considered the specifications of an operating room and tailored our process to meet these standards. Our team believes this project will help women experience an improved quality of life and feel confident in themselves. In the future, the concentrated protein should be tested for regenerative properties, helping to improve its efficacy for clinical use. Once this is confirmed, our process can be developed into a device that can safely be used in the operating room during lumpectomies
Exploring the Role of Food Processing in Transforming Apple Allergens
No full textApples are one of the most common fruit allergies. Allergies are triggered by the binding of allergen proteins to antibodies in the body. Understanding how allergen proteins can be structurally modified is critical to potentially mitigate food sensitivities. This report aims to observe the effect of food processing methods on the physical and structural changes to apple allergens. Apples were processed via the following methods: baking, blanching, boiling, dehydrating, soaking in ethanol, freezing, and soaking in lemon juice. Proteins were isolated through cell lysis and chloroform-methanol extraction to yield a solid pellet. All trials resulted in a successful extraction except ethanol-soaked samples. Pellets were analyzed using multiple analysis techniques, with Fourier-transform infrared (FTIR) and nuclear magnetic resonance (NMR) spectroscopy proving to be the most effective, thus allowing secondary protein structure to be examined. It was found that all process methods except freezing suggested noticeable changes in secondary structure from FTIR peak shifts in the amide I band region, around a wavenumber of 1,600 cm^-1. This change suggests a decrease in alpha helix content and an increase in either beta sheets or aggregated structures. Baking showed the most drastic structural changes in both FTIR and NMR spectra, however, the observed trends were not as consistent throughout trials as were in boiled and blanched samples, which also exhibited changes in the amide I region. Limitations in analysis occurred as a result from difficulties redissolving the pellets into different solvents required for certain techniques. Future studies should refine re-solubilizing the pellet as this could expand analysis, allowing for the collection of data on concentration as well as primary and tertiary structures
Regulation of the DNA repair protein Xrs2 by Cdk1
No full textCell cycle progression is predominantly regulated by cyclin dependent kinases (Cdks) through phosphorylation of key proteins. Among them is Xrs2, a component of the Mre11-Rad50-Xrs2 (MRX) complex in budding yeast, which plays an essential role in maintaining genome stability by facilitating repair of DNA double-strand breaks (DSBs). Xrs2 was previously identified as a Cdk1 substrate, however it remains unclear how phosphorylation of Xrs2 regulates its function. In this study, we examined the biological implications of seven Cdk1 target sites on Xrs2. We found that, while phospho-variants (Xrs2-7A and Xrs2-7E/EE) showed distinct migration patterns on SDS-PAGE, growth of these mutants were indistinguishable compared to the wild-type. Our results reveal that Cdk1 promotes phosphorylation at multiple sites of Xrs2, however effects of this phosphorylation on cellular fitness remain to be further investigated
Developing Grant-Making Strategies to Support Small Charities
No full textIn collaboration with Merton Connected, a charitable organization in the London Borough of Merton, our project identified the fundraising challenges facing small charities in Merton and proposed grant-making strategies to meet these challenges. Through 56 surveys and 20 interviews of representatives of small charities and funders, we identified the key fundraising strategies that small charities apply, the challenges they face securing grants, and funders’ perspectives on these challenges. We recommend that Merton Connected increase awareness of grant opportunities, provide unrestricted grants, offer collaborative grant funds, and implement an optional digital portion in their grant application. We hope these recommendations allow charities to continue supporting the residents of Merton