Engineering – Page 15 – UROP Spring Symposium 2021

Engineering

Does air-breathing constrain skull function and diversity in fishes?

The anabantoids are a diverse clade of tropical and subtropical freshwater fishes distributed throughout Africa, Asia, and the Indian subcontinent notable for possessing air-breathing organs (ABOs). The ABO allows the anabantoids to breathe outside of water, making it a key innovation: a trait critical to diversification within a particular lineage. While it is unknown if these organs developed independently or from a single ancestral phenotype, fishes with ABOs are dependent on atmospheric oxygen, even drowning without it. The size and arrangement of the ABO influences the skeleton around it (phenotypic integration), and we predict that the ABO will covary in size and shape with skull structure. We visualized skull anatomy with micro-computed tomography (CT) scanning and then used linear morphometrics to capture skull shape variation. By plotting ABO shape changes against skull shape, this study investigated whether fishes with ABOs have more strongly integrated skulls. Using these data, this study will reveal the evolvability of the ABO itself, and the skulls of fishes with or without these organs. We expect: (1) that skull shape will co-vary accordingly with ABO shape and (2) that fishes with ABOs will have less diverse head shapes. Our dataset can also demonstrate how many times ABOs have evolved in anabantoids and whether these complex organs are capable of being lost, perhaps if no longer needed in certain habitats. We found that the ABOs have evolved two to three times independently across anabantarians.

Does air-breathing constrain skull function and diversity in fishes? Read More »

Reimagining Current Messaging Systems of Social Networks

While social networks have enabled improved communication globally and the widespread Human-Computer Interaction (HCI), it has also exasperated unsolicited communication and harassing messages which consistently target vulnerable, marginalized groups. Consentful Messaging offers Twitter users the ability to filter potential message senders before the risk of receiving unsolicited messages through a system built on Twitter’s API, Python, and JavaScript. The field deployment of Consentful Messaging will reveal how social media users choose to receive messages and to what extent they wish to customize potential message senders. Consentful Messaging is made of four possible customizable functions: the user can set a threshold of number of followers required for a potential message sender, the Consentful Messaging user can verify whether they follow the potential message sender, the Consentful Messaging user can determine whether the potential message sender is followed by at least one account that the Consentful Messaging user follows, and verify whether the Consentful Messaging user has ever replied to a message from the potential message sender. The result is that Consentful Messaging Twitter extension will be deployed in March of 2021. This field study will show how and to what extent Twitter users choose to protect themselves from unsolicited messages with the help of technology. Consentful Messaging will offer not only a customizable layer of protection to marginalized social media users by providing a board for healthy interactions among users, but also information as to how Twitter users interact with and modify personalized computer settings.

Reimagining Current Messaging Systems of Social Networks Read More »

Reimagining current messaging systems on the social internet

While social networks have enabled improved communication globally, it has also exacerbated unsolicited communication and harassing messages which consistently target vulnerable, marginalized groups. Consentful Messaging offers Twitter users the ability to filter potential message senders before the risk of receiving unsolicited messages through a system built on Twitter’s API, Django, MySQL, Python, and JavaScript. The field deployment of Consentful Messaging will reveal how social media users choose to exercise agency over receiving messages and to what extent they wish to customize potential message senders. Consentful Messaging provides the following customizable functions that a Twitter user can use to carve out the network that can initiate interactions: the user can set a threshold of number of followers required for a potential message sender, the system only allows notifications or messages from accounts that the user follows, the user can determine whether the potential message sender is followed by at least one account that the user follows, and verify whether the user has ever replied to a message/tweet from the potential message sender. In short, Consentful Messaging aims to offer a customizable layer of protection to social media users, especially marginalized groups, by providing a board for healthy interactions among users. We plan to deploy the system and conduct a field study sometime in the spring. This field study will help us understand the effectiveness of the approach, and show how and to what extent Twitter users choose to protect themselves from unsolicited messages with the help of technology.

Reimagining current messaging systems on the social internet Read More »

Consentful Messaging: Giving People Agency over Online Interactions using Network Data

Our project works on developing a Chrome extension that allows Twitter users to have more control over their inbox messages and analyzes other accounts to generate potential warnings. This extension tackles the growing issue of online harassment and unsolicited messages on social media by letting Twitter users apply network-based rules on the messages and notifications. We use coding languages such as HTML, Javascript, CSS, Python, and web framework Django to create the frontend of our extension/website as well as the backend where the server runs. Team members can choose to work on the frontend, backend, or both. We will evaluate the extension by conducting a field deployment study on Twitter. We will recruit active Twitter users and ask them to use the extension for at least two weeks. Once the participants are finished using the system, they will be invited to complete a post-study survey and participate in interviews. There aren’t exactly any results or conclusions at this stage of developing the extension. We are currently still working on different parts of the project and individually working on appropriate functions/code. We plan to provide a fully functional and helpful extension that is able to reduce the issue of online harassment (on the user’s side).

Consentful Messaging: Giving People Agency over Online Interactions using Network Data Read More »

Image mosaic stitching/blending for large scale neuronal tracing from mouse brain images

Existing knowledge of how brains work is limited. Specifically, there are major gaps in our understanding of the fundamental building blocks and the procedures by which they work together to achieve higher-level brain functions. Modern microscopy techniques permit imaging large brain volume by taking overlapping image tiles with high resolution. This study aims to implement image processing algorithms to stitch such overlapping tiles and blend them into a seamless continuous image. The blended images can then be used to map and functionally reverse engineer the lower-level functions of mice brains. Currently, progress is still being made in the model construction phase. When completed, these models will hopefully allow advanced machine learning algorithms to identify patterns in the neural maps to reverse engineer the brains which has can lead to subsequent breakthroughs in areas like medicine and thought-controlled devices.

Image mosaic stitching/blending for large scale neuronal tracing from mouse brain images Read More »

Large scale neuron image storage solution using HDF5 file format

In Neuroscience and Neuro-biomedical research, it is crucial for researchers to understand the connections between different neurons to fully realize the mechanism for the brain’s functionality. We have proposed a light imaging-based approach to resolve neural circuits, that is first gathering Brainbow images from the microscope by scanning brain sample sections, and then neuronal structures, including synaptic connections can be reconstructed from these images. It is crucial to let researchers specifying requests that retrieving the spatial range and spectral channels of a subvolume of the images as the Brainbow image files generated from modern high-throughput microscopes are rather large. This causes bottlenecks not only for the software to deliver its functionality due to lack of memory but also for file transferring via the internet, which all brought the necessity of designing a new data storage system to efficiently store and process such files. Our project introduces a customized back-end infrastructure by storing compressed Brainbow images using the HDF5 file format that enables easy access and modification of subvolume images. We explored the influence of file segmentation size on the read and write performance of the Brainbow images, and the data compression and storing pipeline. Finally, we discuss the possibility of designing a distributed system for large scientific data hosting.

Large scale neuron image storage solution using HDF5 file format Read More »

Multidomain protein analogous templates detection based on TM-align

Protein structure prediction is a crucial step to understanding and transforming biological and cellular functions. Most proteins exist with multiple domains in cells for cooperative functionality. However, due to the technical difficulties in structural biology, most of the multidomain proteins have only single domain structures solved. To guide the multidomain protein modeling, we present a two-step procedure method to detect the analogous templates from the multidomain protein structure library which includes the multidomain proteins with known full-length structures through the structural alignment. In the first step, individual domains are used to evaluate each template by TM-align, regardless of the overlap between the alignments of different domains, and the average TM-score of all domains is calculated as the local score of a template. In the second step, the top 500 templates selected from the first step are evaluated by the TM-align again with no overlap allowed in the alignments of different domains, and the average TM-score is defined as the global score of a template. Finally, the template with the best global score is selected as the best template. We test the method over 2,269 non-redundant proteins with 2 domains. With homologous templates with sequence identity >30% to the targets excluded, the results indicated that >80% of target proteins have at least 1 template with a TM-score >0.5 and alignment coverage >90%. The data demonstrate that most interdomain orientations can be inferred from the template library, which probably can be used to assist the multidomain protein structure assembly from the independently determined/predicted domain models.

Multidomain protein analogous templates detection based on TM-align Read More »

Developing fast and unbiased computer vision algorithms

Computer Vision Algorithms are a fast-developing technology, and we have seen from example that they are currently not as accurate and unbiased as we hope they can be. Our project aims to develop a more efficient system and algorithm to reduce bias in computer vision programs. One way bias may be introduced into these algorithms is through issues with light levels in images and videos. Since many computer vision algorithms rely on video cameras, as opposed to infrared or another type of light, the lack of light in videos introduces uncertainty in a program, which can produce bias, where some categories of images are more accurately processed by the algorithm than others. This bias can manifest itself in different scenarios, such as during nighttime or when recording people with darker skin, and these are the biases that we aim to correct. My part in the project involved labelling the videos that are going to be used for analysis for the algorithms, and attempting to help create a standardized method of labelling in order to have a set of videos with which the algorithm can be trained with. Our sample set was purposefully selected to have a variety of videos with different light levels and skin tones. Our ultimate purpose was to label as many videos as possible to use later on in the project, where other groups are working on developing the algorithm and all other overarching parts of the project. The main project was not completed, and likely will not for some years, but we achieved our loose goal of labelling videos.

Developing fast and unbiased computer vision algorithms Read More »

Identifying Brain Edema in CT Scans Using Machine Learning

Brain edema is the swelling of the brain as a result of traumatic brain injuries, strokes, tumors, and infections. This affects the patients cognitive and motor function and can lead to lasting adverse health risks and death. Early and accurate identification of edema can prevent these hazards. Studies have found that brain edema is difficult for clinicians to accurately identify, as it often blends in with other brain matter. Additionally finding a link between the volume of edema and the effect on the patient is considered valuable, but there is currently no standard software in place for this end. Even when clinicians are able to identify edema, they are not able to quantify the volume present. Convolutional neural networks were used for training the model to segment the edema region. Images from the PROTECT III collection at the University of Michigan hospital were used for this research. Some images were previously annotated by clinicians and these images were subsequently used in the process of training the machine learning model. The performance of the model was evaluated using quantitative techniques such as dice, sensitivity, specificity, accuracy, and AUC. The goal of this software is to decrease adverse effects and death related to brain edema by creating a system to quantitatively measure edema and make informed decisions on how to treat the patient based on the information collected.

Identifying Brain Edema in CT Scans Using Machine Learning Read More »

Developing fast and unbiased computer vision algorithms

In an effort to improve driver safety and autonomous vehicle testing, video recordings of drivers allow for data to be analyzed. These videos are first examined by human coders, but a more efficient, automated algorithm would prevent the need for human coders entirely. However, in order to build the algorithm, human coders need to analyze videos of drivers and label various actions, such as if the driver is turning or tilting their head, or hand movements, such as texting, and if their hand is obscured. Once these labels are implemented, they are tested against each other for accuracy, so that the final algorithm is unbiased enough to be implemented into vehicle safety.

Developing fast and unbiased computer vision algorithms Read More »

lsa logoum logo