Development of an Automatic Tare Weight Estimation System for Dialysis Patients Using Pose Estimation and Image Recognition Kento Soejima Project Mentor(s): Mariko Oda, Kohei Arai Accurate tare weight management is essential in artificial dialysis to determine precise fluid removal amounts. Currently, manual confirmation by medical staff can lead to human errors, directly affecting fluid removal accuracy. To address this, we developed an automatic tare weight estimation system utilizing pose estimation and image recognition.The system uses MoveNet to detect human poses, ensuring stable framing for automated capturing, followed by person segmentation and background removal using YOLOv8-seg. For precise clothing matching, we proposed a 3-axis hybrid feature extraction method combining shape (ConvNeXt, 65%), color (HSV histogram, 20%), and texture (Local Binary Pattern, 15%).To prioritize medical safety by preventing the misidentification of unregistered clothing, a multi-layer guard mechanism was implemented. This includes category matching, a hard rejection for low shape similarity (< 42%), and a penalty for low color similarity (< 35%). An optimal operation point was determined through a threshold sweep experiment, setting the final similarity threshold at 0.53. Evaluation experiments were conducted using 90 test images of 17 adult garments in a real-world indoor environment, including diverse poses and similar colors. Compared to using shape features alone (Accuracy 82.2%, FPR 100%) or color alone (Accuracy 41.1%, FPR 28.6%), our proposed method achieved a Top-1 Accuracy of 75.0% while drastically reducing the False Positive Rate to 14.3%. The estimated weight error was ± 206g. The system demonstrates clinical effectiveness by prioritizing the safe rejection of unknown items over incorrect automatic passing. Presentation Type: Pre-Recorded Presentation (https://www.youtube.com/@cwusource5518) Keywords: Artificial Dialysis, Tare Weight Estimation, Image Recognition, Pose Estimation, Medical Safety SOURCE Form ID: 3N Yakima Valley College Effects of Anti-aging Compounds on Growth of Caenorhabditis elegans Aidan Mathews, Ruby Martinez, Aris Camacho Project Mentor(s): Claire Carpenter The disposable soma theory of aging states that an organism’s energy is distributed into three categories: growth, fertility, and cell maintenance. Previous studies have shown that the compounds metformin, taurine, and theanine increase cell maintenance which causes organisms to live longer. Our research explored how 10 μ M and 100 μ M dosages of these compounds affect growth in the model organisms Caenorhabditis elegans (a small nematode). To get our C. elegans into the same life stage we conducted a bleaching protocol that eliminated all larvae and adults, leaving only the eggs, which ensured an equal starting point. We measured C. elegans length and diameter during larval and adult stages using ImageJ. Our research found that all compounds decreased diameter, with theanine decreasing length as well. Future work would be required to see the effect on fertility. The reduction in C. elegans growth observed
in this study is consistent with the disposable soma theory of aging. Presentation Type: Poster Presentation (May 21, 9:30am–3:00pm) Keywords: Anti-aging, Growth, C. elegans SOURCE Form ID: 2N
172
Made with FlippingBook interactive PDF creator