Scholarlink

Due to its role in metabolizing alcohol and its impact on overall health, the liver is a crucial organ to understand the physiological impacts of drinking behaviors among non-human primates. Studying liver histology can provide insights into potential physiological variation that may arise as a result of alcohol use. This study analyzed sections from hematoxylin and eosin-stained liver slides sampled from a cohort of 8 male rhesus macaques. We trained a deep learning model to classify the primates into previously identified classes of drinking behavior based on the slide images: non-drinkers, low drinkers (LD), binge drinkers (BD), heavy drinkers (HD), very heavy drinkers (VHD). Our model performed with high accuracy, distinguishing between drinking and non-drinking animals with 98% accuracy and with 94% accuracy when classifying LD versus VHD. Our results suggest that differential exposure to ethanol can have subtle, but identifiable, effects on liver tissue. This approach could provide a way to assess the impact of alcohol on liver tissue and contribute to models designed for automated detection of early signs of alcohol-related liver disease. — Track: CSCI-RTCB

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Deep Learning Predicts Non-human Primate Drinking Category from Liver Histology

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Bikram Khanal,
  • Carissa Fong,
  • Sri Manjusha Tella,
  • Rupak Khadka,
  • Steven W. Gonazales,
  • Kathleen A. Grant,
  • Mary Lauren Benton

摘要

Due to its role in metabolizing alcohol and its impact on overall health, the liver is a crucial organ to understand the physiological impacts of drinking behaviors among non-human primates. Studying liver histology can provide insights into potential physiological variation that may arise as a result of alcohol use. This study analyzed sections from hematoxylin and eosin-stained liver slides sampled from a cohort of 8 male rhesus macaques. We trained a deep learning model to classify the primates into previously identified classes of drinking behavior based on the slide images: non-drinkers, low drinkers (LD), binge drinkers (BD), heavy drinkers (HD), very heavy drinkers (VHD). Our model performed with high accuracy, distinguishing between drinking and non-drinking animals with 98% accuracy and with 94% accuracy when classifying LD versus VHD. Our results suggest that differential exposure to ethanol can have subtle, but identifiable, effects on liver tissue. This approach could provide a way to assess the impact of alcohol on liver tissue and contribute to models designed for automated detection of early signs of alcohol-related liver disease. — Track: CSCI-RTCB