Scholarlink

The proposed research fuses IoMT, Few-Shot Learning and ConvLSTM for the creation of an intelligent healthcare robotic system to enhance surgical guiding with customization. Through live video capture, and less errors lead to greater accuracy for better decision-making in recognizing surgical anomalies and predicting the movement of tools which results in fewer manual corrections necessary for a cleaner outcome. Objective: Robotic system design to integrate IoMT for real-time data, FeSL for anomaly detection with few data, and ConvLSTM for temporal prediction in order to maintain higher surgical precision, to reduce human error rates and to enhance safety as well as result of patients. Methods: It have applied anisotropic kernel regression diffusion filter as a technique to pre-process the surgical video data acquired through IoMT devices. ConvLSTM is utilized to forecast the tool movement which captures the temporal dependencies while Few-Shot Learning is employed for OOD detection using scarce data. Post-processing offers real-time feedback for surgeons during procedures. Results: The system predicts 99.41% accuracy of surgical tools movements and detects anomaly, which outperforms standard models. It works to reduce errors and offer real-time feedback thus helping in making better decisions for a physician and delivering advanced surgical outcomes at the same time. Conclusion: The combination of IoMT, Few-Shot Learning, and ConvLSTM increased the surgical precision in real time. It minimizes errors, provides instant feedback and hence increases patient safety during complex surgeries.

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

An Intelligent Healthcare Robotics System Leveraging IoMT, Few-Shot Learning, and Convolutional Long Short-Term Memory (ConvLSTM) for Accurate Surgical Assistance

  • 章节

  • 发表日期：

    2025年

查看全文
  • Durai Rajesh Natarajan,
  • Swapna Narla,
  • Sai Sathish Kethu,
  • Dharma Teja Valivarthi,
  • Sreekar Peddi,
  • N. Purandhar

摘要

The proposed research fuses IoMT, Few-Shot Learning and ConvLSTM for the creation of an intelligent healthcare robotic system to enhance surgical guiding with customization. Through live video capture, and less errors lead to greater accuracy for better decision-making in recognizing surgical anomalies and predicting the movement of tools which results in fewer manual corrections necessary for a cleaner outcome. Objective: Robotic system design to integrate IoMT for real-time data, FeSL for anomaly detection with few data, and ConvLSTM for temporal prediction in order to maintain higher surgical precision, to reduce human error rates and to enhance safety as well as result of patients. Methods: It have applied anisotropic kernel regression diffusion filter as a technique to pre-process the surgical video data acquired through IoMT devices. ConvLSTM is utilized to forecast the tool movement which captures the temporal dependencies while Few-Shot Learning is employed for OOD detection using scarce data. Post-processing offers real-time feedback for surgeons during procedures. Results: The system predicts 99.41% accuracy of surgical tools movements and detects anomaly, which outperforms standard models. It works to reduce errors and offer real-time feedback thus helping in making better decisions for a physician and delivering advanced surgical outcomes at the same time. Conclusion: The combination of IoMT, Few-Shot Learning, and ConvLSTM increased the surgical precision in real time. It minimizes errors, provides instant feedback and hence increases patient safety during complex surgeries.