Real-Time Thyroid Nodule Precise MIoT Leveraging Quad-Node Raspberry Pi Clusters and NVIDIA Jetson Nano for Edge AI

This study introduces the thyroid nodule segmentation grid search based local patch learning (GS-LPL) network as an effective IoT solution for real-time, precise thyroid cancer segmentation. Utilizing the Turing PI and NVIDIA Jetson Nano, we developed a quadruple node distributed computing cluster for real-time cancer segmentation through parallel and distributed computing. GS-LPL features a lightweight neural network, optimized for mini-supercomputers, enabling efficient model training and deployment. The NVIDIA Jetson Nano supports real-time high-definition video processing, ideal for low-powered intelligent video analytics like Network Video Recorders (NVR). Our model is also deployable on mobile devices with limited resources. Experimental results validate our model's fast, efficient, and accurate performance in thyroid nodule cancer segmentation.