This paper aims to develop a novel deep learning model integrating the Internet of Medical Things (IoMT) for cell detection in biomedical images. Given the significance of real-time acquisition and processing of vast amounts of image data in biomedical field, and the existing challenges in missing detection and locating accuracy of automatic cell detection in human cell images, there is a crucial need to propose an effective solution. However, current methods may not be sufficient in fully leveraging the potential of IoMT and accurately detecting cells in complex biomedical images.
This paper proposes a novel deep learning model, CAM-BCFF, for cell detection in biomedical images based on the IoMT framework. The proposed CAM-BCFF uses a coordinate attention mechanism combined with the FReLU activation function to locate the small-scale cell image. To further improve the locating accuracy of the human cell images, this proposed CAM-BCFF introduces bidirectional cross-feature fusion, which uses skip connections to fuse deep-level feature maps and shallow-level feature maps. Meanwhile, the multi-scale feature map mechanism is used in the network detection layer of CAM-BCFF, combined with Gaussian mixture clustering to generate prior boxes, for handling cell image of different scales.
Extensive experiments are conducted to evaluate the performance of the proposed IoMT-based CAM-BCFF model. The results demonstrate that the proposed model can accurately detect the cells in biomedical images and performs better than the existing state-of-the-art methods.
In this study, this paper proposes a new deep learning model, CAM-BCFF, for cell detection in biomedical images within the IoMT environment.
