SAM-based laryngeal tumor detection method and device, electronic equipment and medium

CN119359619APending Publication Date: 2025-01-24SHENZHEN TECH UNIV +1
0 Cites 3 Cited by

Patent Information

Application Number
CN202411256118.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-09
Publication Date
2025-01-24

AI Technical Summary

Technical Problem

Existing methods for detecting laryngeal cancer often fail to distinguish the images from the surrounding normal mucosa when the lesion is superficial or the tumor is small, leading to missed detections and low accuracy.

Method used

A SAM-based laryngeal tumor detection method is adopted. The global endoscopic image is processed by the JCLNet network, the local image is segmented by the lesion localization module, and multi-scale feature extraction and convolutional attention module processing are performed. The global and local features are fused for classifier prediction.

Benefits of technology

It improves the accuracy of laryngeal tumor detection, especially the recognition rate of benign tumors, and enhances the capture of detailed features of lesions and the utilization of global contextual information.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119359619A_ABST
    Figure CN119359619A_ABST
Patent Text Reader

Abstract

The invention provides a laryngeal tumor detection method and device based on SAM, electronic equipment and a medium, and can be applied to the technical field of tumor detection. According to the method, the global endoscope image is processed by setting a JCLNet network, the global endoscope image is segmented through a focus positioning module based on SAM guidance in the processing process to obtain a local image, and then multi-scale feature extraction is performed on the global endoscope image and the local image to obtain a first global feature and a first local feature; processing the first global feature and the first local feature through a convolution attention module to obtain a second global feature and a second local feature, and fusing the second global feature and the second local feature; and inputting the second global feature, the second local feature and the fusion feature into a classifier for prediction to obtain a laryngeal tumor detection result, thereby improving the accuracy of the laryngeal tumor detection result by using local focus information.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Cited By

  • Neural network-based recurrent laryngeal nerve real-time identification and dynamic tracking method and system

    CN120526290A

  • Multi-part identification method and system for nose and throat endoscope image

    CN120808053A

  • Tumor classification method and device, computer equipment, medium and product

    CN120878137A