Applied image processing techniques in video laryngoscope for occult tumor detection. (January 2020)

• Record Type: Journal Article
• Authors: Jeffrey Kuo, Chung-Feng; Li, Yu-Ching; Weng, Wei-Han; Pinos Leon, Kathya Belen; Chu, Yueng-Hsiang
• Is Part Of: Biomedical signal processing and control. Volume 55(2020)
• Publication Date: 2020-01
• Subjects: Video laryngoscope -- Color space conversion -- Image enhancement -- Threshold selection -- Region-based segmentation -- Dynamic information-based segmentation; Signal processing -- Periodicals; Biomedical engineering -- Periodicals; Signal Processing, Computer-Assisted -- Periodicals; Image Processing, Computer-Assisted -- Periodicals; Biomedical Engineering -- Periodicals; 610.28
• DOI: 10.1016/j.bspc.2019.101633
• Languages: English
• ISSNs: 1746-8094

Abstract

Highlights: Locating and circling the tumor of laryngeal part of pharynx in video laryngoscope by automatic detection. The laryngeal part of pharynx was divided into radix linguae, hypopharynx, epiglottis and glottis regions for circling the tumor. The dynamic laryngoscope videos with larynx tumor was used for clinical test for automatic laryngeal tumor detection. According to doctor's validation, the diagnosis efficiency and accuracy is increased effectively. Abstract: This study implemented image processing techniques for detecting small tumors in a video laryngoscope, as small tumor lesions often appear in the video for only a very short duration, and can even be missed. The aim of this research was to develop a larynx tumor detection system that could accurately mark the position of a tumor. In the laryngoscope images, the glottis and epiglottis images have apparent geometric and brightness features for image segmentation. According to the relevant positions of the two regions, the laryngeal part of the pharynx is divided into the radix linguae, hypopharynx, epiglottis, and glottis regions for marking the tumor. It could be adapted in laryngoscope images and effectively segmented the tumor contour in different locations. The tumor-correlated image processing procedure could then be implemented, and the tumor contour could be segmented. This study analyzed video laryngoscope data from 359 confirmed laryngeal cancer films, and the methods used were centered on image color space conversion, image enhancement, threshold selection, region-based segmentation, and dynamic information-based segmentation. This study demonstrated that the proposed system is a powerful tool for video laryngoscope analysis, especially for small and easily unnoticed tumors. The system identified and marked laryngeal tumors instantly and automatically with an accuracy of 97%. Overall, the system could ease and speed up the process of tumor detection as well as minimize the possibility of ENT doctors missing small or early-stage tumors. By being able to give timely and opportune treatment, patient quality of life could be improved.