Enhancing Video Laryngoscopy and Metaheuristic Methods for Improved Navigation With Full and No‐Reference Quality Metrics

The use of laryngoscopy is widespread in the field of airway management, providing improved visual clarity and procedural support in various clinical situations. Endotracheal intubation (ETI) is commonly employed during general anesthesia and the subsequent recovery period. This work employs a focused methodology to investigate algorithms for improving video quality in real‐world clinical situations and to build algorithms for detecting glottis and larynx in video laryngoscopy. The main objective is to mechanize the capture and archiving of video‐assisted intubation. The heuristic strategy involves capturing the ETI video and subsequent preprocessing utilizing 2‐D FIR filtering, adaptive histogram equalization, Gaussian filtering with varying sigma values, and artificial intelligence improving techniques. Video quality was assessed using no‐reference: NIQE, BRISQUE and full‐reference: PSNR metrics. Additionally, a video frame's quality is calculated using XAI. LoG, contrast, edge density, entropy and brightness are taken into account for the video quality score, with weights taken into account as per [0.35, 0.25, 0.2, 0.1, 0.1]. Subsequently, a suitable model was developed to ascertain the optimal approach for enhancing video quality during ETI. A lower NIQE coefficient leads to improved visibility. Optimizing the video quality selection can enhance the success rate of challenging intubations on the initial attempt. Next, the data will be analyzed for application in educational, training, and simulation settings.