Central Asian Research Index

Muhammad al-Xorazmiy avlodlari, 2026

Multi-Metric Approach for Comprehensive Quality Assessment of Endoscopic Image Training Datasets

Temurbek Kuchkorov, Nazokat Sabitova, Avazbek Hasanov

Tashkent University of Information Technologies named after Muhammad al-Khwarizmi (TATU)

Аннотация · Abstract

This study is devoted to the evaluation and analysis of endoscopic image quality. The paper identifies the primary degradation factors affecting image quality, including blur, motion artifacts, uneven illumination, noise, and color distortions. To assess these degradations, no-reference Image Quality Assessment (NR-IQA) metrics were applied, including Laplacian variance, gradient energy, RMS contrast, and entropy measures. The proposed model is optimized for real-time implementation and demonstrates a high correlation with subjective expert evaluations. A large-scale endoscopic image training o’quv tanlanam was utilized to evaluate image quality indicators, thereby enhancing the model’s detection performance under diverse clinical conditions and anatomical variations. The scientific contribution of this work lies in the development of a hybrid architecture that integrates the strengths of conventional assessment methods, resulting in a robust, high-accuracy image quality evaluation system capable of real-time monitoring in clinical practice.

Сведения · Details