Table 1 Summary of existing studies on laryngeal cancer diagnosis.
Ref | Techniques | Metrics | Findings |
|---|---|---|---|
Inceptionv3, deep belief network, aquila optimization algorithm | Accuracy, precision, recall, F-score | The LCDC-AOADL technique achieves improved laryngeal cancer detection and classification | |
Deep EL, CNN | Benchmark metrics | This study detects laryngeal cancer from endoscopic pictures, addressing image variability and interference challenges | |
CNN, classification and regression tree | Standard metrics | This paper improves the accuracy of glottic cancer diagnosis by combining laryngeal image and voice data from PNUH and validating PNUH and PNUYH datasets | |
DL-based mask R-CNN | Accuracy, precision, recall, F1-score | The study introduces a model for real-time identification of laryngeal cancer and its symptoms in patient screening using the ImageNet dataset | |
DenseNet201, TL, recursive feature elimination with random forest, support vector machine, and k-nearest neighbor | Standard metric | Using a benchmark dataset, the study develops a model that outperforms existing methods in distinguishing SCC tissues, healthy tissues, and precancerous tissues | |
DAN, U-Net | Accuracy, sensitivity, precision, recall | The study introduces a model achieving robust end-to-end segmentation of the glottal area, surpassing benchmarks in laryngeal image analysis using the in-house dataset | |
Multiscale cross-layer adaptation and feature denoising module | Benchmark metrics | The study introduces a model to achieve superior preservation of image structure and quality using white light and narrow-band laryngeal image datasets | |
RANT, CNN, RRM, ConvCRF | mIoU, mDSC | The study introduces a technique for accurate multi-organ segmentation in electronic laryngoscope images, surpassing benchmarks in clinical application for larynx segmentation using two laryngoscopy datasets | |
Farmland fertility algorithm, QO-based learning | Standard metrics | The study demonstrated superior results across benchmark functions and real-world engineering problems under a benchmark dataset | |
AVOA, QRG | Standard metrices | The study proposes a mechanism to improve solution diversity and avoid local optima, achieving robust performance on large-scale image datasets compared to other optimization algorithms | |
IAS, CCF | Benchmark metrics | The study introduces an asymmetric clustering approach enhanced with ten chaotic maps and an intra-cluster summation fitness function, demonstrating superior performance using the COVID-19 dataset |
