Kyungpook National University Researchers Update Understanding of Engineering (Genetic Algorithm Guided Image Channel Selection for Skin Lesion Segmentation).

Předmět:	GENETIC algorithms CONVOLUTIONAL neural networks GENETIC engineering ARTIFICIAL intelligence SKIN imaging
Zdroj:	Cancer Weekly; 10/15/2024, p527-527, 1p
Abstrakt:	Researchers from Kyungpook National University in South Korea have developed a new approach to skin lesion segmentation using a Genetic Algorithm (GA) guided feature selection. The researchers replaced the ResNet block with a custom block in the Modified TMU-Net architecture to improve adaptability to variable input sizes. The dual-pipeline design combines transformers for global representations and convolutional neural networks for local contextual representations. The GA is used to optimize image channel selection, resulting in enhanced segmentation. The proposed model achieved successful outcomes when trained on the ISIC 2017 dataset. [Extracted from the article]
Databáze:	Complementary Index
Externí odkaz:	Zobrazit plný text záznamu