Speaker
Mr
Abderrahim EL YAHYAOUY
(Faculté des Sciences, Université Ibn Tofail, Kenitra)
Description
Given the high prevalence of lung cancer, the desire of doctors to use artificial intelligence technologies in the medical field to save time and effort, and the impact of these technologies on people's lives, the need has emerged for highly accurate and reliable automated diagnostic models for segmenting chest CT scans. In this research, a novel auxiliary U-Net with a squeeze and excitation mechanism is proposed, where the loss is exploited at multiple levels using an auxiliary classifier in the bottleneck layer, while the squeeze and excitation block focuses on the most important features resulting from the convolutional layers. The proposed model achieved an accuracy of 98.56% and a Dice coefficient of 98.42% on the LUNA16 dataset.
Author
Mr
Abderrahim EL YAHYAOUY
(Faculté des Sciences, Université Ibn Tofail, Kenitra)
