Classification of liver CT scan images using deep semi-supervised learning methods

نوع: Type: thesis

مقطع: Segment: masters

عنوان: Title: Classification of liver CT scan images using deep semi-supervised learning methods

ارائه دهنده: Provider: atefe eslamian

اساتید راهنما: Supervisors: Dr. Hassan Khotanlou- Dr. Muharram Mansouri Zadeh

اساتید مشاور: Advisory Professors:

اساتید ممتحن یا داور: Examining professors or referees: Dr. Reza Mohammadi - Dr. Mahdi Sakhaei Nia

زمان و تاریخ ارائه: Time and date of presentation: 2024

مکان ارائه: Place of presentation: seminar3

چکیده: Abstract: Diagnosis and classification of focal lesions liver from CT scan images is an important challenge for doctors due to its importance in the human body. This was done with the aim of improving the classification accuracy of focal lesions liver using neural networks and fine-tuning. In this research, three deep neural networks ResNet, AlexNet and EfficientNet were used to classify CT scan images with improved contrast from NC, ART and PV phases. Due to the limited number of images, fine-tuning is used to improve the performance of the models. The results showed that the ResNet model with fine-tuning and cross-entropy loss function had the best performance compared to other advanced models and methods. This research showed that it uses deep neural networks and can be more accurate. . Increasing the variety and number of images in the dataset, investigating new preprocessing methods, evaluating other neural network architectures, providing interpretation methods for training models, developing ethical systems to assist physicians, and investigating issues related to the use of artificial intelligence in diagnosing and treating diseases can have suggestions that this research be successful