Evaluation of Stochastic Gradient Descent Optimizer on U-Net Architecture for Brain Tumor Segmentation DOI Creative Commons

Purwono Purwono,

Iis Setiawan Mangkunegara

International Journal of Robotics and Control Systems, Год журнала: 2023, Номер 3(3), С. 588 - 598

Опубликована: Авг. 18, 2023

A brain tumor is a type of disease that quite dangerous in the world. This one main causes human death and has high risk recurrence. There are several types locations such as edema, necrosis to elevation. Segmenting location this important do support faster recovery efforts. The Convolutional Neural Network (CNN) algorithm, which part deep learning method, can be an alternative segmentation effort. U-Net architecture CNN algorithm specifically works on medical image segmentation. study experimented build special for had been optimized with SGD. data used BraTS2020O contains collection MRI data. optimization aims improve performance U-net segmenting images. results show SGD carried out succeeded providing better than previous studies. seen from value obtained at 0.9879. accuracy indicates increase High SGD-optimized model good prediction performance.

Язык: Английский

Evaluation of Stochastic Gradient Descent Optimizer on U-Net Architecture for Brain Tumor Segmentation DOI Creative Commons

Purwono Purwono,

Iis Setiawan Mangkunegara

International Journal of Robotics and Control Systems, Год журнала: 2023, Номер 3(3), С. 588 - 598

Опубликована: Авг. 18, 2023

A brain tumor is a type of disease that quite dangerous in the world. This one main causes human death and has high risk recurrence. There are several types locations such as edema, necrosis to elevation. Segmenting location this important do support faster recovery efforts. The Convolutional Neural Network (CNN) algorithm, which part deep learning method, can be an alternative segmentation effort. U-Net architecture CNN algorithm specifically works on medical image segmentation. study experimented build special for had been optimized with SGD. data used BraTS2020O contains collection MRI data. optimization aims improve performance U-net segmenting images. results show SGD carried out succeeded providing better than previous studies. seen from value obtained at 0.9879. accuracy indicates increase High SGD-optimized model good prediction performance.

Язык: Английский

Процитировано

1