Residual Network of Residual Network: A New Approach to Improve Deep Learning Based Segmentation of Left Ventricle in MRI Cardiac Image

Publish Year: 1398
نوع سند: مقاله کنفرانسی
زبان: English
View: 435

متن کامل این Paper منتشر نشده است و فقط به صورت چکیده یا چکیده مبسوط در پایگاه موجود می باشد.
توضیح: معمولا کلیه مقالاتی که کمتر از ۵ صفحه باشند در پایگاه سیویلیکا اصل Paper (فول تکست) محسوب نمی شوند و فقط کاربران عضو بدون کسر اعتبار می توانند فایل آنها را دریافت نمایند.

  • Certificate
  • من نویسنده این مقاله هستم

این Paper در بخشهای موضوعی زیر دسته بندی شده است:

استخراج به نرم افزارهای پژوهشی:

لینک ثابت به این Paper:

شناسه ملی سند علمی:

IBIS09_015

تاریخ نمایه سازی: 19 اسفند 1399

Abstract:

Recently, Magnetic Resonance Imaging (MRI) has revolutionized the early detection of heart failure [1]. A vital step of this process is valid measurement of left ventricle properties from captured images, which seriously depends on their accurate segmentation [2]. Deep learning methods lead to better results than classic alternatives, but the gradient vanishing and exploding problems may hamper the effectiveness of deep neural networks in accurate segmentation [3]. In the proposed architectures Residual network and ROR (i.e. Residual network of Residual network) are engaged as a new substructure inside the U-net deep structure in two steps. The residual learning [4] is responsible to address the problem of accuracy saturation in U-net, thanks to its ability in jump over some layers which leads to reducing gradient flow disturbances.

Authors

Maral Zarvani

Faculty of Engineering, Alzahra University, Tehran, Iran

Sara Saberi Moghadam Tehrani

Faculty of Engineering, Alzahra University, Tehran, Iran

Seyed Vahab Shojaedini

Iranian Research Organization for Science and Technology, Tehran, Iran

Reza Azimi

Faculty of Engineering, Alzahra University, Tehran, Iran