Loading...

Weakly Supervised Mammalian Cell Segmentation in Microscopic Images

Mahmoodinia, Erfan | 2021

228 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 54660 (19)
  4. University: Sharif University of Technology
  5. Department: Computer Engineering
  6. Advisor(s): Rabiee, Hamid Reza; Rohban, Mohammad Hossein
  7. Abstract:
  8. Due to the overall progress in the processing of imaging tissue cells, the identification and diagnosis of complex diseases using machine learning methods has become very important. Recognizing cell characteristics such as size, shape, and chromatin design is essential in determining cell type, which can be achieved through learning methods such as deep network training. Finding the nucleus or cytoplasm of cells in medical images is a small but significant part of a long process of diagnosing and treating diseases. Today, artificial intelligence has rushed to the aid of experts in this field and has increased the speed and accuracy of experts in finding these cells and their nuclei. This dissertation aims to deal with segmentation with low surveillance of microscopic images such as fluorescent images. The purpose of segmenting these images is to identify the nucleus in different cell types. The visual data used to determine the core of the Kaggle Data Science Bowl 2018 is now available to the public. Using low-surveillance and self-supervised networks reduces the cost of data annotation and network retraining for different problem areas. Extracting features from insufficient unlabeled data is an essential step in self-supervised methods. The use of contrastive learning methods helps us to extract important features of cellular and medical images from unlabeled data. In recent years, robust networks in the field of data classification have been introduced. With fundamental changes in these networks, we can improve network accuracy for medical image segmentation by up to 10%. Also, considering only 10% of the available data compared to fully supervised networks with 100% of the data, an improvement of 1.5% accuracy has been observed, improving the process of cell nucleation detection
  9. Keywords:
  10. Unsupervised Learning ; Image Segmentation ; Convolutional Neural Network ; Self-Supervised Learning ; Disease Diagnosis ; Artificial Intelligence

 Digital Object List

 Bookmark

No TOC