The role of generative adversarial networks in bioimage analysis and computational diagnostics.

Computational technologies can contribute to the modeling and simulation of the biological environments and activities towards achieving better interpretations, analysis, and understanding. With the emergence of digital pathology, we can observe an increasing demand for more innovative, effective, and efficient computational models. Under the umbrella of artificial intelligence, deep learning mimics the brain’s way in learn complex relationships through data and experiences. In the field of bioimage analysis, models usually comprise discriminative approaches such as classification and segmentation tasks. In this thesis, we study how we can use generative AI models to improve bioimage analysis tasks using Generative Adversarial Networks (GANs). For that purpose, several studies were conducted. The first study is on domain translation, where we proposed a digital pathology system that can detect and quantify fibrosis in Hematoxylin and Eosin-stained digital slides. The proposed system features a comprehensive machine learning pipeline that includes conditional GANs based translation model, whole slide image registration algorithm, and color-based fibrosis detection module. In the second study, we propose a novel GANs-based model that reconstructs the 3D appearance of the tissue from the available interleaved tissue slices. The proposed model has a sandwich-shaped generator that utilizes a transfer learning strategy to learn the initial parameters from MRI domain before it starts training on the microscopic histology images. Deploying the proposed systems into the digital pathology workflow can improve the efficiency in terms of the processing time, labor cost and can improve diagnostic accuracy