Saudi Cultural Missions Theses & Dissertations

Permanent URI for this communityhttps://drepo.sdl.edu.sa/handle/20.500.14154/10

Browse

Search Results

Now showing 1 - 3 of 3
  • ItemRestricted
    Generative AI for Mitosis Synthesis in Histopathology Images
    (University of Surrey, 2024-09) Alkhadra, Rahaf; Rai, Taran; Wells, Kevin
    Identifying mitotic figures has been established as an effective method of fighting cancer at its most vulnerable stage. Traditional methods rely on manual, slow, and invasive detection methods obtained from sectioned tissue samples to acquire histopathological images. Currently, Artificial Intelligence (AI) in oncology has produced a paradigm shift in the fight against cancer, also known as computational oncology. This is heavily reliant on the availability of mitotic figure datasets to train models; however, such datasets are limited in size, type, and may infringe on patient privacy. It is hypothesised that the potential of computational oncology can be realised by synthesising realistic and diverse histopathological datasets using Generative Artificial Intelligence (GenAI). This report demonstrates a comparison of Denoising Probabilistic Diffusion Models (DDPM) and StyleGAN3 in generating synthetic histopathology images, with mitotic figures. The MIDOG++ dataset containing human and canine samples with 7 types of tumours was used to train the models. Quality and similarity of generated and real images was evaluated using as Frechet Inception Distance (FID), Mean Square Error (MSE), Structural Similarity Index (SSIM), and Area Under the Curve (AUC) as a part of Receiver Operating Characteristic (ROC) study were incorporated. Our results suggests that the DDPM model is superior in terms of structural accuracy, however, StyleGAN3 capture the colour scheme better.
    19 0
  • Thumbnail Image
    ItemRestricted
    Ultrasound Nano-Scale Phase Change Contrast Agent for Hepatocellular Carcinoma Radiosensitization
    (Saudi Digital Library, 2023-08-05) Falatah, Hebah; Eisenbrey, John; Wheatley, Margaret
    The purpose of this study was to develop and characterize nano-scale phase change droplets less than 200 nm from commercially available ultrasound contrast agents and demonstrate their ability to enhance hepatocellular carcinoma (HCC) radiosensitization, which is critically needed to enhance the poor outcomes (two years survival < 50%) of current HCC radiotherapy treatments. Primary liver cancer is the third cause of cancer death worldwide with 906,000 new cases and 830,000 deaths annually. Of these, 75-85% of patients present with hepatocellular carcinoma (HCC), while the remaining 15-25% are intrahepatic cholangiocarcinoma and other types. Due to the late clinical presentation of the disease and treatment limitations of chemotherapy and immunotherapies, HCC has a poor prognosis. Localized radiotherapy in the early and mid-stages of HCC has shown some success in treatment response 25-50%. In such therapy, the hepatic artery supplying blood to the cancer is injected with radioisotope yttrium-90 (Y90), a beta particle emitter that provides localized radiation therapy. Another therapeutic option is external beam radiation (XRT) with MRI or CT guidance. XRT has been used cautiously in HCC treatment due to the radiosensitivity of liver tissue and technological limitations. Fractionated approaches are used to overcome the toxicity to the liver or radiation-induced liver diseases caused by high doses of radiation. The result of these limitations is that the overall five years survival for HCC patients in the United States is 20%, and the two years survival is less than 50%.Therefore, developing more effective HCC treatments is essential to improve patient outcomes. In recent years, researchers have been exploring a variety of radiosensitizers as a means of overcoming radiotherapy resistance. One promising radiotherapy enhancement mechanism is ultrasound-mediated microbubble destruction, which has been shown to sensitize solid tumors to radiotherapy through endothelial cell disruption in tumors. Ultrasound microbubbles have a diameter between 1 to 8 (micrometers) μm and consist of a high molecular weight gas encapsulated by a lipid, protein, or polymer shell. However, the relatively large size of the bubbles prevents them from passing into extravascular spaces, and as a result researchers have developed phase-change contrast agents (PCCAs). PCCAs contain a low boiling point such as -37oC liquid in place of the usual gas and can transition from the liquid to the gaseous state under external stimuli. This technology has been widely used in ultrasound medical imaging, vascular occlusion, and cavitation activity enhancement. The small diameters < 400 nm of these PCCAs allow them to diffuse and accumulate in solid tumors via the enhanced permeability and retention effect before the phase transition. Using ultrasound as an acoustic stimulus can provide local vaporization as well as cavitation of the resultant microbubble, thereby generating local force in tumor tissues. The development and characterization of sub-micrometer (< 200 nm diameter) phase change droplets from commercial ultrasound contrast agents, including their ability to decrease tumor vascularity and enhance cell apoptosis are described in addition to their ability to enhance HCC tumor radiosensitization.
    14 0
  • Thumbnail Image
    ItemRestricted
    Transforming Medical Image Segmentation with Enhanced U-Net Architectures and Adaptive Transfer Learning
    (Saudi Digital Library, 2023-05-25) Albishri, Ahmed; Yugyung, Lee
    Medical imaging has revolutionized healthcare by enabling accurate diagnosis, treatment planning, and monitoring of various diseases. Various modalities, such as computed tomography (CT), magnetic resonance imaging (MRI), and ultrasound, visualize diverse anatomical structures and pathological conditions. However, challenges arise in medical image segmentation due to increasing complexity, variability, noise, artifacts, and scarcity of annotated data. The advent of AI, particularly deep learning with Convolutional Neural Networks (CNNs), has facilitated significant advancements in medical image segmentation. U-Net, a prominent CNN architecture, provides accurate segmentation results with relatively low training samples due to its encoder-decoder structure with skip connections. In addition, transfer learning further mitigates limitations imposed by scarce labeled data. In this thesis, we develop innovative custom U-Net models with advanced building blocks and transfer learning strategies, such as AM-UNet for human brain claustrum segmentation from MRI scans, TLU-Net for organ and tumor segmentation from CT scans, and OCU-Net for oral cancer tissue segmentation from whole slide images (WSI) stained with Hematoxylin and Eosin (H&E). Furthermore, we introduce the "U-Framework", a comprehensive guide in designing and optimizing U-Net models. This framework encompasses key decisions related to architecture, transfer learning, module selection and fine-tuning, and evaluation strategies. Finally, by comparing our models with state-of-the-art approaches on benchmark datasets, we demonstrate their significant potential to contribute to medical image segmentation.
    30 0

Copyright owned by the Saudi Digital Library (SDL) © 2024