Presentation
Time:
M.S. Thesis Defense: Yu Shi
Date:
9:30 am –
10:30 am
Zoom Room: ID: 944 697 0308
Target Audiences:
M.S. Thesis Defense: Yu Shi
Tuesday, July 25, 2023
9:30 AM
Zoom: 944 697 0308
“CT IMAGE SYNTHESIS TO ENHANCE PANCREATIC CANCER EARLY DETECTION”
Pancreatic ductal adenocarcinoma (PDAC) presents a critical global health challenge, as it is often detected at late stages with a weak response to current chemotherapy, resulting in a poor prognosis and high mortality rates compared to other solid cancers. Early detection is crucial for improving the 5-year survival rate, but specific screening tests for asymptomatic individuals remain elusive. Conventional imaging methods, like CT scans, have limited capabilities in early diagnosis due to the tumor’s subtle and heterogeneous nature. Recent advances in medical imaging and computational algorithms offer potential solutions.
Deep learning, particularly in the form of convolutional neural networks (CNNs), has demonstrated success in medical image analysis tasks, including classification and segmentation. The transition to 3D models has allowed the preservation of contextual information from adjacent slices, improving efficiency. However, the scarcity of clinical data for training remains a challenge. Data augmentation, generative adversarial networks (GANs), and cross-validation are potential techniques to address this limitation and improve model performance.
In this study, we propose a GAN-based tool for generating realistic 3D CT images of PDAC tumors and pancreatic tissue. PDAC’s challenging characteristics, such as iso-attenuating or hypodense appearance and lack of well-defined margins, make this task complex. The generated synthetic data can be used to augment the small clinical datasets, mitigating overfitting and enhancing the specificity and generality of the CNN classification model.
The development of this GAN-based model holds significant potential for improving the accuracy and early detection of PDAC tumors, which could have a profound impact on patient outcomes. Moreover, the model’s applicability to various tumor types offers valuable contributions to image processing models in medical imaging. Nonetheless, rigorous testing and validation on different datasets are essential to ensure the model’s effectiveness and generalization in clinical applications. Ultimately, this approach represents a promising avenue to address the pressing need for innovative and synergistic approaches to combat PDAC and other challenging medical conditions.
Committee:
Dr. Hongfeng Yu, Advisor
Dr. Chi Zhang
Dr. Juan Cui
Tuesday, July 25, 2023
9:30 AM
Zoom: 944 697 0308
“CT IMAGE SYNTHESIS TO ENHANCE PANCREATIC CANCER EARLY DETECTION”
Pancreatic ductal adenocarcinoma (PDAC) presents a critical global health challenge, as it is often detected at late stages with a weak response to current chemotherapy, resulting in a poor prognosis and high mortality rates compared to other solid cancers. Early detection is crucial for improving the 5-year survival rate, but specific screening tests for asymptomatic individuals remain elusive. Conventional imaging methods, like CT scans, have limited capabilities in early diagnosis due to the tumor’s subtle and heterogeneous nature. Recent advances in medical imaging and computational algorithms offer potential solutions.
Deep learning, particularly in the form of convolutional neural networks (CNNs), has demonstrated success in medical image analysis tasks, including classification and segmentation. The transition to 3D models has allowed the preservation of contextual information from adjacent slices, improving efficiency. However, the scarcity of clinical data for training remains a challenge. Data augmentation, generative adversarial networks (GANs), and cross-validation are potential techniques to address this limitation and improve model performance.
In this study, we propose a GAN-based tool for generating realistic 3D CT images of PDAC tumors and pancreatic tissue. PDAC’s challenging characteristics, such as iso-attenuating or hypodense appearance and lack of well-defined margins, make this task complex. The generated synthetic data can be used to augment the small clinical datasets, mitigating overfitting and enhancing the specificity and generality of the CNN classification model.
The development of this GAN-based model holds significant potential for improving the accuracy and early detection of PDAC tumors, which could have a profound impact on patient outcomes. Moreover, the model’s applicability to various tumor types offers valuable contributions to image processing models in medical imaging. Nonetheless, rigorous testing and validation on different datasets are essential to ensure the model’s effectiveness and generalization in clinical applications. Ultimately, this approach represents a promising avenue to address the pressing need for innovative and synergistic approaches to combat PDAC and other challenging medical conditions.
Committee:
Dr. Hongfeng Yu, Advisor
Dr. Chi Zhang
Dr. Juan Cui