Electronic Thesis and Dissertation Repository

Thesis Format



Master of Science


Computer Science

Collaborative Specialization

Artificial Intelligence


Unsupervised domain adaptation (UDA) is crucial in medical image analysis where only the source domain data is labeled. There is a lot of emphasis on the closed-set paradigm in UDA, where the label space is assumed to be the same in all domains. However, medical imaging often has an open-world scenario where the source domain has a limited number of disease categories and the target domain has unknown distinct classes. Also, maintaining the privacy of patients is a crucial aspect of medical research and practice. In this work, we shed light on the Open-Set Domain Adaptation (OSDA) setting on fundus image analysis while preserving the privacy concern. In particular, we step towards a source-free open-set domain adaptation where, without source data, the source model is utilized to facilitate adaptation to open-set unlabeled data by delving into channel-wise and local features for fundus disease recognition. In particular, considering the nature of the fundus images, we present a novel objective way in the adaptation phase to utilize spatial and channel-wise information to select the best source model for a target domain, even by considering the small inter-class variation between samples. Our approach has achieved state-of-the-art performance compared to other methods.

Summary for Lay Audience

Medical researchers often want to use data from one group of patients (source domain) to understand diseases in another group of patients (target domain). However, this can be difficult when the data from the target domain doesn’t have labels that tell us what the diseases are. In the past, researchers have tried to use labels from the source domain to understand the target domain, but this only works if the diseases in both groups are the same. But sometimes the diseases in the target domain are different from those in the source domain. To solve this problem, researchers have developed a method called open-set domain adaptation. This method allows them to use data from the source domain to understand the target domain without actually looking at the data from the source domain. This is important because it helps protect the privacy of patients. In this study, we apply source-free domain adaptation to understand various types of eye diseases.

Creative Commons License

Creative Commons Attribution 4.0 License
This work is licensed under a Creative Commons Attribution 4.0 License.