자료유형  

 학위논문

Control Number  

0016932972

International Standard Book Number  

9798379610869

Dewey Decimal Classification Number  

616

Main Entry-Personal Name  

Zhang, Haoyue.

Publication, Distribution, etc. (Imprint  

[S.l.] : University of California, Los Angeles., 2023

Publication, Distribution, etc. (Imprint  

Ann Arbor : ProQuest Dissertations & Theses, 2023

Physical Description  

1 online resource(157 p.)

General Note  

Source: Dissertations Abstracts International, Volume: 84-12, Section: B.

General Note  

Advisor: Arnold, Corey W.

Dissertation Note  

Thesis (Ph.D.)--University of California, Los Angeles, 2023.

Restrictions on Access Note  

This item must not be sold to any third party vendors.

Summary, Etc.  

요약Acute ischemic stroke (AIS) is a cerebrovascular disease caused by deceased blood flow in the brain. Treatment of AIS is heavily dependent on the time since stroke onset (TSS), either by clock time or tissue time. AIS treatments aim to restore blood flow in the stroke-affected area to minimize infarction. Current clinical guidelines recommend thrombolytic therapies (e.g. Intravenous(IV) or Intra-arterial (IA) tissue Plasminogen Activator (tPA) for patients presenting within 4.5 hours of TSS and Mechanical Thrombectomy (MTB) (e.g. surgical removal of the clot) for patients with TSS up to 24 hours. This research attempts to use both CT and MRI to predict the eligibility of AIS patients and their response to treatment while addressing several challenges in neuroimaging and AIS diagnosis in clinical settings using novel machine learning and deep learning approaches. A Self-supervised Learning approach, called intra-domain task-adaptive transfer learning, is the first proposed to predict TSS using limited training data. A hybrid transformer model that utilizes spatial neighborhood information in brain regions is proposed to predict MTB success. A pure transformer and a specifically designed Masked Image Model are developed to predict Large Vessel Occlusion (LVO). Last, a transformer-based super-resolution framework is proposed to generate synthesized thin-slice images from thick-slice images. Together, these models demonstrate the effectiveness of the attention mechanism and the usefulness of self-supervised learning for clinical deep learning applications given the limited data resources compared to natural images.

Subject Added Entry-Topical Term  

Medical imaging.

Subject Added Entry-Topical Term  

Biomedical engineering.

Index Term-Uncontrolled  

Computer vision

Index Term-Uncontrolled  

Deep learning

Index Term-Uncontrolled  

Machine learning

Index Term-Uncontrolled  

Acute ischemic stroke

Index Term-Uncontrolled  

Time since stroke onset

Index Term-Uncontrolled  

Mechanical Thrombectomy

Added Entry-Corporate Name  

University of California, Los Angeles Bioengineering 0288

Host Item Entry  

Dissertations Abstracts International. 84-12B.

Host Item Entry  

Dissertation Abstract International

Electronic Location and Access  

로그인을 한후 보실 수 있는 자료입니다.

Control Number  

joongbu:643871