Texas Digital Library
UT Southwestern Medical Center Institutional Repository (University of Texas)Not a member yet
10274 research outputs found
Sort by
Advancing Radiotherapy Treatment Through Artificial Intelligence-Driven Approaches
Get PDFRadiotherapy is a critical treatment regimen for cancer patients. Modern radiotherapy aims to deliver precise radiation doses to specific tumor target volumes while minimizing exposure to surrounding healthy tissues and organs. To achieve precision and high treatment quality, accurate target delineation, streamlined workflows, and timely follow-up are necessary. However, the current clinical standard of radiotherapy has limitations that need to be addressed. These include the slow manual delineation of targets and organs-at-risk, a lack of tools for early prediction of treatment outcomes, and the potential radiation toxicity and side effects. Fortunately, ultra-high dose rates (FLASH) irradiation emerges as a promising new modality of radiotherapy that has the potential to reduce the radiation toxicity to surrounding normal tissues while maintaining tumor control. However, the clinical translation of FLASH is challenging, and precise quantification of radiobiology is urgently needed.
To address these above limitations and needs, this dissertation aims to develop artificial intelligence (AI)-driven techniques to advance current radiotherapy treatment. Recent breakthroughs in AI, including mathematical algorithms and high-performance computing technologies, have led to transformative impacts in healthcare. Various studies have shown significant improvements in the quality and efficiency of image segmentation and treatment outcome analysis with AI approaches. Therefore, this dissertation specifically focuses on leveraging the power of AI to improve the target delineation, predict treatment outcomes, and assist the clinical translation of FLASH in the following three parts:
(1) Streamlining and standardizing Stereotactic Radiosurgery (SRS) workflow with AI. In this part, an AI-driven auto-segmentation and labeling platform was firstly developed for SRS patients with multiple brain metastases (mBMs). This platform can automatically segment out mBMs and auto-label each segmentation with an atlas label in high accuracy compared with manual contours and labels. Secondly, a deep-learning and radiomics ensemble classifier was developed for the false positive reduction in the raw mBMs segmentation, to improve the mBMs detection specificity while maintaining a promising sensitivity.
(2) Effective SRS management through AI predictions. In this part, three AI techniques for SRS treatment outcome prediction were developed including an ensemble learning model for glioma patients overall survival prediction, an unsupervised structure learning model for mBMs SRS treatment response modeling, and an AI model for mBMs patients post-SRS neurocognitive decline prediction, to assist the SRS treatment decision-making and to improve the treatment quality.
(3) Quantitative FLASH radiotherapy with AI. In this part, an AI model was developed to estimate the equivalent conventional (CONV) dose of FLASH irradiation based on tissue histological images. This developed model can accurately estimate the equivalent dose in CONV irradiation and indicates that deep learning can be potentially used to assess the equivalent dose of FLASH irradiation to normal tissue to accelerate its clinical implementation.
In summary, this dissertation presents novel AI technique developments that could potentially address the current needs in radiotherapy treatment. Moreover, the proposed approaches can be transferred to different tumor sites and radiotherapy procedures, thus generating broader clinical impact
Using outliers to unlock autoimmunity
No full textDetailed formal protocol with illustrations and extensive bibliography.A recording of the protocol presentation is available on UT Southwestern’s Mediasite. Note: Access to the video is restricted to authorized UT Southwestern users only.UT Southwestern--Internal Medicin
Genotype-Phenotype Heterogeneity Among Patients with Lipodystrophy Harboring Rare POLD1 Variants
No full textThe attached files include supplementary figures and tables. This submission meets the Extended Data Sets and Supplemental Materials requirements that are included in author guidelines for the Journal of Clinical Endocrinology & Metabolism (Print ISSN 0021-972X, Online ISSN 1945-7197).An updated, combined file was provided and uploaded in February 2026; the original files have been preserved for archival purposes
Dual Receptor Uptake of Low-Density Lipoprotein Docosahexaenoic Acid (LDL-DHA) Elicits Differential Sensitivity Across Breast Cancer Subtypes
Get PDFWe proposed to treat breast cancer with a novel nanomedicine previously shown to be effective in the treatment of primary liver tumors in rats. This nanoparticle is comprised of the low-density lipoprotein which acts as a carrier for the omega-3 fatty acid docosahexaenoic acid (DHA). The LDL-DHA nanoparticle has demonstrated cancer selective cytotoxicity with little ill effects on healthy cells. Our research began by selecting a panel of breast cancer cell lines representative of the various subtypes of clinical breast cancer. This panel was screened for expression of LDL and SR-B1 receptors, in addition LDL nanoparticle uptake kinetics was evaluated in each cell line. Thereafter, the cytotoxicity of LDL-DHA nanoparticles was assessed across the panel of breast cancer cells. Animal experiments showing efficacy of LDL-DHA nanoparticle are currently in progress, with the expectation that LDL-DHA treatment of tumors will induce significant tumor necrosis. Our current data suggests that our LDL-DHA nanoparticles are avidly taken up by breast cancer cells, with triple negative breast cancer cells showing some of the highest uptake, mediated by LDLR and SR-B1. LDL-DHA has a higher affinity for LDLR compared to SR-B1 with both receptors being able to mediate uptake. Concurrently, we showed that there is a reciprocal relationship between LDLR and SR-B1, where one receptor compensates for loss of the other to maintain cholesterol homeostasis. Thus, ensuring uptake of LDL-DHA nanoparticles in breast cancer cells. Moreover, LDL-DHA nanoparticles elicit significant cytotoxicity with triple negative breast cancer subtypes being most sensitive to LDL-DHA. As such, in vivo and clinical use of this nanomedicine is anticipated to be effective for metastatic breast cancer tumors with potentially more utility in triple negative breast cancer
Targeting and Modeling Immune Resistance in Lung Cancer
No full textThe general metadata -- e.g., title, author, abstract, subject headings, etc. -- is publicly available, but access to the submitted files is restricted to UT Southwestern campus access and/or authorized UT Southwestern users.Lung cancer causes most cancer-associated death both in U.S. and worldwide. Immune checkpoint blockade has achieved durable therapeutic effects in some lung cancer patients. The mechanism of immune escape in lung cancer merits further investigation to improve the efficacy of immune checkpoint blockade and other treatments. This thesis is focused on targeting innate immune resistance in small cell lung cancer (SCLC) and modeling human non-small cell lung cancer (NSCLC) in genetically engineered mouse models (GEMMs). I observed that SCLC escapes from innate immune surveillance by down-regulating NK cell-activating ligands. Histone deacetylase (HDAC) inhibitor could restore expression of NK cell-activating ligands and trigger anti-tumor immunity in vivo. HDAC inhibitors can be potential therapeutics for SCLC. Besides, I generated a NSCLC GEMM with high tumor mutational burden (TMB) by incorporating PoleP286R. PoleP286R GEMM with wildtype p53 is sensitive to immune checkpoint blockade (ICB). Loss of p53 and tumor heterogeneity contributed to immune escape in this high TMB mouse model. This ICB-sensitive, high TMB model can be utilized to explore novel immunotherapy and study novel mechanism of immune resistance of NSCLC
Dual-Pathway Training Model
Get PDFThe author submitted this entry in the Creative Non-Fiction category (Amateur division) for the 2025 On My Own Time™ (OMOT) Art Show.As I enter my final year of medical school, I think about alternative training models that may better serve learners, providers, and the patients that they care for
A Marathon Below
Get PDFThe author submitted this entry in the AI Literature category (Amateur division) for the 2025 On My Own Time™ (OMOT) Art Show.The specific prompt that the author submitted to the AI system to generate the story is available as a separate document.On a long run that felt like it would never end, I dictated the idea for a story onto my phone. When I got home, the text was all scrambled, confused, and hard to decipher. I asked ChatGPT to help me put my words in order, with this story as a result
The Rest Stop
Get PDFThe author submitted this entry in the AI Literature category (Amateur division) for the 2025 On My Own Time™ (OMOT) Art Show.The specific prompt that the author submitted to the AI system to generate the story is available as a separate document.This work received a First Place Award in the "AI Literature" category in the 2025 OMOT show.Driving through New Mexico, the idea for a short story came to mind. I used a speech-to-text app to capture my scrambled imagination and thoughts. At a roadblock about how to decipher these ideas, I used AI to help organize my story
Decoding Sequence Determinants of [alpha]-Synuclein Aggregation
No full textThe general metadata -- e.g., title, author, abstract, subject headings, etc. -- is publicly available, but access to the submitted files is restricted to UT Southwestern campus access and/or authorized UT Southwestern users.α-synuclein (aSyn) is a small presynaptic protein (14 kDa, 140 aa) involved in synaptic processing and neurotransmitter release, which abnormally aggregates to form intraneuronal inclusions called Lewy bodies and neurites or glial cytoplasmic inclusions, the hallmark features of neurodegenerative diseases collectively known as synucleinopathies: Parkinson's disease, transitional and diffuse Lewy body disease, and multiple system atrophy. It is proposed that aSyn behaves like a prion, which misfolds and propagates its pathological conformation inside the cell by converting the naïve endogenous aSyn in a process called "seeding." These aggregates can further escape the cell and be taken up by neighboring cells to propagate the misfolded conformation through a process called "amplification." These events can be reproduced in a simple cellular model in vitro and in animal models in vivo. Despite advanced research on this process, the mechanisms of aSyn aggregation and seeding are not fully understood. For this thesis project, I performed a saturation mutagenesis screen to study sequence determinants of aSyn aggregation. This approach let us identify inhibitory and aggregation enhancing regions at the single residue resolution. We discovered several aSyn domains that are crucial for its aggregation and single residues outside the core that control aggregation propensity. Truncation analysis found that aSyn termini inhibit fibril formation. Based on these results we designed new, high-sensitivity biosensors to detect aSyn aggregates in cells and brain tissue. aSyn monomer crosslinking coupled with mass spectrometry revealed a cluster of possible intramolecular electrostatic interactions which could be the origin of aSyn aggregation propensity. NMR spectroscopy confirmed our hypothesis of existing N-terminal local structure which may modulate aSyn aggregation propensity through interactions with the aSyn hydrophobic core. Our study advances the current knowledge about the mechanism of aSyn aggregation and provides new tools which could be useful for diagnosis of synucleinopathies and further research using human brain derived material
Racial Socialization Intervention for Black Children with Experiences of Racialized Trauma: Training Development and Pilot Among Community-Based Clinicians
Get PDFThe frequency of Black children's exposure to race-based traumatic stress coupled with the dearth of appropriate education to prepare clinicians to effectively intervene necessitates development of specialized training. Training should increase the capacity of behavioral health providers to deliver culturally sensitive care that centers the cultural knowledge and practices of the Black community. Following a review of the literature and discussions with clinicians and clinical administrators who treat Black children, an existing training focused on racial socialization (RS; the process by which many Black caregivers socialize Black children to what it means to be a member of the racial group) was adapted for community-based clinicians. The training was piloted with 75 English-speaking clinician-participants and trainees recruited from local community organizations in an urban setting. Pre- and post-training surveys along with a post-training processing group were utilized to measure changes in perceived RS knowledge and comfort as well as clinician perception of the feasibility, acceptability, and appropriateness of the RS training. High mean-item scores (scores ranged from 1 to 5, with 5 being the highest) for Feasibility (M= 4.42), Acceptability (M= 4.68), and Appropriateness (M= 4.38) were found following the training. Also, over 90% of participants endorsed a perceived increase in RS knowledge and comfort working with Black children with experiences of racial trauma. In addition to providing an overview of the training and findings, this dissertation includes a discussion on lessons learned and implications for RS training in the context of behavioral health organizations and systems