Authors:

Aarya Gajendra Sontakke

Date Created:

Not specified

Course Title:

Professor:

Not specified

About Paper:

Rising obesity among women has led to bariatric surgery and glucagon- like peptide-1 receptor agonist (GLP-1 RA) therapy. These treatments induce weight loss and improve metabolic health but may weaken bone structure and raise fracture risk. Early detection of bone fragility is vital. Standard dual-energy X-ray absorptiometry (DXA) scans cannot capture local structural bone changes after treatment, delaying identification of patients at risk for treatment-induced fragility until fractures occur. Novel imaging biomarkers are required for timely monitoring and early intervention to prevent complications. For a preliminary analysis, we collected high-resolution peripheral quantitative computed tomography (HR-pQCT) scans of the distal tibia from eight participants, six women and two men, receiving GLP-1 RA therapy and 61 age- and sex-matched controls. We cropped images to remove background, segmented five anatomical regions (soft tissue and bones) using a previously developed deep learning model, and post-processed to correct anomalies. Furthermore, we subdivided each soft tissue region into muscle, fat, and skin compartments and extracted comprehensive radiomic features, including first-order statistics, second-order texture metrics, three- dimensional shape descriptors, and wavelet and Fourier transform coefficients. Using these features, we aim to extract clinically relevant structural patterns that differentiate the GLP-1 group from controls. Finally, we are developing a convolutional neural network to integrate whole-volume image information with radiomics outputs, laying the groundwork for an automated, personalized framework to detect early treatment-related bone fragility. Our imaging framework will enable non- invasive detection of early bone weakening after rapid weight loss in women with obesity, allowing personalized monitoring to reduce fracture risk. Keywords: [no keywords provided] † Presenting Undergrad Author; ‡ Contributing Undergrad Author; * Undergrad Acknowledgment

Source:

Purdue University / 2025

Topics:

No topics listed

Co-authors:

Aarya Gajendra Sontakke