Interested in the Synergy Research Program?
Wondering if your research interests align with the SBME Synergy Summer program? Take a look at the exciting and diverse projects that our previous students have worked on!
Featured Projects
Quantifying Seated Pregnant Occupant Anthropometry Using 3D Imaging
Dean Harris
Engineering | Peter Cripton Lab
Motor vehicle crashes are a leading cause of death for pregnant people and fetuses in the United States and Canada, yet research on pregnant occupant safety is limited. Seatbelts are still primarily evaluated using male anthropomorphic test devices (ATDs), despite substantial anatomical differences during pregnancy. This study assesses seatbelt fit in a large cohort of pregnant people by examining belt position relative to external anatomy.
Participants sat in a simulated vehicle seat and were 3D‑scanned using a LiDAR scanner, with markers placed on the ASIS and sternoclavicular joint. Scans were analyzed in 3D Slicer and Meshmixer to measure belt distance from key bony landmarks.
Initial findings show that the enlarged pregnant abdomen alters belt routing, preventing proper engagement with the ASIS and shifting the shoulder belt away from the sternum onto soft tissue.
The goal of this work was to improve understanding of seatbelt fit for pregnant occupants, quantify relevant anthropometry, and support public education on proper belt positioning. This study also contributed methods for assessing seatbelt fit to advance motor‑vehicle injury‑prevention efforts.
Deep Learning on PPG Signals for Sleep Apnea Detection in Children
Corliss Chu
BIOEDICAL ENGINEERING | Calvin Kuo Lab
Sleep apnea is characterized by complete pauses in breathing lasting at least 10 seconds during sleep. Polysomnography (PSG) is the current diagnostic standard, but it is costly, less accessible, and dependent on sleep‑specialist expertise. Wearable devices using artificial intelligence (AI) are therefore being explored for automated detection and pre‑screening. Apnea episodes often lead to changes in blood oxygen saturation (SpO₂), a derived metric based on the underlying photoplethysmography (PPG) waveform. Because SpO₂ values are averaged over time, they may mask subtle pulsatile and respiratory changes, and AI models that rely only on SpO₂ can miss certain apnea events. This project investigates the use of raw PPG signals as input to a deep learning model for apnea detection.
A convolutional neural network (CNN) was developed to learn spatial and temporal patterns from time‑frequency representations of the PPG signal. The model uses templates spanning the full frequency range and slides along the time axis to identify where key signal features occur. The current model achieves a sensitivity of 67% for detecting apnea events. While further improvement is needed, combining physiology‑driven data processing with a CNN provides a basis for developing an accessible and reliable apnea detection tool for children.
Developing a Dendritic Cell-Targeting Cancer Vaccine
JAE-YOON KIM
BIOMEDICAL ENGINEERING | YANPU HE Lab
The immune system plays a central role in suppressing tumour growth and metastasis, and cancer immunotherapies seek to activate immune signalling pathways that generate strong anti‑tumour responses. The stimulator of interferon genes (STING) pathway is a promising therapeutic target, and the lab has been developing protein‑based STING signalling complexes for cancer treatment. However, existing designs lack selectivity for immune cells where activation is most effective. To improve targeting, a new STING fusion protein was engineered to specifically engage dendritic cells (DCs), a key antigen‑presenting cell population.
Producing this recombinant protein in E. coli presented challenges, including endotoxin contamination and low purity. These limitations were addressed by creating a new E. coli strain combining an endotoxin‑free background with rare‑codon tRNAs, enabling cleaner, higher‑yield production of the STING fusion protein and reducing toxicity concerns.
The study evaluated DC targeting, antigen‑presentation efficiency ex vivo, and the ability of the fusion protein to elicit anti‑tumour immunity in mouse cancer models.
How AI Uses Tissue Images to Diagnose Rare Metastatic Cancers
Björn Holst
Computer science | Ali Bashashati Lab
Most cancer subtypes have targeted therapies, but treatment becomes increasingly complex once cancer metastasizes. Five‑year survival drops to 17% for stage IV metastatic disease compared to 56% in pre‑metastatic stage III, and metastatic cancer accounts for over 90% of cancer deaths. Effective treatment requires identifying the tissue of origin (TOO), yet in 1–2% of cases a primary site cannot be found. These patients, diagnosed with Cancer of Unknown Primary (CUP), undergo extensive pathology, radiology, endoscopy, and laboratory evaluations before receiving nonspecific chemotherapy, and fewer than 25% survive beyond one year.
As digital pathology expands, hospitals are rapidly accumulating digitized imaging and omics data. Foundation models trained on cancer histology encode rich, tissue‑related features and continue to improve as data grows. This project leverages whole‑slide image encodings from these models to perform inexpensive, explainable TOO discovery using lightweight multitask networks and data‑driven unsupervised learning, generating organ predictions that support pathologists.
IBD Microbiota Modulates Mucus & Butyric Acid Production in Human Microbiota-Associated Mouse Model
Brian Deng
Microbiology and Immunology | carolina tropini Lab
Inflammatory bowel disease (IBD) affects millions worldwide and is characterized by chronic, relapsing inflammation of the gastrointestinal (GI) tract. The GI tract harbors trillions of microorganisms, collectively called the microbiota, which support human health by producing anti‑inflammatory short‑chain fatty acids (SCFAs). Patients with IBD have an altered, less diverse microbiota and reduced SCFA levels, particularly butyric acid. This study explored how an IBD‑derived microbiota modulates gut physiology and SCFA production in the absence of inflammation. Germ‑free mice were colonized with fecal samples from patients with IBD or healthy controls, and after four weeks of stabilization, bacterial composition, cecal SCFA concentrations, and intestinal health were assessed.
Mice colonized with IBD microbiota exhibited reduced bacterial diversity, lower butyric acid concentrations, and altered intestinal mucus. Gene‑function analysis showed enrichment of predicted mucus‑degradation pathways and depletion of butyrate‑production pathways, aligning with experimental findings. Overall, the data indicate that loss of specific microbial members drives changes in mucus integrity, SCFA levels, and microbiota function.
SYNERGY RESEARCH DAY
At the end of the program, Synergy students present their work at the Synergy Research Day. This is an excellent opportunity for students to gain valuable presentation skills, as well a chance to network with SBME’s vibrant community of researchers and industry partners. Below are the programs from previous Synergy Research Days; they’ll give you an idea of the type of research that past Synergy students have conducted during the program.