Our lab is affiliated with three different research institutes Djavad Mowafaghian Centre for Brain Health, Center for Heart Lung Innovation (HLI), and The Center for Aging SMART. Some active projects are listed below. Prior to 2021, our laboratory was located at Rutgers University: CompastLab
Improved Management of Musculoskeletal Disease
The annual economic burden of musculoskeletal disease accounted for $213 billion and $ 22 billion in the United States and Canada respectively. We are developing computational methods for improved management of musculoskeletal diseases by incorporating information extracted from ultrasound data. Our specific focus for this project is on arthritis, spine disorders, and traumatic injuries.
Keywords: Musculoskeletal disease, ultrasound, machine learning, spine, pedicle screw, cartilage, arthritis.
Surgical Tool Enhancement and Localization from Interventional Ultrasound
Minimally invasive surgical procedures (percutaneous biopsy, spinal anesthesia, and/or intracardiac beating heart procedures) often involve the insertion of thin surgical instruments, such as needles, into the surgical area of interest. Due to its real-time imaging capabilities, ultrasound imaging is being used as one of the standard imaging modalities during these procedures. However, simultaneous and reliable visualization of these surgical instruments and underlying anatomical structures still continues to be a very challenging task. For this project, we are developing fully automated and real-time image processing methods for accurate, robust, and real-time localization of surgical instruments from ultrasound data.
Keywords: Needle localization, ultrasound, needle, localization, enhancement, biopsy, machine learning, deep learning
Artificial Intelligence (AI)-guided Point-of-care Ultrasound (POCUS) for Out of Hospital Traumatic Brain Injury Risk Assessment
According to the recent statistics provided by the CDC, traumatic brain injury (TBI) is a significant cause of death and disability in the United States and contributes to about 30% of all injury deaths. On the other hand brain disease, disorder, or injury will affect 1 in 3 Canadians in their lifetime with the burden of neurological disorders increasing substantially over the last 25 years with a total annual cost estimated to be $61 billion in Canada alone. We are developing AI-based POCUS imaging for TBI risk assessment. Our envisioned system will be deployed at healthcare centers, care homes, and hospitals with an active focus in collecting a comprehensive dataset encompassing a diversity of patients ranging from different racial and ethnic minorities, service members and veterans, the unhoused, survivors of intimate partner violence, older adults living in care homes, professional and amateur level athletes (varsity teams) from sport training centers and UBC athletics.
Keywords: AI, ultrasound, POCUS, Brain injury, TBI
Improved Risk Assessment and Diagnosis of Early-Stage Endometriosis Using Artificial Intelligence(AI)
Endometriosis is a disease where tissue similar to the lining of the uterus grows outside the uterus, causing pain and/or infertility. According to WHO endometriosis affects 10% (190 million) of reproductive-age women and girls globally. Early diagnosis is important for effective treatment. However, currently, endometriosis sufferers often experience a diagnosis delay of more than five years. For this project, we are developing machine learning methods for early-stage endometriosis diagnosis and risk assessment from ultrasound data.
Keywords: Endometriosis, ultrasound, machine learning, deep learning, artificial intelligence
Application-driven Ultrasound Imaging for Early-Stage Liver and Kidney Disease Diagnosis
Advanced liver disease is expected to increase by 65% in 10 years and liver cancer is estimated to increase 80% by 2030 Recent studies show that, in Canada, the healthcare cost to manage chronic kidney disease (CKD) exceeds $40 billion annually. The main objective of this project is the development and validation of new signal and image processing methods and the integration of these developed algorithms into the traditional ultrasound imaging system for early-stage liver and kidney disease diagnosis.
Keywords: Liver disease, kidney disease, CT, ultrasound, machine learning, image processing
Lung Disease Management Using Computational Models
The main objective of this research topic is the development, validation, and deployment of new computational tools, based on deep learning, for processing multi-feature multi-anatomy longitudinal ultrasound data to derive models that yield individual-level accurate risk assessment and treatment monitoring in the context of lung disease management. We are also interested in deploying these methods to various other medical data such as CT.
Keywords: Lung disease, CT, ultrasound, machine learning, image processing
Point-of-care Ultrasound (POCUS) for Decentralized Healthcare
Noncommunicable diseases (NCDs) account for 71% of all deaths worldwide, claiming nearly 41 million people annually. People with chronic NCDs often face several challenges in accessing healthcare, resulting in an increased risk of mortality. For this project, we are developing an AI-guided point-of-care ultrasound (POCUS) imaging platform for improved out-of-hospital management of NCDs. We are aiming to deploy our imaging solution at healthcare centers, care homes, pharmacies, and athletic centers.
Keywords: Decentralized healthcare, accessible, POCUS, artificial intelligence