Research

Research Spotlight

Carousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel imageCarousel image

Current Research

A Fully-Automated Endoscopic Scoring System for Ulcerative Colitis

In this project, a computer-aided system for colonoscopy video analysis is under development, which will facilitate disease severity measurement and aid in treatment selection and clinical outcome prediction.

Project Lead: Heming Yao

Clinical Lead: Ryan Stidham

Project Team: Reza Soroushmehr, Jonathan Gryak, Jie Gui

Research Theme: Decision Support Systems

Colonoscopy is a standard medical examination used to inspect the mucosal surface and detect abnormalities of the colon. Objective assessment and scoring of disease features in the colon are important in conditions such as colorectal cancer and inflammatory bowel disease. However, human disease assessment and measurement is hampered by interobserver variation and several biases.

In this project, a computer-aided system for colonoscopy video analysis is under development, which will facilitate disease severity measurement and aid in treatment selection and clinical outcome prediction. The proposed system consists of non-informativeness classification, Mayo severity estimation, image-wise biopsy forceps detection, image-wise location estimation, image segmentation and camera’s motion tracking. By analyzing image features and camera motion, we could derive the contextual understanding of the colonoscopy video such as severity and surgical regions distribution, which can provide richer information for patient’s condition evaluation and outcome prediction.

A Multimodal Integrative Platform for Continuous Monitoring and Decision Support during Postoperative Care in Cardiac Patients

The project will develop an innovative, real-time clinical decision support (DSS) platform. The platform will utilize Big Data analytic methods, novel algorithms, and software tools to integrate and analyze disparate sources of continuous and non-continuous patient data.

The DSS platform will be able to understand and predict potential complications and recovery trends for patients following cardiac surgery, other open vascular operations in the chest and abdomen, and other trauma surgeries involving the cardiovascular system.

Project Lead: Jonathan Gryak

Clinical Lead: Michael Mathis

Project Team: Harm Derksen, Gang Liu, Renaid Kim, Maryam Bagherian

Research Theme: Decision Support Systems

The project will develop an innovative, real-time clinical decision support (DSS) platform. The platform will utilize Big Data analytic methods, novel algorithms, and software tools to integrate and analyze disparate sources of continuous and non-continuous patient data.

The DSS platform will be able to understand and predict potential complications and recovery trends for patients following cardiac surgery, other open vascular operations in the chest and abdomen, and other trauma surgeries involving the cardiovascular system.

Algorithms for Tensor-Based Modeling of Large-Scale Structured Data

In this proposal, the PIs develop novel theoretical foundations and scalable algorithms for tensor decomposition and apply them to some major Big Data tasks such as noise removal, data imputation, information integration and dimension reduction. To evaluate the performance of the proposed tensor methods, they will be applied to early detection of sepsis, which represents a spectrum of complex problems in medicine and science.

Project Lead: Harm Derksen

Clinical Lead: Kevin Ward

Project Team: Jonathan Gryak, Olivia Alge, Maryam Bagherian, Cristian Minoccheri

Research Theme: Artificial Intelligence Techniques

Most approaches to Big Data do not exploit all of their structure. Specifically, Big Data tasks such as denoising, imputation of missing data and classification can greatly benefit from the information contained in the structure of the data. Typically, arrays of dimension $3$ or higher are flattened in order to apply conventional linear algebra methods. In doing so, one loses the rich tensor structure.

The Canonical Polyadic (CP) decomposition is a powerful tool for analyzing tensor data. CP decompositions have been applied in many areas such as psychometrics, chemometrics, signal and image processing, computer vision, neuroscience and finance. Unfortunately, CP decompositions are computationally inefficient and numerically unstable. The lack of scalable algorithms to apply tensor-based methods is a major obstacle in utilizing the structural information.

In this proposal, the PIs develop novel theoretical foundations and scalable algorithms for tensor decomposition and apply them to some major Big Data tasks such as noise removal, data imputation, information integration and dimension reduction. To evaluate the performance of the proposed tensor methods, they will be applied to early detection of sepsis, which represents a spectrum of complex problems in medicine and science.

AngioAid – A Computer-Based Platform Interpreting Coronary Angiograms

This project has two aims. First, the construction of AngioAid, a fully automated computer-based platform to assist with the interpretation of coronary angiogram videos. Secondly, the creation of a curated dataset of coronary angiogram videos, made publicly available via Amazon Web Services, that will spur development of new algorithms to advance the field of angiogram interpretation.

Project Lead: Alexander Wood

Clinical Lead: Brahmajee Nallamothu

Project Team: Reza Soroushmehr

Research Theme: Decision Support Systems

Development and Assessment of an In-Vehicle Cardiac Monitoring and Severe Event Prediction System

The proposed system will accurately predict or detect the onset of severe cardiac events (e.g., hemodynamic instability) prior to complications experienced by the driver in order to ensure driver safety.

Project Lead: Jonathan Gryak

Clinical Lead: Hamid Ghanbari

Project Team: Harm Derksen, Zhi Li, Alexander Wood, Julia Pagnucco, Gang Liu, Winston Zhang

Research Theme: In-Vehicle Systems

Image Processing and Machine Leaning for Prediction of Wound Healing

We have developed wound segmentation and deep neural network models that can accurately predict infection and wound healing. The image segmentation methods provide an accurate “hands-free” measurement of wound surface area. Our objective here is to refine our algorithms and test the suitability of image analysis as a biomarker for healing or infection. We will do this by performing image analyses of DFU treated by total contact cast (TCC) and by comparing the image features to those not treated by TCC. Our working hypothesis is there will be unique features from image analysis that can accurately predict infection or healing beyond surface area changes over time, known clinical predictors (Brownrigg, Hinchliffe et al. 2016) or candidate biomarkers.

Project Lead: Jonathan Gryak

Clinical Lead: Rodica Busui

Project Team: Renaid Kim, Heming Yao

Research Theme: Decision Support Systems

Can one tell if a wound is healing? As basic as this question is, it is not easily answered using current practice for the treatment of diabetes-related foot ulcers (DFU). It is well known that DFU surface area changes at 4 weeks are highly predictive of subsequent wound closure.(Sheehan, Jones et al. 2003, Robson, Cooper et al. 2006, Steed, Attinger et al. 2006, Snyder, Cardinal et al. 2010) Simply providing this information to wound care providers improves wound healing outcomes.(Kurd, Hoffstad et al. 2009) Additionally, the quality and quantity of the tissue in the wound bed also offers important prognostic information.(Sherman 2003, Valenzuela-Silva, Tuero- Iglesias et al. 2013) Unfortunately outside of the research setting, accurate measurement of wound surface area changes and quantification of the wound base are out of reach in the busy wound practice setting. Currently, clinicians estimate wound size by estimating wound width and length using a scalpel after wound treatment. There is a great deal of measurement error in assuming all wounds are rectangles. Rarely do clinicians take the next step in estimating wound surface area and changes over 2 and 4 weeks; crucial time points for wound healing assessment.(Sheehan, Jones et al. 2006)

Our long-term goal is to implement cognitive computing advancements in image processing and machine learning into a busy wound care setting to fundamentally change the quality of wound care. We have developed wound segmentation and deep neural network models that can accurately predict infection and wound healing.(Wang, Yan et al. 2015) The image segmentation methods provide accurate “hands-free” measurement of wound surface area. Our objective here is to refine our algorithms and test the suitability of image analysis as a biomarker for healing or infection. We will do this by performing image analyses of DFU treated by total contact cast (TCC) and by comparing the image features to those not treated by TCC. Our working hypothesis is there will be unique features from image analysis that can accurately predict infection or healing beyond surface area changes over time, known clinical predictors (Brownrigg, Hinchliffe et al. 2016) or candidate biomarkers. The rationale for the proposed research is that, image analysis features along with surface area changes over time will accurately identify non-healers at an early stage where more advanced wound care strategies can result in improved patient outcomes.

Improving Care for Heart Failure Patients Using Tropical Geometry and Soft Computing

In this project, we aim to develop an automated decision-making system that can identify patients eligible for an HT/MCS that would facilitate primary care physicians or general cardiologists referring those patients for consideration of advanced therapies.

Project Lead: Heming Yao

Clinical Lead: Jessie Golbus

Project Team: Keith Aaronson, Jonathan Gryak, Renaid Kim, Harm Derksen

Determining the appropriate timing of mechanical circulatory support (MCS) device implantation or heart transplantation (HT) for patients with advanced heart failure is essential as there may be a mortality cost to delayed care. Recognizing patients eligible for an HT/MCS requires expertise from providers trained in transplant cardiology. In this project, we aim to develop an automated decision-making system that can identify patients eligible for an HT/MCS that would facilitate primary care physicians or general cardiologists referring those patients for consideration of advanced therapies. We have developed a hybrid of fuzzy logic and neural networks with a genetic algorithm to identify patients that are eligible for HT/LVAD. Our model achieved better classification performance than other traditional ML algorithms. To overcome the optimization difficulty and computational burden in the existing model, we are developing a novel soft computing architecture with tropical geometry. The proposed network can be not only optimized by gradient-based methods but also close to piecewise linear to ensure the model’s interpretability.

Novel Noninvasive Methods of Intracranial Pressure and Cerebrovascular Autoregulation Assessment: Seeing the Brain through the Eyes during Postoperative Care in Cardiac Patients

Among other image modalities used for TBI assessment, a number of recent studies reported that hand-held optic nerve ultrasound (ONUS) may help identify elevated intracranial pressure (ICP). In the largest study of ICP prediction using ONUS to date, we found that optic nerve sheath diameter can accurately predict ICP > 25 mmHg in a mixed population of patients in a neurosurgical ICU. Our preliminary work has demonstrated the feasibility of fully automated algorithms to measure optic nerve sheath diameter. In this project we will further develop these algorithms to predict ICP from measuring the optic nerve sheath diameter.

Project Lead: Reza Soroushmehr

Clinical Lead: Craig Williamson

Project Team: Hakam Tiba, Krishna Rajajee

Since 2000, more than 300,000 service members have been diagnosed with TBI, and over 34,000 have experienced moderate to severe injury. In 53% of acute combat deaths between 2001 and 2011, the brain was the primary site of injury. Despite the disproportionately high incidence of TBI injury and deaths among U.S. warfighters, there is not a reliable noninvasive ICP monitoring technique available for use in the field. Elevated ICP following TBI is a powerful predictor of subsequent neurological deterioration and overall outcome.There is an urgent need to develop noninvasive methods to estimate and prevent elevated intracranial pressure (ICP) to prevent secondary brain injury and ensure adequate cerebral perfusion pressure as a fundamental component of TBI care. However, invasive ICP monitoring is not without both controversy and complications, and is unavailable in Role 1 settings. Although there have been technological advancements, no single noninvasive ICP monitoring technique has yet proven sufficiently reliable to adopt into routine practice.

Among other image modalities used for TBI assessment, a number of recent studies reported that hand-held optic nerve ultrasound (ONUS) may help identify elevated ICP. In the largest study of ICP prediction using ONUS to date, we found that optic nerve sheath diameter can accurately predict ICP > 25 mmHg in a mixed population of patients in a neurosurgical ICU. Our preliminary work has demonstrated the feasibility of fully automated algorithms to measure optic nerve sheath diameter. In this project we will further develop these algorithms to predict ICP from measuring the optic nerve sheath diameter.

Osteoarthritis of the Temporomandibular Joint

In this project, we will identify quantitative imaging, biochemical and clinical markers and design predictive models of TMJ OA progression. We develop image processing and machine learning methods and integrate all clinical, imaging and biological to classify patients and to make a prediction on the progress of the disease.

Project Lead: Reza Soroushmehr

Clinical Lead: Lucia Cevidanes

Project Team: Winston Zhang

Osteoarthritis (OA) is a top cause of chronic disability with aging: The National Center for Disease Control estimates that 78.4 million (26%) adults, 18 and older, will have a diagnosis of arthritis by the year 2040, compared with the 54.4 million adults in 2013-2015. However, aging alone does not cause OA; rather the multifactorial mechanisms responsible for OA may include an age-related pro-inflammatory state. The continuum of changes modulated by both systemic and local factors in age-related inflammation, with transition from healthy aging to disease, remains understudied. The disease cannot be diagnosed until it becomes symptomatic, at which point structural alterations already are advanced.

We will identify quantitative imaging, biochemical and clinical markers and design predictive models of TMJ OA progression. We develop image processing and machine learning methods and integrate all clinical, imaging and biological to classify patients and to make a prediction on the progress of the disease.

Polytrauma Decision Support System for Pelvic and Abdominal Traumatic Injuries

Trauma is the leading cause of death among Americans under 44. The mortality rate for patients with pelvic injuries is increased by the risk of further complications, especially severe hemorrhage. However, even a single trauma patient generates large volumes of information, including vital signs, injury severity scores, demographic details, lab reports, and in particular, complex medical images such as CT scans and X-rays—all of which impact diagnosis and treatment, as well as costs. A computer-assisted decision support system (DSS) capable of rapidly analyzing large volumes of patient information to generate accurate treatment recommendations and outcome predictions has the potential to improve both patient care/survival and resource utilization. The proposed Decision Support System (DSS) technology is envisioned to significantly improve pelvic/abdominal trauma decision-making using facilitated and prompt analysis of complex and heterogeneous patient medical data.

The current project will extend the algorithms of our DDS technology to:

  1. Integrate our algorithms and methods for analysis of data for pelvic/abdominal injuries with those of TBI such that the resulting DDS software can be used in polytrauma cases.

  2. Further develop the planned software implementation of our technology as a user-friendly software tool to be used in the clinical setting.

Project Lead: Alexander Wood

Clinical Lead: Craig Williamson

Project Team: Negar Farzaneh, Cheng Jiang, Reza Soroushmehr, Heming Yao

Prediction and Assessment of Acute Respiratory Distress Syndrome

This project aims to create novel computational methodologies that synthesize and integrate longitudinal electronic health data streams for real-time and continuous health status monitoring and early detection of disease. We will utilize these technologies to address the problem of monitoring patients with lung disease to detect the presence of the acute respiratory distress syndrome (ARDS) at early stages. The project combines the capabilities of two emerging fields of machine learning, privileged learning and learning from uncertain data, both highly relevant to healthcare applications. While the current project focuses on improving diagnosis of ARDS, the proposed learning methods will generalize across healthcare settings, allowing for better characterization of patient health status in both in-hospital and in-home settings via portable electronic monitoring devices.

Project Lead: Jonathan Gryak

Clinical Lead: Michael Sjoding

Project Team: Nick Reamaroon, Zijun Gao

Racial Disparities in Health: The Roles of Stress, Social Relations and the Cardiovascular System

Hypertension is the number one cause of racial disparities in mortality in the U.S. Understanding the mechanisms by which race is linked with the cardiovascular system is key for potentially reducing race disparities in hypertension and hypertension related mortality. The present study is guided by existing theories of racial health disparities, which suggest that racial health disparities are due to variations in long-term exposure to stress that are moderated by social relations and age.

Project Lead: Maryam Bagherian

Clinical Lead: Kira Birditt

This project has three aims:

  1. Test links between long-term stress exposure and short-term stress reactivity among Eurpoean American (EA) and AA adults.

  2. Examine age differences in long-term stress exposure and short-term reactivity by race.

  3. Determine how long-term social relationships moderate individual differences in stress exposure and reactivity.