Blood flow in microcirculation is a significant physiological parameter that reflects the adaptive response of organs to disease, trauma, and cancer. Although ultrasound Doppler imaging was previously unable to assess blood flow in the microvasculature (< 0.5 cm/sec), the introduction of microbubble contrast agents has removed this limitation. However, blood flow of the entire vascular tree is mixed together during imaging. We present a method that segments and visualizes the entire vascular tree, including capillary blood flow, larger sub-spatially resolved vasculature and larger vasculature (>50 µm). In this work, we present an approach that decomposes nonlinear Doppler acquisitions into different groups of velocity projections. We demonstrated the ability to segment these different levels of vasculature in a rat spinal cord injury model where the varying rates of low velocity microbubble decorrelations captured by our high frame rate acquisitions enable us to quantify microvascular blood flow. This approach overcomes limitations encountered in conventional imaging methods by removing tissue signal before Doppler processing by combining high-frame rate plane wave imaging, microbubble nonlinear pulse sequences, and Doppler segmentation of blood flow. Singular value decomposition was used to segment the nonlinear Doppler signal. Our results successfully illustrate the segmentation of lower velocity sub-resolution microvascular flow and higher velocity flow in larger vessels in a rat spinal cord injury model. We isolated low and mid-velocity flow in sub-resolution vasculature (<20 µm). We observed different spatial distribution and bolus kinetics between low- mid- and higher velocity Doppler projections. We are assessing the utility of these different blood flow features for the management of spinal cord injury and other applications (e.g. oncological).

With cardiovascular disease as the leading cause of death worldwide, there is a need for improved wearable monitoring tools for assessing the health of the cardiovascular system. Photoplethysmography (PPG) is a continuous, non-invasive measurement that encodes a multitude of informative vital signs, including heart rate, heart rate variability, respiratory rate, cardiac output, and arterial stiffness. Although existing PPG sensing technologies record from the finger or wrist, the face presents a promising and underutilized location for wearable pulse sensing. This work presents a novel wearable PPG sensing system that records at multiple wavelengths and facial locations. As a proof-of-concept, we seek to evaluate a potential application of our system incorporated in a surgical face mask for use in intra-operative hemodynamic monitoring. By collecting data with our system alongside ground truth cardiovascular vital signs, we can build and test non-invasive inference algorithms. After validating our system’s heart rate detection accuracy with a standard error of 2.84 beats per minute, we now proceed to test our device’s ability to infer additional cardiovascular parameters. In addition to showing promise for novel non-invasive, continuous surgical monitoring, this work has broader implications for wearable health applications based on face-worn form factors such as glasses, helmets, and headsets.

Blood pressure (BP) serves as the primary indicator of a patient’s cardiovascular health. Today, cuff-based BP monitors are the gold standard for routine blood pressure monitoring. However, they are prone to inaccuracies and cannot provide continuous readings. Continuously monitoring BP would allow patients to observe their BP fluctuations from eating, medicine intake, and exercise, thus empowering individuals diagnosed with hypertension to make better-informed health decisions. This motivates the need for a non-invasive, continuous blood pressure monitor. Prior studies have already shown the potential for pulse transit time (PTT), which is the time for a pulse wave to travel between two arterial sites, to be used for non-invasively measuring BP. In this work, I hope to improve upon this technique. To do this, I focus on testing and improving the pulse detection accuracy of a system incorporating an optical sensor array in a surgical eye protection face mask. By getting a better resolution of the pulse waves, I believe the estimate of BP will be more accurate and, in turn, provide a valuable dataset to further investigate the relationship between PTT and BP.

Alzheimer’s disease (AD) is an irreversible neurodegenerative disease that affects memory and thinking, but can develop undetected long before clinical symptoms appear. There are no current cures for AD and treatments to slow the progression of the disease have greater chances of success if the disease is caught early, but current methods, such as PET and lumbar puncture, are limited in terms of applications as preventative screening methods for early detection. Default Mode Network (DMN) functional connectivity obtained through resting state functional magnetic resonance imaging (rsfMRI) has been proposed as a non-invasive biomarker for AD. Therefore, we explored the dynamic FC between the DMN nodes in healthy control, mild cognitive impairment (MCI), and AD subjects using a sliding window analysis of ultrafast resting state fMRI (rs-fMRI) data. Using machine learning methods, we were able to identify highly occupied states through clustering and developed a classifier to distinguish between the groups using the insights gleaned from clustering. Further development of this classifier can potentially benefit both clinical and research settings in AD diagnosis and treatment by providing a non-invasive imaging method for early detection of AD or by allowing researchers to monitor the progression of the disease or therapeutic effects potential treatment options.

Cellular regulation of transcription is a complex phenomenon, with a wide range of biological determinants influencing the binding of transcription factors (TFs). Being able to accurately predict TF binding is important to pinpoint noncoding DNA regions where mutations are likely to cause disease. Existing approaches are able to accurately predict TF binding across the genome, but are not focused on interpretability. Our approach, EXPERT: Explainable Prediction of Transcription Factor Binding based on Histone Modification Data, achieves state-of-the-art predictive performance while simultaneously offering local interpretations that reveal biologically meaningful relationships by using a stage-wise boosting technique and local additive feature attributions. We utilize a performant non-linear model (XGBoost) and an efficient local feature attribution method (TreeSHAP) to demonstrate how to increase the signal learned from a set of biologically meaningful features (histone binding in DNA) using a stage-wise gradient boosting scheme while maintaining high performance. EXPERT enables understanding predictive models through biological explanations and can further our foundational understanding of the epigenome by highlighting novel biological relationships.

The dystrophin protein protects cardiac and skeletal muscle from damage during normal contraction and relaxation by acting as a shock absorber in the cell. Mutations in the dystrophin gene lead to Duchenne muscular dystrophy (DMD), an X-linked recessive disease. Boys suffering from the disease become ventilator dependent at a young age and usually succumb to cardiac failure in their thirties. Other symptoms of DMD include muscle wasting, cardiomyopathy and respiratory failure. Currently there is no cure for DMD and while gene therapy has shown great promise, it still needs to be complemented with additional therapeutic interventions in order to fully address the symptoms of DMD. Previous work by our lab identified several drugs that blocked a certain type of calcium channel in cardiac muscle and protected the cells from damage following injury by correcting calcium movement into and out of the cell during muscle contraction. In this study, the leading drugs will be tested further using a cross-platform approach. The drugs will initially be tested in cardiac and skeletal muscle differentiated from healthy and DMD patient-derived induced pluripotent stem cells (iPSCs). The top three drugs that restore normal contraction and relaxation kinetics in vitro will then be tested in the DMDmdx rat. This novel, small animal model has a similar progression of DMD symptoms to human patients. The rats will be fed the drugs in their chow. Skeletal and cardiac muscle performance will be evaluated to determine whether correcting muscle contraction kinetics ameliorates the symptoms of DMD. Furthermore, we will show whether the same drug is equally effective in treating both cardiac and skeletal muscle. The cross-platform approach may help to better predict drug efficacy, leading to a reduced rate of failure in clinical trials. Moreover, this approach may serve as a benchmark for drug discovery in other neuromuscular diseases.

Heart disease is the leading cause of death in the United States, contributing to 1 in every 4 deaths. One of the most common causes of heart failure is dilated cardiomyopathy (DCM), and approximately 35 percent of these cases are caused by mutations in genes encoding for a variety of sarcomeric, cytoskeletal and nuclear envelope proteins. There are nearly 1,500 separate mutations in sarcomere protein genes that can lead to DCM, but patients are not currently treated differently based on these unique mutations. In this study, I use two mouse models of genetic DCM, D230N and I61Q, which have mutations that occur in cardiac Tropomyosin (Tm) and cardiac Troponin C (cTnC), respectively. These mutations result in desensitization of the cardiac myofilament, leading to negative health consequences such as left ventricular dilation and reduced ejection fraction. I am testing the hypothesis that the ionotropic Levosimendan will be more effective in the I61Q cTnC DCM model because the drug binds directly to the N-terminal domain of cTnC, increasing its affinity for Ca2+ which results in increased force generated. At 3-4 months of age, cardiomyocytes are isolated and suspended in Tyrode’s buffer, an isotonic physiological solution. Cell shortening and relaxation was measured using video microscopy (IonOptix, Milton, MA) with pacing at 1.0 Hz and 37°C in presence of Levosimendan ( 1 µM) or control (DMSO). Compared to DMSO, 1 µM Levosimenden increased myocyte shortening in NTG mice (9.36 to 13.02 %, n=5) and I61Q cTnC (7.3 to 8.4 %, n=3) but had no impact on D230N Tm mice (10.4 to 8.2 %, n=5). My results suggests that Levosimendan is more effective in I61Q cTnC compared to D230N Tm model. In future studies I plan to make similar measurements using the myosin activator Omecamtiv Mecarbil.