skip to main content

Title: CNN-Based LCD Transcription of Blood Pressure From a Mobile Phone Camera
Routine blood pressure (BP) measurement in pregnancy is commonly performed using automated oscillometric devices. Since no wireless oscillometric BP device has been validated in preeclamptic populations, a simple approach for capturing readings from such devices is needed, especially in low-resource settings where transmission of BP data from the field to central locations is an important mechanism for triage. To this end, a total of 8192 BP readings were captured from the Liquid Crystal Display (LCD) screen of a standard Omron M7 self-inflating BP cuff using a cellphone camera. A cohort of 49 lay midwives captured these data from 1697 pregnant women carrying singletons between 6 weeks and 40 weeks gestational age in rural Guatemala during routine screening. Images exhibited a wide variability in their appearance due to variations in orientation and parallax; environmental factors such as lighting, shadows; and image acquisition factors such as motion blur and problems with focus. Images were independently labeled for readability and quality by three annotators (BP range: 34–203 mm Hg) and disagreements were resolved. Methods to preprocess and automatically segment the LCD images into diastolic BP, systolic BP and heart rate using a contour-based technique were developed. A deep convolutional neural network was then more » trained to convert the LCD images into numerical values using a multi-digit recognition approach. On readable low- and high-quality images, this proposed approach achieved a 91% classification accuracy and mean absolute error of 3.19 mm Hg for systolic BP and 91% accuracy and mean absolute error of 0.94 mm Hg for diastolic BP. These error values are within the FDA guidelines for BP monitoring when poor quality images are excluded. The performance of the proposed approach was shown to be greatly superior to state-of-the-art open-source tools (Tesseract and the Google Vision API). The algorithm was developed such that it could be deployed on a phone and work without connectivity to a network. « less
; ; ; ;
Award ID(s):
Publication Date:
Journal Name:
Frontiers in Artificial Intelligence
Sponsoring Org:
National Science Foundation
More Like this
  1. Abstract Background Numerous cardiometabolic factors may underlie risk of hearing loss. Modifiable risk factors such as non-optimal blood pressure (BP) are of interest. Purpose To investigate early auditory evoked potentials (AEPs) in persons with nonoptimal BP. Research Design A cross-sectional nonexperimental study was performed. Study Sample Fifty-two adults (18–55 years) served as subjects. Individuals were classified as having optimal (systolic [S] BP < 120 and diastolic [D] BP < 80 mm Hg, n = 25) or non-optimal BP (SBP ≥=120 or DBP ≥=80 mm Hg or antihypertensive use, n = 27). Thirteen subjects had hypertension (HTN) (SBP ≥130 or DBP ≥80 mm Hg or use of antihypertensives). Data Collection and Analysis Behavioral thresholds from 0.25 to 16 kHz were collected. Threshold auditory brain stem responses (ABRs) were recorded using rarefaction clicks (17.7/second) from 80 dB nHL to wave V threshold. Electrocochleograms were obtained with 90 dB nHL 7.1/second alternating clicks and assessed for summating and compound action potentials (APs). Outcomes were compared via independent samples t tests. Linear mixed effects models for behavioral thresholds and ABR wave latencies were constructed to account for potential confounders. Results Wave I and III latencies were comparable between optimal and non-optimal BP groups. Wave I was prolonged in hypertensive versus optimal BP subjects at stimulus level 70 dB nHL (p = 0.016). ABR wave V latencies weremore »prolonged in non-optimal BP at stimulus level 80 dB nHL (p = 0.048) and in HTN at levels of 80, 50, and 30 dB nHL (all p < 0.050). DBP was significantly correlated with wave V latency (r = 0.295; p = 0.039). No differences in ABR amplitudes were observed between optimal and non-optimal BP subjects. Electrocochleographic study showed statistically comparable action and summating potential amplitudes between optimal and non-optimal BP subjects. AP latencies were also similar between the groups. Analysis using a set baseline amplitude of 0 μV showed that hypertensive subjects had higher summating (p = 0.038) and AP (p = 0.047) amplitudes versus optimal BP subjects; AP latencies were comparable. Conclusion Elevated BP and more specifically, HTN was associated with subtle AEP abnormalities. This study provides preliminary evidence that nonoptimal BP, and more specifically HTN, may be related to auditory neural dysfunction; larger confirmatory studies are warranted.« less
  2. Abstract

    In two-thirds of intensive care unit (ICU) patients and 90% of surgical patients, arterial blood pressure (ABP) is monitored non-invasively but intermittently using a blood pressure cuff. Since even a few minutes of hypotension increases the risk of mortality and morbidity, for the remaining (high-risk) patients ABP is measured continuously using invasive devices, and derived values are extracted from the recorded waveforms. However, since invasive monitoring is associated with major complications (infection, bleeding, thrombosis), the ideal ABP monitor should be both non-invasive and continuous. With large volumes of high-fidelity physiological waveforms, it may be possible today to impute a physiological waveform from other available signals. Currently, the state-of-the-art approaches for ABP imputation only aim at intermittent systolic and diastolic blood pressure imputation, and there is no method that imputes the continuous ABP waveform. Here, we developed a novel approach to impute the continuous ABP waveform non-invasively using two continuously-monitored waveforms that are currently part of the standard-of-care, the electrocardiogram (ECG) and photo-plethysmogram (PPG), by adapting a deep learning architecture designed for image segmentation. Using over 150,000 min of data collected at two separate health systems from 463 patients, we demonstrate that our model provides a highly accurate prediction of themore »continuous ABP waveform (root mean square error 5.823 (95% CI 5.806–5.840) mmHg), as well as the derived systolic (mean difference 2.398 ± 5.623 mmHg) and diastolic blood pressure (mean difference − 2.497 ± 3.785 mmHg) compared to arterial line measurements. Our approach can potentially be used to measure blood pressure continuously and non-invasively for all patients in the acute care setting, without the need for any additional instrumentation beyond the current standard-of-care.

    « less
  3. Background: Deformational plagiocephaly and brachycephaly (DPB) is manifested in ~20% of newborns in the US. DPB can be effectively corrected by repositioning and/or physical therapy if detected and monitored before 4 months of age. The cranial index (CI) and cranial vault asymmetry index (CVAI) are used for DPB diagnosis and monitoring. As there is no current tool available for pediatricians or parents to quantitatively measure these indices at the point-of-care, we developed a smartphone app, called SoftSpot, that measures CI and CVAI from photographs of a child’s head to increase the chances of early detection and treatment. Objective: To prospectively evaluate the accuracy of the smartphone measurements of CI and CVAI in a clinical setting. Methods: Bird’s eye-view head photos of 117 infants aged 2-11 months (42 female, 75 male) were captured at a large multidisciplinary craniofacial center with the SoftSpot app (PediaMetrix Inc. Rockville, MD) using an iPhone X (Apple Inc., Cupertino, CA). The study was IRB approved and parent consent was obtained. Measurements included width, length, and diagonals of the patients’ head were obtained by a single CRNP and were used to calculate CI and CVAI as the ground truth. At least five images for each patient weremore »chosen by an analyst, CI and CVAI were automatically measured by the proprietary algorithms of the app, and results were averaged for each patient. Automated and ground truth CI and CVAI measurements were compared using the Bland-Altman method and Spearman Correlation Coefficient after excluding outliers with mean absolute error (MAE) greater than two standard deviations. Results: MAE was 2.47 ± 1.68 for CI, 1.55 ± 1.03 for CVAI. Spearman correlation coefficients were 0.93 and 0.91 (p-values < 0.001) for CI and CVAI, respectively (see Fig. 2). Bland-Altman analysis (see Fig. 2 resulted in limits of agreement of [-4.41, 6.53] for CI and [-3.64, 3.68] for CVAI, with respective biases of 1.06 and 0.02. Conclusion: Our app measures CI and CVAI from head 2D photos with very high correlation to caliper-based measurements obtained in the craniofacial clinic. This prospective study demonstrates the clinical feasibility of using a smartphone app for cranial measurements at the point-of-care with the potential to early detect and monitor DPB. The app can potentially be used in telemedicine encounters when in-person visits are difficult due to circumstances like COVID-19 or for remote and underserved areas.« less
  4. Frequent blood pressure monitoring is the key to diagnosis and treatments of many severe diseases. However, the conventional ambulatory methods require patients to carry a blood pressure (BP) monitoring device for 24 h and conduct the measurement every 10--15 min. Despite their extensive usage, wearing the wrist/arm-based BP monitoring device for a long time has a significant impact on users' daily activities. To address the problem, we developed eBP to measure blood pressure (BP) from inside user's ear aiming to minimize the measurement's impact on users' normal activities although maximizing its comfort level. The key novelty of eBP includes (1) a light-based inflatable pulse sensor which goes inside the ear, (2) a digital air pump with a fine controller, and (3) BP estimation algorithms that eliminate the need of blocking the blood flow inside the ear. Through the comparative study of 35 subjects, eBP can achieve the average error of 1.8 mmHg for systolic (high-pressure value) and -3.1 mmHg for diastolic (low-pressure value) with the standard deviation error of 7.2 mmHg and 7.9 mmHg, respectively. These results satisfy the FDA's AAMI standard, which requires a mean error of less than 5 mmHg and a standard deviation of less than 8more »mmHg.« less
  5. Abstract

    Hypertension is associated with more end-organ damage, cardiovascular events, and disability-adjusted life years lost in the United States compared with all other modifiable risk factors. Several guidelines and scientific statements now endorse the use of out-of-office blood pressure (BP) monitoring with ambulatory BP monitoring or home BP monitoring to confirm or exclude hypertension status based on office BP measurement. Current ambulatory or home BP monitoring devices have been reliant on the placement of a BP cuff, typically on the upper arm, to measure BP. There are numerous limitations to this approach. Cuff-based BP may not be well-tolerated for repeated measurements as is utilized with ambulatory BP monitoring. Furthermore, improper technique, including incorrect cuff placement or use of the wrong cuff size, may lead to erroneous readings, affecting diagnosis and management of hypertension. Compared with devices that utilize a cuff, cuffless BP devices may overcome challenges related to technique, tolerability, and overall utility in the outpatient setting. However, cuffless devices have several potential limitations that limit its routine use for the diagnosis and management of hypertension. The review discusses the different approaches for determining BP using various cuffless devices including engineering aspects of cuffless device technologies, validation protocols to testmore »accuracy of cuffless devices, potential barriers to widespread implementation, and future areas of research. This review is intended for the clinicians who utilize out-of-office BP monitoring for the diagnosis and management of hypertension.

    « less