Continuous Monitoring of Vital Signs Using Cameras: A Systematic Review

Abstract

In recent years, noncontact measurements of vital signs using cameras received a great amount of interest. However, some questions are unanswered: (i) Which vital sign is monitored using what type of camera? (ii) What is the performance and which factors affect it? (iii) Which health issues are addressed by camera-based techniques? Following the preferred reporting items for systematic reviews and meta-analyses (PRISMA) statement, we conduct a systematic review of continuous camera-based vital sign monitoring using Scopus, PubMed, and the Association for Computing Machinery (ACM) databases. We consider articles that were published between January 2018 and April 2021 in the English language. We include five vital signs: heart rate (HR), respiratory rate (RR), blood pressure (BP), body skin temperature (BST), and oxygen saturation (SpO₂). In total, we retrieve 905 articles and screened them regarding title, abstract, and full text. One hundred and four articles remained: 60, 20, 6, 2, and 1 of the articles focus on HR, RR, BP, BST, and SpO₂, respectively, and 15 on multiple vital signs. HR and RR can be measured using red, green, and blue (RGB) and near-infrared (NIR) as well as far-infrared (FIR) cameras. So far, BP and SpO₂ are monitored with RGB cameras only, whereas BST is derived from FIR cameras only. Under ideal conditions, the root mean squared error is around 2.60 bpm, 2.22 cpm, 6.91 mm Hg, 4.88 mm Hg, and 0.86 °C for HR, RR, systolic BP, diastolic BP, and BST, respectively. The estimated error for SpO₂ is less than 1%, but it increases with movements of the subject and the camera-subject distance. Camera-based remote monitoring mainly explores intensive care, post-anaesthesia care, and sleep monitoring, but also explores special diseases such as heart failure. The monitored targets are newborn and pediatric patients, geriatric patients, athletes (e.g., exercising, cycling), and vehicle drivers. Camera-based techniques monitor HR, RR, and BST in static conditions within acceptable ranges for certain applications. The research gaps are large and heterogeneous populations, real-time scenarios, moving subjects, and accuracy of BP and SpO₂ monitoring.

Keywords: blood pressure; body temperature; camera; contactless; continuous monitoring; heart rate; noncontact; oxygen saturation; remote health care; respiratory rate; vital sign.

Figure 1

PRISMA diagram of literature screening.

Figure 2

General flow diagram of vital sign measurements from video/image data.

Figure 3

Data acquisition according to study design, hardware, and ground truth.

Figure 4

Distribution of subject information including obtained data, number of subjects, participant type and available (A) and not available (NA) info of skin tone.

Figure 5

Distribution of hardware parameters, namely camera type, frame rate, resolution (r), and camera-subject distance.

Figure 6

Chord diagram of conceptual semantics between camera and vital sign.

Figure 7

Distribution of ground truth.

Figure 8

Comprehensive work flow from various camera images to HR vital sign (Note: FF: full frame, VJ: Viola Jones algorithm, HoG: Histogram of oriented gradients; CNN: convolutional neural network; ManualD: manual ROI detection, NoD: no ROI detection, OtherD: Other ROI detection, ManualT: manual ROI tracking, KLT: Kanade–Lucas–Tomasi; KCF: kernel correlation filter; featT: feature tracking, OtherT: other ROI tracking, NoT: no ROI tracking, BSS: blind source separation algorithms, otherSP: other signal processing, ML: Machine learning FD: frequency domain, TD: time domain).

Figure 9

(a) RMSE of HR in static and dynamic conditions and, (b) varying distance between the subject and camera.

Figure 10

RMSE of RR estimation with varying distance.

Figure 11

Factors affecting the vital signs performance.

Figure 12

The framework of main research gaps and future directions. In this, red color boxes represent the research gaps, and green color boxes refer to future research directions.