Measure vital physiological changes with computer vision
The measurement of vital physiological signs is a fundamental task to inform clinical decision-making. Numerous medical devices have been invented to perform this task, while many fitness and wellness devices, such as smart watches, have also entered the consumer market in recent years.
Despite incredible advances in technology, outside of critical care respiratory monitoring is still mostly done manually by counting chest movements over a minute or less with a stopwatch.
A camera and software can count breaths in a similar way to humans. As the chest expands and vibrates during breathing, our algorithm automatically maps out landmarks on the chest and proceeds to track their movement, from which a respiratory waveform is generated. This waveform is then used to calculate the respiratory rate. The system’s ability to monitor chest movements is not hindered even when a patient is clothed or under bed linen.
A pulse signal can be extracted from a digital camera in a similar way as reflective pulse oximeters do, but with the advantage of not requiring specialized equipment or physical contact with a patient. As pulsatile blood flows into the face, small changes in the face colour occur which are especially visible in the green channel but are undetectable to the human eye. The rate at which these changes occur gives us the pulse.
This technique is known as remote photoplethysmography. In 2008, Wim Verkruysse and his colleagues published the seminal paper ‘Remote plethysmographic imaging using ambient light’ with promising results in lab conditions. ‘R-PPG’ can be used to indirectly measure respiratory rate but is less accurate than motion analysis.
In real-world use, many adjustments need to be made. For example, people don’t sit still or face in the same direction continuously. It’s natural to move and look away during conversation so we needed to add tolerance for this. Lighting varies in different settings, and some lights have stronger blue colours than red, or vice-versa. Skin colours absorb and reflect light differently, and so a wide variety of test subjects are required
Basics of functionality
Vitacam software will search for a human face in the camera live view or from a recorded video clip. When a face is located, it will be tracked and digital landmarks will be placed over the face- and chest area. By tracking these landmarks, the software is able to calculate the vital signs of the patient. As long as the face and upper chest are visible in the live view or the recording, Vitacam will automatically measure every 15 seconds. The measurements are available both as a timestamped list and in trend format. Median values are automatically generated.
Direct integration to electric patient records is possible, or a selected result can be manually copied to any chosen platform.
Different ways to use vitacam
With our remote monitoring solution, a camera can be used anywhere from an assessment room to a waiting area or even a patient’s own home.
In-person observations can be gathered during patient assessment. Vitacam can provide measurements throughout the assessment to aid clinical decision-making through a Windows application.
Distancing and reach
Measuring patients from a separate room or another location. A mounted camera is placed in the same space with the patient, while the clinician can evaluate measurements remotely. Particularly useful when dealing with contagious patients.
Mobile device measurements
Vital signs are often best measured when the subject is relaxed and seated. Video can be recorded on a personal mobile device and sent to a cloud server for processing. Solution can be integrated to an existing mobile application or be deployed
as a standalone solution.
In elderly care, subjects may be bedridden when a concern is raised and a carer wishes to obtain data to share with another healthcare professional. The Vitacam smartphone app can be used on a stand or hand-held by a carer.
How it works
A software-based monitor can be run any hardware with enough processing power and a high enough quality digital camera.
Our solution includes all necessary software components and is easy to deploy. The first part is getting a video of the patient. With mobile devices, a short video clip can be recorded using the dedicated Vitacam mobile app. For continuous monitoring, the system uses a live view from a regular web camera that can be placed up to 2.5 metres away from the patient.
Our processing server then analyses the sent video or live feed to obtain the respiratory rate and pulse. This server can be situated in the cloud or physically in a hospital, on a mini-PC running Linux. Processed results can be returned to a nurse’s pc, Vitacam web dashboard or the chosen mobile device. As an additional service, the measurements can be written to a patient record system.
The solution can be run on any hardware that meets the requirements on performance and security. Since there are no demands on specific model or manufacturer on any individual item, there is flexibility in both setting up the system and altering the setup later on. In a situation that a piece breaks, it can be replaced with another approved 3rd party device without issue.