Motion Artifact-Resilient Embedded Sensors by Mechanically Rendered Stable Zone

Recently, miniaturized and lightweight sensors have changed the conventional hard and bulk medical equipment to flexible form, allowing users to engage in daily activities while having sensors equipped or implanted. This simple change has provided continuous data acquisition of human conditions for the high yield of diagnosis. However, the continuously measured physiological/electrophysiological signals from a body-interfaced device are inevitably embedded into physical noise from body movement. This is because the body always moves by the skeletal system, which is controlled by muscle contraction and relaxation. Therefore, human-interactive devices that are closely placed to the body are also affected by skin and muscle movement. Consider that, in many cases, the transducer of the sensors is supposed to be deformed by the amount of the targeted physical quantity. However, unintended deformation of the sensor induces unexpected tensional or compressive force, generating errors. For these reasons, dynamic motion always diminishes the quality of the data, and reduces the signal-to-noise ratio. As an example, a tactile sensor aims to recognize touch. However, dynamic movement is also a factor that generates signal change without touch, and results in the user recognizing a virtual touch that does not really exist. The canceling of dynamic noises and extraction of the target signal are essential issues in body-interfaced devices.<br/>The most common method is signal processing for noise canceling. However, this requires additional data processing, which potentially incurs signal loss of the original data. Alternatively, unconventional signal filtering using structures, materials, or designs of electrodes has been suggested. Representative examples are bending-insensitive devices using neutral planes, unidirectional sensors with precisely aligned patterns to have rather high sensitivity in one direction, and stretching insensitive sensors using wavy or origami forms that enable the sensors to change along with the deformation, while maintaining their performance. However, these methods concentrated on a single dynamic effect, and they also faced other problems. To resist dynamic movement, the sensor should be resilient to various deformations, like stretching, compression, and bending. The neutral plane method has high effectiveness for bending insensitivity, but no resistivity to stretching and compression. Unidirectional sensors are difficult to expand to various sensors, because they need to accompany processes that are of complicated structure, or difficult. In the case of origami and wavy 3D structures, some operating space over the whole device is required to properly work, and it is barely used for implantable devices.<br/>Notably, motion-induced noise operates through the body surface direction, and the film shape sensor has varying sensitivity according to the orientation of the device. To eliminate dynamic motion artifacts more efficiently and universally, in this research, the film sensor is placed in a position where motion noises apply to a relatively insensitive direction, while letting the target stimuli apply to the sensitive area. The principles of the unidirectional sensors mentioned above are expanded from 2D body surface to 3D space, and the study is conducted on motion artifact-minimized sensors that do not require specifically aligned patterning to take a strict process and limit the variety of fabrication. The study aims to show sensors that focus on human-interactive devices while showing insensitivity to 75 % stretching, 25 % compression, and 5 mm radius of curvature bending, while maintaining the sensing ability of the sensor. As an demonstration, the pulse signal on a wrist is well measured while the hand is rotating, and artificial skin implanted on rat can distinguish external pressure from the effect of movement noise.