Large Scale Ultrafast Manufacturing of Wireless Soft Bioelectronics Enabled by Autonomous Robot Arm Printing Assisted by a Computer Vision‐Enabled Guidance System for Personalized Wound Healing

Background

Electrical stimulation can mimic or enhance the endogenous wound currents that guide cell migration and angiogenesis, but its clinical use has been limited by bulky hospital-based equipment, and patch-type devices must also adhere to tissue without damaging it on removal. Lithography and screen printing require masks and are poorly suited to wounds whose size and shape differ from patient to patient. Earlier in situ printing combined with 3D scanning used 3-axis systems that cannot easily reach wounds on the side of the body, and separate hand-held scanners need markers and coordinate transformations that add processing steps.

Design and approach

CARE is fabricated by INSIGHT, a system that mounts a depth camera and a pneumatic multi-syringe printhead on a 6-axis robot arm. The camera scans the body, the wound contour is extracted by color contrast with the surrounding skin, and a bounding box gives the wound width and length; curved or large wounds are captured from several angles and the point clouds are merged. The wound size is passed to a MATLAB script that sizes the interdigitated electrodes and the receiving coil so that the patch resonates with a 13.56 MHz excitation coil. Three inks, a silver conductive ink, a PDMS dielectric ink and a photocurable poly(glycerol-co-sebacate) acrylate (PGSA) adhesive, are then printed as five layers (coil, dielectric, electrodes, passivation and adhesive) on a flexible substrate on a hotplate at 80 °C, with the arm rotating to ±45° to switch between syringes.

How it was evaluated

Scanning accuracy was tested on wound phantoms with five skin and five wound colors and on curved phantoms imaged from three angles (0° and ±30°). Printed conductive lines were cycled under 50% bending and 0–10% stretching, adhesion was measured by lap shear on porcine skin, and extract cytotoxicity was tested with human dermal fibroblasts (HDFs). Wireless performance was assessed by electric field simulation, inductance measurement and the transmission coefficient S21. Cellular effects of a 20 Hz pulsed radio frequency signal on a 13.56 MHz carrier were examined in HDFs and green fluorescent protein-expressing human umbilical vein endothelial cells (GFP-HUVECs) through proliferation, migration, tube formation, CD31 expression and growth factor secretion. Fit to an irregular wound was demonstrated ex vivo on the dorsal skin of a discarded mouse.

Key results

Pre-printing processing rose only from 13.9 s to 17.0 s as the wound size increased from 3 cm × 3 cm to 8 cm × 8 cm, and scanning finished within 20 s regardless of size. Width and length were recognized with less than 2.6% error across all 25 skin and wound color combinations, and multi-angle scans of curved phantoms gave 94% to 99% accuracy. Printing a patch took about 11 min for a 3 cm × 3 cm wound and approximately 42 min for an 8 cm × 8 cm wound, and the patch could be attached after 1 min of UV curing. Coils were also printed on curved substrates and on a substrate tilted at 45 degrees.

The resistance of printed conductive lines remained stable over 1000 bending cycles at 50% strain and increased gradually by up to 10% as tensile strain rose from 0 to 10%, returning to its original value on release. The PGSA adhesive gave a lap shear strength of 5.24 kPa on porcine skin, which the authors described as five times that of commercial fibrin glue (1.09 ± 0.04 kPa), and extracts at 100%, 50% and 25% concentration showed no significant difference in HDF viability.

Simulations indicated a field of 1.37 V/mm through the passivation layer and 88.8 mV/mm at 10 mm above the electrodes, above the 25 mV/mm level cited as promoting cell migration. The measured S21 was approximately 10 dB at 13.56 MHz, and about 3 V reached the patch when 10 V was applied to the external coil. For a 1.5 cm square wound, the target, optimized and measured coil inductances were 2.27, 2.03 and 1.75 μH; the authors attributed the lower measured value to removing half of the innermost turn to connect the electrodes.

In vitro, stimulation at a 99% duty cycle for 3 days yielded higher DNA content for both HDFs and GFP-HUVECs than controls. At a 20% duty cycle, migration was 1.86-fold faster for HDFs after 24 h and 1.64-fold faster for GFP-HUVECs after 8 h according to the Results, whereas the abstract gives factors of 1.6 for fibroblasts and 1.9 for endothelial cells. Tube formation after 6 h of stimulation produced over twice the number of junctions, CD31 expression increased 1.8-fold at a 99% duty cycle, and stimulated HDFs secreted more fibroblast growth factor (FGF) and vascular endothelial growth factor A (VEGF-A).

Open questions

The authors stated that the wound information was transferred manually to the MATLAB script and that a new algorithm is needed to design electrodes and coils automatically within the system. The excitation coil was on a rigid printed circuit board, and printing external coils on flexible substrates is required for translation. Future devices are planned to add temperature, impedance and pH sensors for closed-loop monitoring. Biological effects were evaluated in cell culture, with an ex vivo fit demonstration on a mouse.

How it connects to other research

A wireless closed-loop smart bandage combined sensing and stimulation with switchable adhesive hydrogel electrodes and was tested in mouse wound models, whereas CARE concerns how stimulation electrodes and receiving coils are dimensioned and printed to match an individual wound.