Blueprint for a Semi-automated Image Processing Tool to Characterize Stent Features: Application to a Pediatric Growth-Adaptive Stent.

Abstract

The goal of this work was to develop a general blueprint for a semi-automated image processing tool (SIPT) to measure small, complex features of stent prototypes that can replace the current gold standard of manual measurements. The stents were designed using CAD software and manufactured via laser cutting. Stent prototypes were imaged using a Keyence microscope in top and side view orientations. The SIPT algorithm was developed in MATLAB to extract and measure 4 dimensions of the stent (inner and outer diameter, spring bend outer radius, spring bend width). The same dimensions were also manually measured by an experienced metrology technician as a gold standard comparison. We successfully made over 5000 unique measurements across the 4 key dimensions of 15 stents using the SIPT algorithm. Compared to the gold standard manual method, SIPT reduced measurement time by nearly 90% and increased the total number of measurements captured by over 2300%. The two one-sided test and Bland-Altman analysis demonstrated that SIPT achieved equivalency against the manual method of measurement for all 4 dimensions. In summary, we found that our SIPT software could be used to replace manual measurements and provided substantial time savings with consistent accuracy. Overall, this paper presents a generalizable workflow to isolate and measure critical features of stent prototypes that we believe will provide a valuable, cost-effective tool to other medical device designers seeking to rapidly iterate on unique stent designs or other manufactured parts with small and complex structures.