Approaches and challenges in semi-automatic evaluation of NiTi microstructure
The knowledge of NiTi microcleanliness is of significant importance to the fatigue properties of medical devices. For a long time, microcleanliness for NiTi milled product has been specified within ASTM F2063, however without a standardized definition on the methodology and a precision and bias statement.
Therefore, an ASTM work group is currently discussing approaches to the standardization of a test procedure. Key steps of this destructive test method include sample collection, sample preparation, image acquisition and image analysis, which must be evaluated separately with their respective influencing factors.
While small sample sizes might be sufficient for random sampling during quality assurance, a deeper understanding of larger areas and volumes is crucial during the development of new products or alloy quality grades. Moreover, semi-finished products further down the manufacturing line usually contain smaller non-metallic inclusions, that are more difficult to evaluate on a statistically significant basis.
The primary objective of this work is to evaluate approaches and limitations on the image analysis for large data sets with both AI-assisted tools as well as automated workflows for classical threshold value methods. An overview of different image analysis methods is given and necessary adjustments during sample preparation and image acquisition are discussed.