Background: As manual cell counting lacks objectivity in the assessment of positive marker cells in immunohistologic sections, there has been a shift to automated computer analysis solutions. However, quantifying inflammation around dental implants is still often done by manual cell counting.
Method: With mucosal sections stained against MRP8 harvested around dental implants, we developed an automated method (AM) to identify positive marker cells. In this proof-of-concept study, we developed a procedure for its validation on an exemplary data set. Therefore, the sections were also analyzed with the manual method (MM). Intrarater and interrater reliability as well as time analyses were conducted.
Results: The newly developed AM was based on a color deconvolution in the open-source software ImageJ2. We embedded the determination of the most appropriate filter setting into the systematic validation procedure, implementing the intraclass correlation (ICC) and the Bland-Altman bias (BA). The newly developed validation procedure carried out on the data set of this proof-of-concept study resulted in an excellent reliability of the AM (ICC = 0.97). Both the reliability and time analyses' results were in favor of the AM.
Conclusion: Our newly developed AM showed advantages in terms of repeatability and objectivity combined with a shorter duration. The detailed descriptions of its application and its validation procedure offers the opportunity to apply it for further immunohistologic questions. The prerequisite for the replacement of the MM is that the validation, carried out on a sufficient number of samples, leads to satisfactory results.
Keywords: Automatic cell counting; Computer-assisted image analysis; Digitalization; Peri-implant inflammation; Reliability; Validation.
Copyright © 2022 The Authors. Published by Elsevier Ltd.. All rights reserved.