Development and Evaluation of an Automated Histomorphometric Analysis Method for the Assessment of Implant Osseointegration

Abstract

This study evaluated an automated histomorphometric analysis method for assessing implant osseointegration. Sixty-eight histological images of implants installed in the tibiae of 40 rats were analyzed using the manual method and ImageJ software to evaluate the %BIC (bone-implant contact) and %BBT (bone between the implant threads). In the automated histomorphometric analysis method, the similarity of the RGB color channels was analyzed, and the black background region of the image was excluded, to select the region of interest. The contour curve was extracted, excluding points that do not belong to the contact between the implant and the bone, i.e., the image edges. A Gaussian filter was applied to smooth small imperfections in the curve. An analysis of the curve points in relation to their neighboring points was conducted, and the local maximum of the curve was selected, excluding points that were too close together and keeping only those that represent the peaks of the implant curve. The peaks and the corresponding portion of the curve were used to select the region of the original image and calculate the %BIC and %BBT values. A comparison of the results obtained for both parameters in both evaluation methods was then performed using Pearson’s correlation test. A strong correlation was observed between the analyses (r = 0.99). Thus, the proposed automated histomorphometric analysis method was shown to be reliable for assessing implant osseointegration.