Methods: Specimens were prepared on sound human third molars (<6 months extracted). Cavities were restored with Gradia® (GC) composite and either Clearfil® SE (Kuraray) adhesive (Class V) or MaxCem® (Kerr) composite cement (Class II). After 5000 thermocycles (5-55°C) specimens were infiltrated (50% AgNO3, 24h). After sectioning, light-microscopy microleakage scores (range 0-4) were assigned. To reduce radiodensity, sections were demineralised (2 weeks EDTA) and embedded (Epon). Micro-CT scans (Skyscan1172, Skyscan) were taken (1) before sectioning at 100kV, 7.8µm pixel size (PS), (2) after sectioning at 85kV, 7.8µm, (3) after demineralisation/embedding at 50kV, 7.8µm, (4) after further size reduction at 50kV, 3.6µm, and (5) idem at 50kV, 1µm. Micro-CT data sets were reduced to maximum-intensity projections in the section planes (thickness=360µm) and regions of interest (ROI) close to restoration margins were selected. Grey-value histograms over the ROI were decomposed into Gaussian peaks (PeakFit 1.2).
Results: In conditions (1)-(2), no histogram peaks attributable to Ag could be discerned. In conditions (3)-(5), Ag peaks appeared as shoulders in the peak caused by attenuation from demineralised dental tissue, and average relative error on Ag peak areas decreased from 0.030 (PS 7.8µm) to 0.007 (1µm) (Pearson, n=9, r=0.69, p<0.05).
Conclusion: When micro-CT of silver tracer is used to visualise microleakage, specimen demineralisation and a pixel size of 3.5µm or less are recommended. By extrapolation of the observed trend, micro-CT with sub-micrometer resolution can be recommended for detailed quantitative 3D-study of microleakage. Supported by K.U.Leuven Grant OT/06/58.