implants0112

other optical system, have optical limits and owing toCT’sMTFandintrinsiclimits,CTscanscanbecon- sidered low-resolution 3-D images. They also achieve spatial resolution levels far from those needed in our field to ascertain placement preci- sion. Consequently, statistical inferences based on superimposition cannot be said to deliver valid proof. _High-contrast spatial resolution I scanned an implant using the latest NewTom CBCT(CB3DVG-IMARK3),andviewedthescanus- ing SimPlant Crystal (Materialise Dental) to verify the resolution and the precision of the measure- ment. The best I was able to achieve was 0.1 mm. This means that a real measurement of 1.43 mm could be achieved on CT within 1.33 and 1.53 mm, and 0.3 mm is the possible measurement error (Fig. 17a). The same difficulties also arise with MSCT scans (Fig. 17b). Spatial frequency is evaluated by means of MTF, the ratio between the output and the input signal, with one describing an ideal system with no loss of informationattheoutput.MTFdefineslimitingres- olution, which describes the ability of a system to perceivetwoobjectsasdistinct.Athighfrequencies, that is a high number of line pairs per mm (lppm), MTF will approach zero (Figs. 18a & b). When taking MTF into account, we must evaluate a CT scan ac- cording to its optical performance. When the fre- quencyisincreased,aseriesofsquarewaves,corre- sponding to a 1:1 ratio with combined white and black lines, changes into a series of bell-shaped waves.Thisprocessistermedthepointspreadfunc- tion. As a result, the contrast decreases, which makes it increasingly difficult to visualise the edge ofthelines.MTFistheFouriertransformationofthe point spread function. When the frequency is low and the quality ratio is one, the wave corresponds perfectly to the square waves. When the frequency increases,theratiodecreasesandthewavebecomes increasingly bell shaped. At an MTF of 2%, the im- agewillbeofauniformlygreycolour(Figs.18c&d). The CT scan limiting resolution is therefore 2 lppm at best (Fig. 18e). _Low-contrast spatial resolution Moreover, we can extend our discussion to the contrast level at which an image is observed and analyse low-contrast spatial resolution.6 When the contrast decreases at high frequencies, we have to cope with a low-contrast level image that is noise dependent. Furthermore, the optical spatial resolu- tion properties depend on the part of the screen at which we are looking. The resolution is at its best at theisocentre,worseningbothintheradialdirection and along the circumference, the azimuthal direc- tion(Fig.19). Whilethisphenomenonholdstruefor the cone beam in particular, a cone-beam effect is also achieved with MSCT: the more slices we have, that is, the greater the fan beam width of each sub- sequent MSCT scan, the greater the cone-beam ef- fect(Figs.20a&b).Whentheisocentreisconsidered the central part of the radiation fan, this effect can be seen in the outermost slices of the radiation fan beam especially (Fig. 20c). Axial reconstruction al- gorithmsreportthiscone-beameffectinrelationto a spiral path in the axial images (Fig. 20d).7 Compensating cone-beam reconstruction algo- rithms or spiral interpolation algorithms help to solve this problem, for instance the multi-row Fourier reconstruction. Similarly, an extension of theadvancedsingle-slicerebinningmethod(ASSR), which combines the idea of ASSR with a z-filtering approach, has been proposed as a solution to this problem, but its validity has not been adequately I case study Fig. 18a–e_MTF concept. 24 I implants1_2012 Fig. 18bFig. 18a Fig. 18c Fig. 18d