558
S. Grigg et al.
location, shown clearly in Fig. 49.2 where some regions are very dense and others
very weak. Other factors will also have affected the result. The axial load recorded
will have been affected by the force applied by the operator as was the case with
the Sawbones testing. The load cell itself is also likely to not have been completely
accurate, this is due to factors such as the angle of the hole relative to the surface,
the fact that the surface itself is not flat and was covered in a non-uniform thin layer
of tissue. These factors all affect the distribution of load over the cell, decreasing its
accuracy.
Comparing the linear regression models of Sawbones and cadaveric testing
showed a good relationship for the large fragment set was very good. The error
could be a result of the aforementioned issues with acquiring load during the cadaveric testing, or the conversion of HU to BMD being inaccurate. The small fragment
set testing produced significantly different models between Sawbones and cadaveric
testing. The reason for this is not fully clear, however most likely due to variations
in the BMD across the bone being more influential on the screws smaller tooth area.
It was proposed by Pullin et al. [12] that cumulative AE energy could be used as a
stop criteria to prevent stripping of the screw. For this to be possible AE energy given
off would need to be constant no matter the density of the bone. Figure 49.7 shows
that this is not the case in Sawbones or cadavers. The linear regression models show
an increase in AE energy recorded with increase BMD for all test configurations,
however it is always with a low correlation. Although this result is not positive, there
is a flaw in the testing that could have caused this; the testing was performed by hand.
This means that the rate at which the screw was tightened will have varied. It also
meant that tightening had to be stopped every half a turn, in some cases the screw
had load applied and stripped within one turn, in others it spanned multiple turns;
this will have influenced the AE acquired. One way to combat this would be to test
with a constant rate machine driven device. Collecting the torque in the screwdriver
would also be beneficial, as it is not dependent on some of the factors mentioned.
The frequency results presented in Fig. 49.8 show that peak frequency is dictated
by the set used, a peak of 220 kHz is seen on the small fragment set and 230 kHz when
using the large fragment set, be it when inserting screws to Sawbones or cadavers.
A lot of differences in frequency component can be seen between cadavers and
Sawbones. One obvious difference is the significantly lower low frequency component within the Sawbones compared to the cadavers testing. This additional low
frequency is likely the reason for the higher cumulative energy within the cadaveric
testing. The Sawbones however have a strange trend in their higher (300–500 kHz)
region; as BMD increases as does the average high frequency component, as shown
in Fig. 49.9. This strong correlation would enable a prediction of the BMD, and
so required torque for optimal tightening, to be made in the initial screw turning
process, prior to load being applied. If this could be applied within cadavers it could
stop surgeons overtightening and stripping screws. Unfortunately, this relationship
is not present within cadavers and it is unclear why, however if a similar relationship
could be found it would be very beneficial.
S. Grigg et al.
location, shown clearly in Fig. 49.2 where some regions are very dense and others
very weak. Other factors will also have affected the result. The axial load recorded
will have been affected by the force applied by the operator as was the case with
the Sawbones testing. The load cell itself is also likely to not have been completely
accurate, this is due to factors such as the angle of the hole relative to the surface,
the fact that the surface itself is not flat and was covered in a non-uniform thin layer
of tissue. These factors all affect the distribution of load over the cell, decreasing its
accuracy.
Comparing the linear regression models of Sawbones and cadaveric testing
showed a good relationship for the large fragment set was very good. The error
could be a result of the aforementioned issues with acquiring load during the cadaveric testing, or the conversion of HU to BMD being inaccurate. The small fragment
set testing produced significantly different models between Sawbones and cadaveric
testing. The reason for this is not fully clear, however most likely due to variations
in the BMD across the bone being more influential on the screws smaller tooth area.
It was proposed by Pullin et al. [12] that cumulative AE energy could be used as a
stop criteria to prevent stripping of the screw. For this to be possible AE energy given
off would need to be constant no matter the density of the bone. Figure 49.7 shows
that this is not the case in Sawbones or cadavers. The linear regression models show
an increase in AE energy recorded with increase BMD for all test configurations,
however it is always with a low correlation. Although this result is not positive, there
is a flaw in the testing that could have caused this; the testing was performed by hand.
This means that the rate at which the screw was tightened will have varied. It also
meant that tightening had to be stopped every half a turn, in some cases the screw
had load applied and stripped within one turn, in others it spanned multiple turns;
this will have influenced the AE acquired. One way to combat this would be to test
with a constant rate machine driven device. Collecting the torque in the screwdriver
would also be beneficial, as it is not dependent on some of the factors mentioned.
The frequency results presented in Fig. 49.8 show that peak frequency is dictated
by the set used, a peak of 220 kHz is seen on the small fragment set and 230 kHz when
using the large fragment set, be it when inserting screws to Sawbones or cadavers.
A lot of differences in frequency component can be seen between cadavers and
Sawbones. One obvious difference is the significantly lower low frequency component within the Sawbones compared to the cadavers testing. This additional low
frequency is likely the reason for the higher cumulative energy within the cadaveric
testing. The Sawbones however have a strange trend in their higher (300–500 kHz)
region; as BMD increases as does the average high frequency component, as shown
in Fig. 49.9. This strong correlation would enable a prediction of the BMD, and
so required torque for optimal tightening, to be made in the initial screw turning
process, prior to load being applied. If this could be applied within cadavers it could
stop surgeons overtightening and stripping screws. Unfortunately, this relationship
is not present within cadavers and it is unclear why, however if a similar relationship
could be found it would be very beneficial.