This study has found that there is a good correlation of normalized regional particle deposition (nC/P) by gas ventilation versus transmission scan for both healthy (>80% FEV1
predicted) and CF lungs (>50% FEV1
predicted), even though there was an absolute difference in values. Although the absolute values for nC/P with both groups of subjects were different for the two methods, XV and TT, the good correlation indicates that, for most of the 2D scintigraphic analyses performed using either gas equilibrium ventilation images or radioactive transmission images as references for lung and chest geometry, direct comparisons between the two may be made using simple corrections. For example, it should be possible to use the regression equations found in and to relate nC/P data generated from the use of the XV or TT methods, such as for a meta-analysis, if the definition for lung outlines used in the analysis are similar to those presented here. Some deviation from these ROI definitions should still result in a good correction due to the insensitivity of nC/P to ROI shape and size.(5)
Based on the two arguments presented and discussed below, i.e
., normalizing C/P and the tight correlation between XV and TT methods, we propose that the equations of and can be used to compare population data of regional lung deposition generated from either of the two methods.
Although the size of the ROIs in this study tended to be less for TT versus XV scans for either the healthy subjects or CF patients, it should be emphasized that our patients had relatively mild lung disease (mean FEV1 % predicted=77). In patients with more severe disease where ventilation may be almost absent from large portions of the lungs, the size of ROIs determined by TT may, in fact, be larger than by XV, being more representative of the true lung outline. As shown in , apices in CF patients can have reduced ventilatory capacity in that region of the lung. This may reduce the size of the ROI by truncating or attenuating the edges, although in this case there was still sufficient activity for the ROI to include much of this region. Transmission imaging is somewhat less sensitive to the ventilation reduction, as seen in . As our ROIs were set at boundaries of 20% of peak xenon counts or at 3×soft tissue background in transmission images, the TT ROIs averaged smaller than the XV ROIs.
It is interesting to note that the nC/P values of regional particle deposition for the two methods correlate so well despite the density patterns of the reference images being very different. The C/P of the reference image for the TT method was lower than that for the XV method for both study groups (), indicating a much more curved structure by gas ventilation, or alternatively, a flatter TT structure, yet the nC/P for the deposition images for the two methods correlates very well. This is likely due to the fact that the difference in nC/P is predominately due to differences in the reference C/P, whereas the C/P for the deposition images is very similar between the two methods ().
For the normal subjects, the area of the ROI was highly correlated with FVC, nearly to the same degree for both the XV and TT methods. In contrast, the ROIs for the CF patients were poorly associated with FVC, most likely due to the degradation of lung function with the disease, but a stronger correlation was observed for the TT method. For these subjects, poor lung function results in poor xenon gas ventilation that alters the XV outlines due to nonequilibrium conditions. For this reason, if the results can be inferred to other subjects with compromised lung functions such as asthma and chronic obstructive pulmonary disease, the TT method would be a better choice for defining lung outline and calculating a regional deposition parameter such as C/P in patients with lung disorders.
The use of the isocontour outline of the lung periphery, rather than a rectangular outline, reduced the variability in nC/P between the XV and TT method, especially in lungs of CF patients whose ventilation is compromised, making a better comparison between the two methods. The higher variability with the rectangular ROI using the TT method is presumably due to the high variability associated with the nonlung corners, which is eliminated when using the isocontour outline.
Compared with gas ventilation scans, transmission scans with either 57
Co or 99m
Tc are the easiest and most accessible method for defining the lung borders, because all laboratories undertaking deposition studies should have access to a flood-field source for quality control purposes. An additional advantage of the transmission scan over the xenon ventilation scan is the lower radiation dose received by the subject. Our estimation of radiation dose to the lungs for the xenon scan for 5
min of ventilation with 10
mCi of 133
Xe is approximately 65 mRem. For the 90-sec transmission scan with 2
mCi of 99m
Tc, it is estimated to be less than 3 mRem. Although the results presented here used 99m
Tc rather than 57
Co for the transmission scans, differences between 57
Co and 99m
Tc transmission scans are expected to be negligible, with linear attenuation coefficients changing from 0.15 to 0.14
The C/P values reported here are specific to the shapes and sizes of ROIs derived from the criteria of our analysis technique. The method for determining the rectangular central ROI was chosen to allow for reproducibility. The counts in the central region, with its high mean count per pixel, may be very sensitive to which pixels are included. The ambiguity of central ROI shape and position is reduced by making the central ROI a simple geometric fraction of its rectangular whole right lung ROI with unambiguous placement in relation to the whole right lung ROI. In this study, the same rectangular central ROI defined for each subject was used for both the rectangular and isocontour whole right lung analysis. As seen in the values in , the difference in values for nC/P for the two methods can be minimized by choosing a whole right lung isocontour outline in combination with a rectangular central region. However, differences between the XV isocontour method and the TT isocontour method persist in the data based on the CF population. Biddiscombe et al
have recently shown that the size of the central (C) versus whole lung (WL) ROI can clearly affect the magnitude of the C/P for the deposition image alone (i.e
., non-normalized C/P). But when the nC/P is determined by Equation 1
(normalized to a reference image), the relative size of C versus WL is much less important, emphasizing the importance that the C/P indices be normalized by reference images (either gas or transmission scans). The tight correlation of nC/P between the differently sized rectangular and isocontour outlines, for either XV or TT method, shown in and , confirms this observation.
The determination of the lung boundary in a 2D image created by a radioactive gas that fills the lung, such as 133
Xe in our XV method, is based on where the gas-containing region meets the solid tissue. In a logical sense, this is a determination based on where the “lung is detected,” i.e
., a top-down determination with respect to counts. On the other hand, a transmission image detects the lung by passing a planar radiation source through the whole body, attenuating most where the denser body tissue is located and less in the gas-containing regions. Therefore, our method for determining the boundary from a transmission scan is based on the assumption that “the lung is where the solid tissue is not,” i.e
., a bottom-up determination, with respect to counts, to locate the tissue boundary. Complicating the generation and comparison of XV and TT lung boundaries is the fact that the XV counts from the lung are attenuated by one layer of body tissue, and the TT counts pass through two layers of tissue. Were it not for the fact that the lungs are tapered around the edges, setting the boundary would be elementary. However, a limit must be set at where the taper ends. We have used 20% of peak lung count as the isocontour limit for lung boundary for the gas ventilation method and 3× abdominal count as the isocontour for tissue boundary. Many studies set the limit even lower, e.g
., where the “radioactive counts decreased to background levels.”(5)
These are well used but somewhat arbitrary values and may need further evaluation in future studies.
Our findings show that transmission scans are similar to and can be easily related to gas ventilation scans for defining lung borders and lung thickness required for measuring regional particle deposition. Based on the results of this study, the following conclusions can be drawn: (1
) using a transmission scan instead of a ventilation scan to define the lung borders may limit problems associated with poorly ventilated lung regions; (2
) choosing a right lung isocontour outline in combination with a rectangular central region appears to be more precise in terms of defining the nC/P compared with choosing a right lung outline based on the rectangular method; and (3
) using simple corrections based on the appropriate regression equations of similarly generated data may allow direct comparison between nC/Ps based on ventilation versus transmission scans. These suggestions should be appropriate to other deposition parameters, such as nP/C or penetration index.