A simple manual method to estimate water-equivalent diameter for calculating size-specific dose estimate in chest computed tomography

Abstract

Objectives:

The American Association of Physicists in Medicine (AAPM) Task Groups (TG) 204 and 220 introduced a method to estimate patient dose by introducing the Size-Specific Dose Estimate (SSDE). They provided patient size-specific conversion factors that could be applied to volumetric CT Dose Index CTDI_vol to estimate patient dose in terms of SSDE based on either effective diameter (D_eff) or water equivalent diameter (D_w). Our study presented an alternative method to manually estimate SSDE for the everyday clinical routine chest CT that can be readily used and does not require sophisticated computer programming.

Methods:

For 16 adult patients undergoing chest CT, the method employed an average relative electron density (ρ_e^lung = 0.3) for the lung tissue and a ρ_e^tissue of 1.0 for the other tissues to scale the lateral thickness and compute the effective lateral thickness on the patient’s axial image. The proposed method estimated a “corrected” D_eff (D_eff^corr) to replace D_w and compared results with TG220 and a second method proposed by Huda et al, for the same set of CT studies.

Results:

The results showed comparable behavior for all methods. There is overall agreement especially between this study and TG220. Largest differences were +13.3% and+15.9% from TG220 and Huda values, respectively. Patient size correlation showed strong correlation with the TG220 and Huda et al methods.

Conclusions:

A simple, quick manual method to estimate CT patient radiation dose in terms of SSDE was proposed as an alternative where sophisticated computer programming is not available. It can be readily used during any clinical chest CT scanning.

Advances in knowledge:

The paper is novel as it presents simple, quick manual method to estimate CT patient radiation dose in chest imaging. The process can be used as alternative in cases no sophisticated computer programming is available.

Introduction

The latest National Council on Radiation Protection and Measurements (NCRP) Report 160 reports that medical exposure to patients is one of the largest sources of radiation dose to Americans, with CT being the largest contributor to this exposure.¹ The latest United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR) 2008 Report also states that CT is the major contributor to the collective patient dose.²

The latest European Basic Safety Standards (BSS) on radiation protection³ and the International BSS from the International Atomic Energy Agency (IAEA)⁴ have underlined the need to monitor and optimize radiation dose in CT examinations. More specifically, IAEA BSS report clearly states that medical X-ray equipment must have a means to inform the practitioner of the relevant parameters for assessing the patient dose and even more important to have the capacity to transfer this information to the record of the examination.

The dose indices that were used for many years to monitor radiation dose were the volumetric Computed Tomography Dose Index (CTDI_vol) and Dose–Length Product (DLP).⁵ CTDI_vol was developed to compare CT radiation output levels between CT scanners using a reference phantom, and DLP is the product of CTDI_vol and total scan length of each patient. These dose indices cannot be used to assess accurately patient dose, as they are both independent of patient size. In 2011, the American Association of Physicists in Medicine (AAPM) Task Group 204 (TG204) proposed a more accurate method to estimate patient dose by introducing the Size-Specific Dose Estimate (SSDE).⁶ The TG204 developed a set of patient size-specific conversion factors (f_size^16X for head examinations and f_size^32X for body examinations) that could be applied to CTDI_vol to estimate patient dose. More specifically, the report described a size metric that involved the physical dimensions of the patient ((anteroposterior (AP), lateral (LAT), AP +LAT, or effective diameter (D_eff)), in combination with scanner output (CTDI_vol), to determine SSDE. These factors were limited to patient geometric size without taking into account the X-ray attenuation within the body. The methodology did not consider the body region attenuation characteristics (lung region in the CT chest exam attenuates the X-rays less compared to similar geometrical size of the abdomen area in the abdominal CT). The issue of patient attenuation was addressed in more recent publications in 2012.⁷ One of these introduced the water equivalent diameter (D_w) and its use in determining object radiation dose.⁸ D_w and its relation to actual patient dose was further elaborated within the latest AAPM Task Group 220 (TG220) in 2014.⁹ The main objective of this TG was to develop a sound metric for automatically estimating patient size that would account for patient attenuation and allow routine determination of SSDE for all patients, with little or no user intervention. The report described methodologies to calculate D_w from either a CT image or a CT localizer radiograph image. The first method requires extensive manual intervention or automatic segmentation algorithms. The second method uses the CT localizer radiographs which are generated after the patient is positioned on the CT table, before the actual CT scan and form the anatomical basis for the tube current modulation for the specific scan (automatic adjustment of tube current during a single rotation or part of the rotation of CT tube around the patient taking into account patient anatomy in each direction of the tube). TG220 extensively describes several important issues that may limit the accuracy of this approach.⁹

Our study describes an alternative and practical method for estimating patient “corrected” effective diameter based on the computation of the “radiological depth” or “radiological thickness” in the presence of tissue inhomogeneities, such as the chest (chest) area that includes lung tissue. This radiological thickness will facilitate the “correction” of the patient effective diameter, defined as D_eff^corr, which will be equivalent to the TG220 D_w.¹⁰ The water-equivalent tissue thickness in the LAT dimension on the CT image can be obtained by scaling the tissues involved with the appropriate relative electron density of the tissue.^11–14 Similarly, the AP dimension can also be scaled if needed, however this was not performed in our study. It practically provides a simple calculation of SSDE that can be done without using sophisticated software or cumbersome methodology, easily applicable during the everyday clinical routine, complementary and potentially alternative to the AAPM TG204/TG220 methodology.

Methods and materials

Sixteen (16) adult patients (female and male) who underwent chest CT examinations were selected with their CT studies anonymized. The study was conducted in a collaboration between the University of Pennsylvania, Perelman School of Medicine, Philadelphia, United States of America (USA) and the Konstantopoulio Hospital, Medical Physics department, Athens, Greece. The CT scanners and scanning protocols used are listed in Table 1. The mid-thoracic CT image for these patients was identified from the multiplanar reconstructions (sagittal or coronal views) or the CT localizer radiograph and, the axial slice was selected for each patient above (example is shown in Figure 1a). The water-equivalent tissue thickness in the LAT dimension on the CT image can be obtained by scaling the tissues involved with the appropriate relative electron density of the tissue.^11–14 Similarly, the AP dimension can also be scaled if needed. The relative electron density for lung tissue used was ρ_e^lung = 0.3 and for all other tissues ρ_e^tissue = 1.0, both being values well-reported in literature.^13,15,16

Table 1.

Konstantopoulio Hospital CT scanners and chest scanning protocol of the study

CT scanner	Type	kV	Recons action thickness (mm)	Pitch	Rotation time (sec)	Collimation (mm)
Canon Medical systems, USA	Asteion 4	120	5.0	1.5	0.75	4*3
Philips Medical systems, the Netherlands	Brilliance 64	120	2.5	0.891	0.50	64*0.625

Figure 1.

(a) The sagittal and coronal reconstructions (middle and right image) are used to identify the level of the mid thoracic image (left image), (b) The distances (lengths) of soft and lung tissues in the mid thoracic CT image are measured using the measuring tool of the CT scanner (d1, d2L, d3, d4L, (d5). These values are used to compute ${LAT}_{corr}^{eff}$ as per Eq. (4) in the text ($LA{T}_{phys}=d_1+d_{2L}+d_3+d_{4L}+d_5$).

The definition of radiological depth or thickness in a heterogeneous medium is defined as¹⁰:

$$d={\int }_{blue}^{yellow}\rho \left(r\right)dl$$ (1)

where, dl is the length element of integration and the integration limits are between the “blue” and “yellow” points in Figure 1b along the lateral path and ρ(r) is the medium density. For arbitrary, discrete heterogeneous medium with different segments, the above integral has been shown to be approximated as^12,14,17:

$$\mathrm{d}=\sum _{\mathrm{j}=1}^{\mathrm{N}\mathrm{s}}{\rho }_e^s{\left(\mathrm{j}\right)\mathrm{l}}_{\mathrm{s}}\left(\mathrm{j}\right)$$ (2)

for most substances including human tissues (muscle, fat, lung).¹² In Eq. (2), is the density of the tissue relative to water (called, “relative electron density”), N_s is the number of line segments along the dimension of interest, l_s(j) and ρ_s(j) are the length and physical density of the j-segment, respectively, with subscript “s” to refer to the tissue segments.

For the axial image at the mid-scanning range of the CT study, LAT thicknesses were measured manually with the distance tool of the scanner software (standard software tool provided by all CT manufacturers and all scanner types), along with the individual distances d₁, d_2L, d₃, d_4L, d₅ (all in mm) and in the middle of the image along the AP and LAT direction, as shown in Figure 1b. The subscript L refers to the distance within lung tissue. The total physical LAT_phys is given by:

$${\displaystyle LA{T}_{phys}=d_1+d_{2L}+d_3+d_{4L}+d_5(\mathrm{i}\mathrm{n}\mathrm{m}\mathrm{m}),}$$ (3)

To address the presence of various inhomogeneities along the LAT direction, calculation of the radiological distance instead of the physical one is required. Applying the above Eq. (2) to Figure 1b, the physical distance (Eq. (3)) is modified to the inhomogeneity-corrected effective distance (thickness) as follows:

$${\displaystyle \begin{array}{c}LA{T}_{eff}^{corr}=d_1+{\rho }_e^{lung}\cdot d_{2L}+d_3+{\rho }_e^{lung}\cdot d_{4L}+d_5=\\ d_1+d_3+d_5+{\rho }_e^{lung}\cdot \left(d_{2L}+d_{4L}\right)\end{array}(inmm),}$$ (4)

where, ρ_e,^lung = 0.30 is the relative electron density of lung tissue relative to water (averaged between subject’s ages, inhalation, exhalation and normal breathing modes) as this is reported in literature.^13,15,16 Due to limited low density inhomogeneities in the AP direction, $A{P}_{eff}{}^{corr}\cong A{P}_{phys}$, is used in this study. To account for inhomogeneities, we propose the modification and replacement of the patient D_eff⁶ with the proposed D_eff^corr to be equivalent to D_w⁹ to be made as follows:

$${\displaystyle D_{eff}^{corr}\left(=D_w\right)=\sqrt{A{P}_{phys}\cdot LA{T}_{eff}^{corr}}(\mathrm{i}\mathrm{n}\mathrm{m}\mathrm{m}),}$$ (5)

One other method of estimating D_eff that accounts for tissue inhomogeneities in the thoracic region has been proposed by Huda et al¹⁸ is given by, D_Huda:

$${\displaystyle D_{HUDA}=2\times 0.428\times \sqrt{A{P}_{phys}\cdot LA{T}_{phys}}=0.856\times D_{eff}(\mathrm{i}\mathrm{n}\mathrm{m}\mathrm{m}),}$$ (6)

Note that, $D_{eff}=\sqrt{A{P}_{phys}\cdot LA{T}_{phys}}$ is the effective diameter of the patient, as defined in TG204,⁶ while the correction to the D_eff is a simple multiplicative factor independent of the size of the inhomogeneity.

Considering attenuation only along central axis, the same axial image was used to apply the TG220 proposed method for determining the D_w as⁹:

$${\displaystyle D_w=2\times \sqrt{\left(\frac{RO{I}_{mean}}{1000}+1\right)}\cdot \sqrt{\left(\frac{A_{ROI}}{\pi }\right)}(\mathrm{i}\mathrm{n}\mathrm{m}\mathrm{m})}$$ (7)

where, region of interest (ROI) ROI_mean and A_ROI are the mean CT number and the total area of the axial image ROI, respectively. To calculate ROI_mean and A_ROI, the free-hand ROI tool in the analysis utility of the scanner was used. The outline of the axial slice was drawn as shown in Figure 2. Once the area was drawn, the scanner software tool automatically calculated ROI_mean and A_ROI values. D_w values were calculated for the 16 patients included in the study using the three different methods a) current proposed methodology, b) TG220⁹ and c) Huda et al method.¹⁸ The results of these calculations and comparison of the three methods is shown below.

Figure 2.

An axial CT image selected at the mid-scanning range of a patient in our study that shows the individual measured PA and LAT distances. LAT, lateral.

Results

Computation of the patient diameter based on the three methods and percent differences of our method compared to that of TG220 and that of Huda are show in Figure 3. The scanner software distance tool has a 0.1 mm precision for all distances measured and the relative electron density has a SD = 0.04.¹⁶  The statistical variation of the patient diameter for each method within the 16 patient sample, is shown in Table 2. The figure shows comparable behavior for all three methods (proposed method: blue line diagram, AAPM TG220: orange line diagram and Huda method: gray line diagram). There is an overall agreement between the features of the data plotted, especially between our values (blue data) and the TG220 ones (orange data), as shown in Figure 3a. For most patients in the study, the proposed method results in patient diameter values (D_eff^corr) that are slightly smaller than the D_w(TG220) values, with patient diameter values of the Huda et al. (D_HUDA) method having the largest variation compared to the other two methods.

Figure 3.

a) Computations between D_w (TG220), D_eff^corr (this work) and D_HUDA methods are shown. (b) Percentage difference between our method (D_eff^corr) and TG220 and Huda methods for the 16 patients in the study (the %diff is calculated as: $\mathrm{\%}\mathrm{}\mathrm{d}\mathrm{i}\mathrm{f}\mathrm{f}=100\bullet \left[\right(\mathrm{T}\mathrm{G}220\mathrm{o}\mathrm{r}\mathrm{H}\mathrm{U}\mathrm{D}\mathrm{A}-\mathrm{t}\mathrm{h}\mathrm{i}\mathrm{s}\mathrm{}\mathrm{S}\mathrm{T}\mathrm{U}\mathrm{D}\mathrm{Y})⁄\mathrm{T}\mathrm{G}220\mathrm{o}\mathrm{r}\mathrm{H}\mathrm{U}\mathrm{D}\mathrm{A}]$) are presented. (See, statistical analysis in Table 2).

Table 2.

Statistical variations of the patient diameter computed by the different methods (sample size: 16)

Patient diameter (mm)	Mean	Median	SE	SD	Min/Max	SV	CL (95%)
Deff (TG204)	302.5	303.3	8.79	35.17	253.4/358.9	1236.82	18.74
Dw (TG220)	263.9	255.9	9.78	39.12	191.8/343.1	1530.59	20.85
DHUDA (Huda et al.)	258.9	259.7	7.53	30.10	216.9/307.2	906.26	16.04
Deffcorr (this Work)	248.6	245.2	9.06	36.26	197.3/310.3	1314.52	19.32

CL, confidence level; SD, standard deviation; SE, standard error; SV, sample variance.

Figure 3b shows the percentage differences between the proposed method and those of TG220 and of Huda. The gray bars represent the % difference of current method compared to the Huda et al, method. The orange bars represent the % difference of current method to the TG220 methodology. Largest differences were +13.3% and +15.9% from the TG220 and the Huda values, respectively as shown in Figure 3b.

Patient size correlation was also investigated by plotting the patient diameter computed with of each of the three methods (D_eff^corr, D_w, D_HUDA), in the Y-axis against the values of D_eff (from TG204), in X-axis, as shown in Figure 4. Results of our method (strong correlation: R² = 0.8713) line up well with the TG220 ones (R² = 0.7689) in a parallel pattern indicative of mutual correlation (our method vs TG220), as well. The Huda et al method linear fit is almost exact since the method depends on a single multiplicative factor to D_eff but is positioned between our method’s and TG220 method’s linear fit lines, going from smaller to larger patient diameters. Similar correlations have been noted by other investigators especially in the chest area.^19,20 The mutual correlation (R² = 0.8803) that verifies the equivalency of D_eff^corr (proposed method) to D_w (TG220) is shown in Figure 5. A possible reason for the noted differences between our proposed method and the TG220 one may be, the way that each method accounts for the inhomogeneity. The proposed method is based on linear scaling related to one-dimensional photon attenuation instead of a two-dimensional attenuation correction.

Figure 4.

D_eff^corr show strong correlation with D_eff (of TG204) for the values of our study (blue points and line: R² = 0.8713, 95% CI, SD = 13.46). Comparison of equivalent relationship with the values of D_w of TG220 (orange points and line: R² = 0.7689, 95 % CI, SD = 17.50) and D_Huda (gray points and line: R² = 1.00, 95% CI, SD = 4.27×10⁻⁵), in the chest area, is shown. CI, confidence interval; SD, standard deviation.

Figure 5.

D_eff^corr show strong correlation (R² = 0.8803, 95% CI, SD = 12.98) with D_w for the patients in our study that confirms the equivalence of the proposed scale-corrected D_eff^corr (with the relative electron density via Eqs. (4) and (5)) diameter, with the water equivalent diameter (D_w), derived by TG220.

The computation of the SSDE according to TG204⁶ and TG220⁹ involves the CTDI_vol reported for the 16 cm or 32 cm diameter dose index phantom and the patient size-specific conversion factor, f_size(D_eff) and f_size(D_w) respectively via the relationships (same relationships hold for f_size(D_eff) as well):

$${\displaystyle f_{size}\left(D_w\right)=3.70469\times e^{-0.03671937\times D_w}\mathrm{f}\mathrm{o}\mathrm{r}32\mathrm{c}\mathrm{m}\mathrm{d}\mathrm{i}\mathrm{a}\mathrm{m}.\mathrm{p}\mathrm{h}\mathrm{a}\mathrm{n}\mathrm{t}\mathrm{o}\mathrm{m}}$$ (8a)

$${\displaystyle f_{size}\left(D_w\right)=1.874799\times e^{-0.033871313\times D_w}\mathrm{f}\mathrm{o}\mathrm{r}16\mathrm{c}\mathrm{m}\mathrm{d}\mathrm{i}\mathrm{a}\mathrm{m}.\mathrm{p}\mathrm{h}\mathrm{a}\mathrm{n}\mathrm{t}\mathrm{o}\mathrm{m}\mathrm{a}\mathrm{n}\mathrm{d}}$$ (8b)

$${\displaystyle SSDE=f_{size}\left(D_w\right)\times CTD{I}_{vol}(\mathrm{i}\mathrm{n}\mathrm{m}\mathrm{G}\mathrm{y}),}$$ (9)

Alternatively, appropriate values for f_size(D_eff) or f_size(D_w) can be chosen from Tables 1 and 2 of TG204 or TG220. The relationships above (Eqs. (8) and (9)) can be applied to our proposed D_eff^corr and to D_HUDA to compute the SSDE. The variation of the conversion factor follows the variation of the patient diameter values between the three methods (D_eff^corr, D_w, D_HUDA) as shown in Figure 6, with the lower values of D_w to correspond to a higher value of the conversion factor.

Figure 6.

Conversion factors-f_size (Eq. (8a)) selected based on the computation of patient diameter by our method (D_eff^corr), TG200 (D_w) and Huda (D_HUDA) methods are drawn. X-axis is the patient number and Y-axis is the value of each conversion factor. The overall error in the f_size values was ± 0.001.

Discussion

The proposed methodology is a short and simple alternative to the TG220 methodology for the everyday clinical routine. It is based on the use of the axial CT image in the mid-scanning range of the CT scan, for measuring individual distances in the chest inhomogeneities (mostly lung) regions and scaling those distances with a value of ρ_e = 0.3. The rest of the tissue thickness except lung were scaled using a value of ρ_e = 1. All these distances can be added to estimate total LAT_eff^corr (it accounts for tissue inhomogeneities). The user would only need the distance tool of the CT scanner analysis utility. And she/he would not have to rely on the existence or use of the ROI tool (not all scanners have this tool and specially the free hand tool). The next steps for computing the D_eff^corr are identical to the steps found in TG220 with D_eff^corr replacing D_w.

Furthermore, it has been shown that the average SSDE computed from slice-by-slice application of TG220 method, for the entire scanning range of the patient agrees well with the value obtained using the single image at the mid-scanning range.^{21 22 23} Thus, it should be sufficient to apply our proposed method to the same central image (mid-scanning range) as well. Finally, our method can potentially be extended to other body areas that include tissue inhomogeneities such as the abdomen and pelvis if average relative electron densities for tissues are known. For example, in the pelvis area, the pelvic bones are the largest tissue inhomogeneity that will result in a water-equivalent diameter for the image larger than that of water or soft tissue.

There are a few limitations to our method, the most important being that the method is investigated only for chest CT. Limitations similar to those found in other methods, such as images with truncated anatomy (limited field-of-view) and different patient morphologies across and image (i.e. large tumor invading the mediastinum) that need to be addressed in the computation of the patient diameter, are valid in our study, as well. Those limitations will eventually under- or overestimate the SSDE. Correction methods suggested in literature for truncated patient anatomy in order to account for the attenuation of the missing anatomy,¹⁹ need to be employed. The existence of other devices in the image, such as a table surface, does not alter our results. Finally, our proposed manual method included data from only two CT manufacturers with one set of acquisition parameters for each. Due to the inherent simplicity in our method that needs only the lateral lung thickness to be measured, we concluded that can be used for any manufacturer and set of acquisition parameters. However, verification of this claim by direct comparison to the simple option of TG220⁹ is prudent.

Conclusions

A simple and quick computation method to estimate effective diameter or water-equivalent diameter primarily in the thoracic area, was proposed. The method was based on the use of relative electron density for lung to compute the effective thickness of the patient for the lateral dimension and use that to compute the patient effective diameter or water-equivalent diameter, before the f_size correction factor was selected from TG204 tables or formulas. The method was compared against TG220 methodology and other methods in literature by applying it in several clinical cases. The proposed method showed good correlation with the water-equivalent diameter computed by TG220 and can be used as a simple and fast alternative to estimating SSDE, without the need for automatic image segmentation or for ROI_mean estimation. The method is potentially expandable to other image areas, such as pelvis. The simplicity of our method makes it readily applicable to daily routine clinical CT chest examinations. It takes a few moments to identify the image of the mid-scanning range and measure the required distances through the lungs. The water equivalent diameter is then calculated with a calculator and the SSDE estimated with the available tabulated correction factors from TG220.⁹ The method can be used for example, during a CT dose audit or for any dose investigation reasons.

Keypoints

Alternative simple method to manually estimate patient dose in terms of SSDE for the everyday clinical routine in chest CT.