| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Medical Physics Department, Agios Savvas Hospital, 171 Alexandras Ave., Athens
115 22, Greece.
2 Medical Physics Unit, Konstantopoulio Hospital, Athens, Greece.
3 Computed Tomography Department, Konstantopoulio Hospital, Athens,
Greece.
Received April 7, 2008;
accepted after revision May 28, 2008.
Address correspondence to I. A. Tsalafoutas
(j_tsalas{at}hotmail.com).
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. In each of the 12 interventions studied, a Kodak EDR2 film was positioned under the patient's anatomic area of concern. After processing, each film was scanned with a medical-grade scanner to produce a digital image from which the gray-scale profiles were obtained using the appropriate software. From these data and respective data from a series of calibration films, the entrance skin dose (ESD) profiles were derived. These ESD profiles were compared with the ESD profiles produced using a theoretic model and its revised version, which utilizes the DICOM data of each slice (i.e., tube potential, tube loading, slice thickness, slice location, pitch, and table height) and air-kerma output measurements from the CT tube.
RESULTS. In general, the ESD profiles calculated using the revised theoretic method were in better agreement with the profiles derived from the verification films than the ESD profiles derived from the original theoretic method. The deviations from the peak skin doses (PSDs) derived from the digital film images were within –3% and 27% of the PSDs derived from the verification films. The respective deviations of the ESD profiles calculated with the original theoretic method were quite larger.
CONCLUSION. The theoretic model provides a useful tool for estimating skin doses during CT-guided interventions with a reasonable level of accuracy. With further refinement and a little automation this method could be implemented for daily use.
Keywords: CT guidance • entrance skin dose • interventional procedures • medical physics • peak skin dose • radiation dose
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Although the major concern of interventionalists is the successful outcome of these procedures, the patient dose should always be of concern. Interventional procedures can result in effective doses (Es) and peak skin doses (PSDs) that are quite large compared with those resulting from simple radiographic procedures [1–3]. Thus, the possibility of stochastic and deterministic effects occurring cannot be ignored, and methods for estimating patient doses during these procedures are required.
Although information in the literature concerning patient dose calculations in CT interventions is rather limited, these procedures can also result in high patient doses as a result of the acquisition of a large number of images, most of which repeatedly expose the same piece of skin [4–7]. Indeed, in a recent study in which 14 biopsies, 14 radiofrequency ablations, 14 abscess drainages, and seven nephrostomies performed under CT guidance were studied, a maximum overlap of 37, 30, 15, and 15 slices over the same piece of skin was recorded, respectively, and the maximum PSD estimated was approximately 1 Gy [7]. In another study, the maximum PSDs estimated for the drainage and biopsy procedures studied were 1.61 and 1.44 Gy, respectively [6]. These skin doses are quite high, taking into account that transient erythema may occur at a threshold skin dose of approximately 2 Gy but has also been observed at 1 Gy [8, 9].
Patient dose calculations in diagnostic or interventional CT procedures require CT-specific dosimetric quantities, such as the CT dose index (CTDI) and the dose–length product (DLP), as well as appropriate conversion coefficients derived using Monte Carlo techniques to estimate the E [10]. For diagnostic CT examinations, DLP and E can be estimated using the ImPACT CT Patient Dosimetry Calculator (CTDosimetry.xls; henceforth referred to as CTDosimetry), which is a Microsoft Excel–based program freely available on the Internet [11].
In a recently published study, a method was presented for the retrospective calculation of DLP, E, and entrance skin dose (ESD) profiles in CT-guided interventional procedures [7]. The theoretic model and the respective computer program developed for this purpose (using Excel and Microsoft Visual Basic) used the DICOM data—that is, the tube potential, tube loading, slice thickness, slice location, and pitch—of the images stored in the CT department's PACS and the CTDosimetry.
The initial purpose of our study was to investigate the accuracy of the aforementioned model with respect to the calculation of the ESD profiles using film dosimetry [12, 13]. However, in view of the comparison results, the theoretic model was revised to take into account the effect of table height position on slice width and ESD. Thus, the ESD profiles derived from films were also compared with those obtained using the revised theoretic model.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In this CT department, all data from interventional procedures are routinely transferred and stored in the CT department's PACS using the DICOM format; thus, all the technical parameters used for the acquisition of each image are available. According to the aforementioned model [7], the ESD profile can be derived using the DICOM data, referring to the tube potential, tube loading, slice thickness, pitch, and slice location of each image, and assuming that the ESD from each slice is equal to mAs / 100 x nCTDIp within the geometric limits of the nominal slice width and zero beyond it. The nCTDIp is the periphery CTDI normalized per 100 mAs, and its value is given in CTDosimetry separately for the head and body phantoms for almost all commercially available CT scanners and their various operating tube potentials.
This method was again used to calculate the ESD profiles in 12 CT-guided interventions studied including seven biopsies and five radiofrequency ablations, all of which were performed in the abdominal or thoracic region. All slices corresponded to an anatomic position on the mathematic phantom included in the CTDosi metry to simulate human anatomy. In each of these procedures, a 35 x 43 cm (14 x 17 inches) radiation therapy verification film (EDR2, Eastman Kodak) was positioned on the CT table under the patient's anatomic area of concern. After the end of each procedure, the film was developed in the processor of the radiology department.
Calibration Procedure of EDR2 Films
For air-kerma measurements, a calibrated dosimeter (model 3036, Radcal)
equipped with a pencil-type ion chamber (model 10 x 5-3CT, Radcal) was
used. The ion chamber was positioned with the center of its active volume at
the gantry's isocenter and measurements, both with and without the table in
the scanning plane, were acquired. The dosimeter readings for exposure factors
120 kV, 100 mAs, and 10- and 7-mm collimation were divided by the nominal
slice width and expressed per 100 mAs to correspond to the normalized CTDI
values found in CTDosimetry. Because CTDI measure ments provide the line
integral of the air kerma along the length of the pencil chamber (i.e., the
air-kerma–length product [AKLP])
[14] and not the maximum
air-kerma value, air-kerma mea surements at the gantry isocenter were also
performed using a common calibrated digital dosimeter (PMX-III R/CT
multimeter, RTI Electronics) with a solid-state detector (R25, RTI
Electronics). The detector was positioned at the isocenter with the x-ray tube
locked in its upper most position (0°) during scanning performed with 120
kV, 100 mAs, and 10-mm collimation.
For calibration of the EDR2 films, six films were positioned on top of the CT table with the table height adjusted midway in the gantry and were scanned from 1 to 60 times in the same position with the tube rotating using 120 kV, 100 mAs, and 7-mm collimation. In each calibration film, two or three series of scans were obtained, allowing a distance of at least 10 cm among adjacent scans, to minimize the scatter contribution. The maximum optic density (OD) of the calibration films and the maximum OD of the films used in interventional procedures (henceforth referred to as patient films) were determined using a calibrated optical densito meter (RMI 331, X-Rite).
Calibration films and patient films were scanned using a medical-grade film scanner (Dosimetry Pro Advantage, Vidar) and commercial software designed for brachytherapy applications using a 300-dpi scanning resolution and 8-bit digital output. The resulting digital images (in TIF format) were exported to a CD-ROM and were then loaded to a PC equipped with commercial software that can read the brightness level of each pixel in an array defined on a digital image and can export these data in an Excel spreadsheet. The resolution of the digital images was adjusted to have pixel dimensions of 1 x 1 mm and 8-bit depth (256 gray levels).
The brightness level of each pixel decreases with the increase of its OD; therefore, it inversely varies with the incident dose. To overcome this problem, the brightness level values of all pixels were subtracted from the maximum brightness level value of the 8-bit gray scale, which ranges from 0 to 255, to derive the pixel values (PVs), which increase with the incident dose. Using the calibration films and the data from the air-kerma measurements, a mathematic formula was obtained for the translation of the OD and the PVs to ESD values (in mGy). The same formula was also used in patient films to convert the PV profiles to the respective ESD profiles. Because the patient films lack any positional information, to compare the ESD profiles from the digital images with those theoretically calculated, the location of the maximum ESD on the digital images was matched to the location of the maximum ESD value in the theoretically calculated profile.
The Revised Theoretic Model for the ESD Profile Calculation
The underlying assumptions of the original theoretic model used for the ESD
profile calculation [7] are
that the patient's trunk is similar to a 32-cm-diameter cylindric polymethyl
methacrylate phantom used for CTDI measure ments, and the phantom is
positioned in the center of the gantry. However, the shape of the CT images of
the trunk matches the shape of an ellipse more closely than that of a circle
of 32 cm diameter, as can be easily seen from the differences commonly
observed among lateral and anteroposterior (AP) diameters. As a result, the
focus-to-skin distance varies among patients and even varies for the same
patient across the body in the x-, y-, and z-axis
directions. Thus, even if the patient appears perfectly centered in the gantry
in one CT image, this is seldom valid for all the CT images acquired during a
diagnostic or an interventional procedure. To position patients with different
AP diameters centrally within the scanning plane, the table height must be
adjusted to a different value. Furthermore, for CT-guided interventions, the
table is often lowered during the interventional stage to facilitate the
manipulations required.
The aforementioned facts have a direct impact on the actual width of the x-ray beam entering the patient's body because the CT x-ray beam is divergent. Therefore, for two patients with dif ferent AP diameters scanned with the same slice thickness, the x-ray beam will have a smaller width in the skin of the patient with the larger AP diameter. Furthermore, the ESD will also be smaller for the larger patient for two main reasons: First, because of the intensity-shaping filter of the CT tube assembly (bowtie filter), the beam output is gradually reduced when moving away from the isocenter; therefore, the more distant a point is from the isocenter, the less the air kerma at that point. Second, during tube rotation, a given point on the patient's skin will be irradiated with the primary beam for a smaller arc portion [14, 15]. The same arguments are valid for patients of the same size when the table height is different. The farther the patient's skin is from the isocenter, the smaller will be the ESD and the DLP to which it is exposed.
Finally, it is well known that the actual dose profile extends beyond the geometric limits of the beam; thus, the steplike function assumed in the original model—being maximum and zero within and beyond the nominal geometric beam limits, respectively—can serve only as a first approximation. This assumption is in agreement with the CTDI definition, in which the air-kerma value recorded from a 10-cm-long active volume of the ionization chamber is divided by the nominal slice thickness. In this way, the gaussian-shaped profile is reduced to the step-shaped profile assumed in the original model. For a given point within the gantry, the air-kerma profile along the z-axis will also be modified by the presence of the patient's body, which produces variable attenuation and scatter of the incident x-rays. However, for a given point on the patient's skin entrance surface, the primary beam will always be the major contributor of the dose at that point.
The impact of the aforementioned parameters on the ESD profile were studied using EDR2 films and the pencil-type ion chamber in the presence and absence of two cylindric water phantoms with diameters of 20 and 25 cm to obtain the appropriate correction factors for different table height positions. The measurements without phantoms were performed with the pencil-type ionization chamber taped along the table's midline, whereas the measurements with phantoms were performed with the ionization chamber taped to the phantom's bottom.
The results of the aforementioned measurements were accounted for in the revised model used for calculating the ESD profile during CT-guided interventions. The dose profile of each individual slice was separately considered, and for each point in the z-axis the sum of the contributions of all slices was calculated. To model the ESD profiles, a formula was used to fit the ESD profiles recorded in the EDR2 films derived with and without phantoms, as described in Appendix 1.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
6% larger) with the respective value measured
using the digital dosimeter. This agreement can be explained considering that
in free-air measurement there is no significant contribution of scatter (in
contrast to the respective measurements made with phantoms) and thus the
air-kerma profile is well confined in the nominal geometric limits of the
x-ray beam.
|
|
In Figure 1, the results of the OD measurements in the calibration films are given versus the respective Dmax values in mGy as well as the formula that was used to fit the maximum OD values (ODmax) to ESD. The Dmax is the value of maximum ESD as derived from the PV-to-ESD conversion. Similarly, in Figure 2, the maximum PV (PVmax) values obtained from measurements in the digital images of the calibration films are given versus the respective Dmax values as well as the formula that was used to fit the PVmax values to the respective ESD values. It must be noted that both OD and PVs decreased toward the film edges as a result of the respective ESD decrease for points distant from the isocenter.
|
|
|
In Table 1, the results of the dosimetric calculations from the 12 interventional procedures studied are given including the procedure type, patient sex, anteroposterior and lateral diameters, and PSD values derived from film OD measurements, digital images, the original theoretic model, and the revised theoretic model. Typical examples of the deviations observed among measured and theoretic ESD profiles are shown in Figure 6A, 6B, 6C, 6D for four interventional procedures (two biopsies and two radiofrequency ablations), including the two procedures for which the largest PSDs were observed.
|
|
|
|
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
At this point, it is worth mentioning that large offsets increased considerably the presence of artifacts in patient images, whereas in water phantoms large offsets had an adverse impact on the CT attenuation value accuracy and noise. In extreme cases in which the table was positioned at its lowest position or quite close to it, deviations of the water CT attenuation value as large as 20 HU from 0 HU were observed. Thus, off-center positioning should be used with caution and only when absolutely necessary for the successful outcome of an intervention.
From the AKLP measurements made with and without phantoms to derive the OSCF(TH) function (shown in Fig. 5), it is evident that Dmax values continuously decrease when moving away from the isocenter. Although the Dmax ratios with respect to the maximum values obtained, respectively, with or without the two cylindric phantoms fall fairly well on the same curve, the absolute Dmax values derived with a phantom were about half the respective values derived in the absence of a phantom and were indeed slightly smaller for the 25-cm-diameter phantom.
Avilés Lucas et al. [14] studied the variation of AKLP values measured at the surface of three elliptic phantoms (16 x 30, 24 x 30, and 28 x 30 cm) with off-center positioning. They concluded that the observed variations were dependent on the offset of the entrance surface from the gantry isocenter independent of the patient size (within experimental error) and that the flattening filter was the major contributor to these variations. However, all the phantoms that were used in their study [14] had the same major axis. Therefore, whether the same conclusions would apply for an elliptic phantom having a smaller major axis is not certain.
The EDR2 films provide a relatively cheap way to determine the
PSD—that is, simply by measuring the OD at the points of maximum film
density. With the use of a scanner, a detailed 2D map of the skin dose
distribution can be obtained at the expense of the time required for film
scanning and data analysis. The estimate of skin dose with films seems to be
the method of choice when CT fluoroscopy is used because the contribution of
fluoroscopy to the patient skin dose cannot be accounted for with the
theoretic method given that, at present, only the last fluoroscopic image is
stored in the hard disk [4]. A
potential disadvantage of the EDR2 films is their dependence on the processing
conditions. With the OD-to-ESD conversion equation obtained from the
calibration films, we deduced that a 0.1 change in OD, which may result from a
small drift in processing conditions, will lead to a dose difference of
approximately 30 mGy (for ODs up to 2.5) and up to 80 mGy for OD values
close to film's saturation. For very small ODs (i.e., up to 0.7), the
respective error in the ESD can be in the order of 50%; however, for OD values
greater than 1, the respective error will be approximately 10%.
Both the ESD profile derived from digital images measured with the films and the ESD profile calculated with the revised theoretic method refer to the surface of the patient when the patient is lying on the table. The other patient entrance surfaces will be exposed to different ESD levels for the reasons analytically described earlier. Thus, when the patient table is lowered, the PSD on the skin of the patient when the patient is lying on the table will be reduced, but the respective ESD on the upper patient surface will be increased [15]. In this context, the original calculation method can be considered appropriate for providing an approximate average of the ESD on both patient surfaces. The accuracy of the original method could be improved if the CTDI value at the periphery of the body phantom (nCTDIp) used for approximating the ESD was corrected for the different patient body diameters and the consequent differences of nCTDIp with table height position.
The revised model could also be used to provide the ESD profile on the upper body surface by using the OSCF(TH), which corresponds to its distance from the isocenter. Because the upper patient surface does not have a constant offset from the isocenter as does the surface lying on the table, for accurate calculations an image-by-image measurement of the AP diameter would be normally required; however, as a first approximation, the AP diameter of the most exposed body region could be used.
In conclusion, all the methods presented in this study for the assessment of the radiation dose to the patient's skin during CT-guided interventions present certain limitations and each has advantages and disadvantages. The revised theoretic method, however, provides a useful dosimetric tool that is worth further development for producing software able to automatically use the DICOM data and calculate the skin dose all around the patient-exposed surfaces. Similar software for skin-dose mapping is already available in some sophisticated angiographic x-ray systems [2].
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
