Hellenic Journal οf Radiology

1.Introduction

CT is widely used in diagnosis of disease and has been identified as the major contributor to the collective ra­diation dose to population from medical exposures [1-7]. Depending on clinical needs and technical protocols, pa­tient radiation dose differs, even for the same anatomical region, clinical indication, technical protocol and some­times even for the same CT unit [8-12]. Furthermore, high rates of repeated CT scans are reported in the liter­ature (such as in trauma patients) with additional hospi­tal charges and additional radiation exposure [13]. The urge to monitor medical practices in CT scanner has ris­en significantly. The effective dose (E) for the most typ­ical CT examinations varies tremendously even for the same technical protocol or the same CT scanner (1 mSv- 25 mSv) [8-23]. Therefore, the necessity of tracking pa­tient dose data and CT scanning protocols is essential. It is worth noticing that CT radiation dose is linked to an in­creased risk of radiation-induced cancer [24-26].

To assess the radiation dose in CT, volumetric comput­ed tomography dose index (CTDIvol), dose length product (DLP) and E are the most widely used radiation dose indi­ces [8, 10, 27]. In order to optimise radiological practices more efficiently in terms of radiation dose, the term Diag­nostic Reference Level (DRL) is defined both in the Europe­an [28] and International [29] Basic Safety Standards (BSS). It is also clearly stated 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 [29].

Due to the upcoming implementation of the EU BSS [29], the need for an automated dose monitoring solu­tion rises. This can be an extremely time consuming and complex task. Nowadays, sophisticated software packag­es with friendly interface can assist to this task, result­ing in a much easier and quicker way to monitor all data included in the Digital Imaging and Communication in Medicine (DICOM) header of the CT scanner or data re­corded in the Picture Archiving and Communication Sys­tem (PACS) of the hospital [30-37]. These studies focus on radiation dose data analysis.

One of the commercially available dose tracking software was recently installed in our CT department. Our study fo­cused on evaluating the software features and capabilities and on assessing staff performance in every day routine. It also investigated whether the system can be used solely for patient radiation dose analysis, or could also assist in the general management of the CT department.

2. Material and Methods

2.1 CT scanner and dose tracking software system

A dose tracking software (DOSE, Qaelum, Belgium) was recently connected to a Philips Brilliance 64 CT scan­ner (Philips Systems, the Netherlands), installed in 2013 in Konstantopoulio General Hospital in Athens, Greece. This is a software tool that has the capability to connect with a variety of medical ionising and non-ionising ra­diation units that are available in a hospital and to com­bine all available data into a single configuration where every examination is analysed and archived. The con­figuration provides a direct overview and makes infor­mation available at any time for all users, depending on their level of authority. Although initially intended to be connected to a number of X-ray modalities such as an an­giography system, an older four (4) slice CT scanner and certain computed radiography systems, this could not be done in real practice. This was due to the fact that these devices are old and they could not export dose data in the DICOM formats that the software supports. It must be noted that our hospital does not have a PACS system and therefore the software was connected directly to the Philips 64-slcie CT scanner.

2.2 Technical factors and dosimetric quantities collected

For each examination, the tube voltage (kV), tube out­put (mAs), Field of View (FOV), pitch, collimation, num­ber of slices, series information, scanogram and oper­ator’s name are examples of clinical data that can be recorded. Regarding radiation dose metrics, CTDIvol and DLP are the dosimetric quantities transferred from the CT scanner to the software workstation. The soft­ware uses the patient CT exam data and calculates E and organ doses. Organ doses are estimated using conver­sion factors derived from Monte Carlo simulations for a standard-size model. More specifically, the conversion factors are derived by an in-house Monte Carlo simula­tion model that has been developed based on vGate sim­ulation tool. A reference adult phantom from the phan­tom library of MIRD is used for the simulations. The model calculates the organ doses for every body part examined according to typical z-ranges. Specifically, for every organ, dose profiles are generated. Simulat­ed scans are performed per 1 cm in the z-direction forvarious photon energies up to 140 keV obtained by com­bining different energy values according to the realis­tic spectrum for a given tube voltage. For organs away from the scan range, extrapolation of sparse data is per­formed. Organ doses are derived by adding the values of the dose profile over the typical z-range (determined by typical examination types). The validation of the in-house model was already performed against commer­cially available dosimetric tools before the study to en­sure accuracy of organ dose estimation [38].

2.3 Patient sample

The study included 6,010 CT examinations from 13th March 2016 to 14th March 2017. The study was per­formed according to the ethical standards as described by the Declaration of Helsinki. Initially an overall eval­uation of all CT examinations and utility of the software was performed. Then a more focused analysis was per­formed for the most frequent CT exams independently of clinical indication. Due to the higher radiosensitivity of organs in examinations of the trunk, the authors decid­ed to focus on these exams for further analysis. In order to estimate organ doses, the examinations were divid­ed in 3 broad anatomical part categories: Chest, Abdo­men and Chest/Abdomen/Pelvis (CAP) CT examinations.

3. Results

3.1 Overall evaluation

The percentage of men and women were 41% and 59%, respectively. The software offered easy and quick sta­tistical overview of all clinical and technical data of CT examinations. This overview could be provided for a specific date, for a time period within the day, or for a whole date range selected by the user. An example of this workload overview can be seen inFig. 1a and 1b.Fig. 1ashows the evaluation of each hour and day of the week in the CT room, in terms of number of performed exam­inations and average DLP. The number of examinations is shown by the size of the boxes; the mean accumulated DLP is colour coded. It is practically a moment evaluation chart of the actual usage of the CT scanner plotted to­gether with the corresponding DLP on a certain day and hour. Thus, the volume and dose trends are visualised to facilitate the analysis. The more red the box becomes, the more radiation dose (in terms of DLP) is given to the pa­tients the specified time of the day. As the box becomes bigger in size, the higher is the number of exams. This provided a valuable overview of the workload statistics which occasionally required discussion with the related staff and in some cases, corrective actions.Fig. 1bshows a weekly workload analysis graph, that practically mon­itors the activity in blue and green colours and also the inactivity of the CT scanner (orange and red colour). Reg­ular evaluation of this graph helped optimise the use ef­ficiency of the scanner. It also allowed making changes in the number of technologists in every day routine for better management of the department. This also identi­fied in a more structured way the time periods the scan­ner was not in use, in order to schedule routine quality control tests.

The software allowed evaluation of patient demo­graphics, in multiple ways depending on the operator or hospital management needs.Fig. 2shows part of pa­tient demographics analysis. It presents the distribution of patient sample according to age, separately for men and women. It shows that most of patients are in the range of 60-90 years, whereas the percentage of patients in the 40-60 years range is much smaller. This could be partially explained by the fact that European population is aging and a rise of chronic diseases is recorded [39].

As far as technical protocols and parameters used are concerned, a very large variability of CT protocols was noticed. Forty six (46) different scanning protocols were recorded for 16 different anatomical regions. These pro­tocols are related both to anatomical region and clinical indication. It must be noted that effective dose and or­gan doses estimation are related to anatomical region and not clinical indication. The analysis revealed mis­takes in the technique mainly related to the technical protocol (extra image series, longer scans than actually needed, wrong protocolused, or even choosing the cor­rect protocol but typing the wrong name). All these er­rors were communicated to the operators and helped in the optimisation of procedures and better organisation of the department.

2 Dose evaluation

Table 1shows median value and range for CTDIvol, DLP and E for routine chest and abdomen, as well as CAP examina­tions. Minimum dose values were found for Chest CT, fol­lowed by Abdomen CT and CAP exam. CTDIvol values did not have big differences between exam protocols. On the other hand, DLP almost doubled from Chest to Abdomen and more than tripled from Chest to CAP exam. For E, the value increased about 1.6 times from Chest to Abdomen and 2 times from Chest to CAP exam. The large range in DLP andE values especially in CAP reflected the protocol variabili­ty. This could be partially explained by different number of series (pre-contrast, post-contrast studies, multiphase studies, etc.) depending on clinical indication and physi­cian choice. Another reason was the different scan range applied by some operators. This was further investigated and results are presented in the last paragraph of this sec­tion. In general, all findings were discussed with radiolo­gists and technologists and corrective actions were taken.

Table 2defines typical local DRLs in terms of CTDIvol (mGy) and DLP (mGy.cm). They are defined based on the methodology described in the European CommissionGuidance document on DRLs for medical exposures [40].

Table 3presents the comparison of typical local DRLs to national and international data found in the recent litera­ture. It is noted that in almost all cases our local values are lower than both national and international data. The only local DRL higher than international data was for the CAP exam. As CTDIvol was comparable with the literature data, it is obvious that the scanning length was the reason for the higher CAP DLP.

Table 4presents the comparison of E with international literature [18, 19, 21-23]. It must be noted that our data are based on the fact that our routine examinations are per­formed with a pre-contrast and a post-contrast scan. Some of the papers listed in the table [23] specify values both per scan and per patient. Others do not clarify if values reported are for one or two scans (pre contrast or both pre and post contrast scans). It is evident that this plays a very important role for the final value of E. The large variability of doses re­flects the large differences in technical protocols that most often depend on clinical indication, or the differences in scan length (usually the choice of operator).

The software also offers individual organ doses (mGy) estimation in routine basis. Apart from individual patient organ doses, the software calculates the 25 percentile, me­dian values and 75 percentile for the organ doses. There is also the possibility to calculate these values for the whole sample of patients or apply a filter based on the specific body part or the same description of the examination. An example is given inFig. 3a and 3b. InFig. 3a, organ dos­es are shown for a male patient undergoing a multi phase CAP examination receiving an E of 96.9 mSv. The figure shows the organ doses in orange bars compared to the 25% of the whole patient sample distribution (yellow bars), the median (blue bars) and 75% (green bars). InFig. 3b, organ doses are shown in a similar way for a male patient un­dergoing a single phase CAP examination receiving a low E dose of 6.5 mSv. The bar graphs of the 2 figures clearly show that organ doses in a high dose exam can reach val­ues over 300 mGy in more than one organ (breast, lung, etc.) whereas there are CT examinations in which no or­gan receives a dose above 10 mGy (Fig. 3a). This tool was used in specific patient to patient dose evaluation and as an educational means for optimisation purposes. In cer­tain cases, it was helpful to estimate quickly organ doses in women of reproductive age in case of anxiety following a CT exam (this reassured the patient that radiation dose was minimal).

DLP is the dose value that represents more efficiently the differences between radiographers due to the differences in scan length (Fig. 4). Chest CT is a straightforward standard examination, which is performed in a similar way in terms of DLP by all radiographers. Abdomen and even more the CAP examinations vary in terms of DLP, due to the differ­ent scan region. This was also communicated and further discussed with radiographers and radiologists. As the soft­ware has a simulation tool that estimates organ and effec­tive dose for different technical CT factors, it was used for educational purposes to show in a more efficient way the effect of scan length (as well as all technical CT factors) in radiation dose.

4. Discussion

The dose management system proved to be an effective, powerful tool that facilitated the evaluation of gener­al practice and workflow of the CT department and re­vealed the habits of operators, so that corrective actions are made for the benefit of the patient. Before installing the software, analysis of data was a cumbersome pro­cess that required manual entry of values, with signifi­cant outflow of resources and the risk of typing errors. The fact that all these had to be done manually patient by patient was the main reason why it could not be done in a systematic way.

One could argue that the implementation of a dose management system is a cost for the hospital, especially with the current financial situation in Greece. This cost, which usually increases with the number of X-ray devic­es or examinations performed, includes not only the in­stallation but also the maintenance. However, the gain of the hospital is also important. In this study, we high­lighted some parts of the benefit, mostly in terms of im­provement of workload, personnel efficiency and patient dose evaluation. As far as radiation doses are concerned, assessment of examination doses in terms of CTDIvol and DLP for a certain number of patients is presented and dis­cussed extensively in the international literature for at least 20 years. The use of the dose tracking tool to eval­uate the whole sample and not only a limited number of patients helped to easily identify weaknesses and mis­takes in our every day routine. The immediate estima­tion of E and organ doses proved to be a valuable tool in this process. It allowed us to take a deep dive into the pa­tient data for optimisation of safety, quality, training and practice of the department.

The traditional estimation of average CTDIvol, DLP and E and comparison with national or international values seems no longer relevant, if one considers the numerous and very different technical protocols applied for differ­ent clinical indications for the same anatomical region. Furthermore, the number of scanning phases depends also on the clinical indication which is basically the main determining factor of radiation dose. Various clinical in­dication–specific CT technical protocols have been devel­oped to help tailor CT dose and image quality on the basis of specific clinical indications. It seems essential to de­fine diagnostic reference levels for each of these clinical indications for more accurate comparison of practice

Figure Figure Figure Figure Figure Figure Figure Figure