Justifying multidetector CT in abdominal sepsis: time for review?
The further development of multidetector row CT (MDCT) has led to changes in the application and examination technique, leading to a need to justify the level and frequency of radiation exposure associated with MDCT. A literature review of how the use of modern scanners has affected diagnosis was undertaken, followed by a year-long retrospective study of MDCT scans of patients presenting with symptoms of abdominal sepsis. The diagnostic accuracy of detecting causes of abdominal sepsis using this technology was sought. Scans were performed using a LightSpeed 16 system (GE Healthcare Medical Systems, Slough, UK and Milwaukee, WI). Clinical diagnoses were based upon surgical and histopathological findings, treatment outcome and follow-up scans. System dose parameters recorded were the dose–length product (DLP) and volume CT dose index. The literature on investigating suspected abdominal sepsis has not been updated significantly since the time of conventional CT. 94 patients were included in the study; causes of abdominal sepsis could be detected with a sensitivity of 0.95 and a specificity of 0.91. Repeat examination and cumulative exposure was a key finding. Patients with abscesses and acute pancreatitis had the highest number of scanner visits; patients with diverticular disease had the lowest number of visits, lowest cumulative DLP and shortest stay in hospital. Cumulative DLP was affected by scan length, number of scans and patient size. In conclusion, diagnostic accuracy data for MDCT scans using 16 slices confirm that CT remains a suitable modality for imaging abdominal sepsis but scope for dose constraint exists.
Intra-abdominal abscess or infection is highly variable in presentation. Depending upon a patient's symptoms and condition when they present for diagnosis, a range of microbiology tests and imaging studies, including ultrasound, may be employed. However, in most instances, CT of the abdomen and pelvis is the diagnostic choice for diagnosis of abdominal sepsis or abscess [1–3]. CT is further recommended in the National Guidelines Clearinghouse of the American College of Radiology  as the most appropriate examination in all but pregnant patients. CT is used to locate diagnostic features when a diagnosis cannot be established by clinical assessment and microbiology testing, which in practice is the majority of cases. In general, CT is used in preference to ultrasound to search the peritoneal cavity and pelvic region for sources of infection. Similarly, CT may be requested when there is doubt relating to diagnosis. Abdominal sepsis is a condition that may require drainage or monitoring scans, which means that there is scope for multiple visits and large cumulative patient doses.
In 2002, CT examinations accounted for 40% of the collective population dose from medical exposures in the UK but for only 3.3% of examinations [5–6]. In North America, up to 67% of diagnostic ionizing radiation is used for 16% of examinations . The development of multidetector row CT (MDCT) has extended the use of abdominal examinations and there is the perception that this has contributed to increasing exposure of populations.
Under IRMER (Ionising Radiation (Medical Exposure) Regulations) 2000, there is a requirement to perform regular clinical audits. To do meaningful dose audit/surveys, exposure must be related to clinical justification and our study was undertaken to review the current role of MDCT in abdominal sepsis with respect to the justification of the radiation exposures occurring in practice.
The purpose of our study was to review the current literature pertaining to the use of MDCT in the diagnosis of abdominal sepsis and to audit the use of this technology within our own practice, with a specific focus on patient exposure. This approach is relevant to current practice and the potential for dose constraint.
The focus of the search was diagnostic accuracy data relating to the diagnosis of abdominal sepsis with MDCT. Minimum requirements were sensitivity and specificity data. Beyond this, the aim was to establish a basis for a clinical audit and to review work relating to patient exposure as a consequence of abdominal MDCT. PubMed Medline (US National Library of Medicine) and Google Scholar searches were performed initially. Searches were later expanded based on references in suitable papers, related links in Medline and review papers etc. Abdominal sepsis covers a group of conditions; those sought in the literature search were the ones coded and classified by the ICD10 (International Statistical Classification of Diseases and Health Related Problems) system, including abdominal sepsis, abscess or infection. A summary of the search terms used is provided in Appendix A. Medline searches were limited to papers on human subjects. The study period was 2000 to 2007. Our aim was to include papers based on MDCT systems with a minimum of 16 slices, but there were too few available and so a minimum of 4 slices was accepted and recorded in the first review. The abstract of each paper identified by Medline at the first review was evaluated. Papers were included in the second review if the paper covered primarily the abdomen or abdominal structures and sepsis. Studies needed to comprise at least 10 subjects. If CT was used as the “gold standard” technique or diagnostic reference, papers were excluded. The diagnostic accuracy data recorded were sensitivity, specificity and accuracy. If the diagnostic accuracy data were missing and could not be calculated from data presented in the papers, the papers were excluded. Furthermore, studies that lacked data on the scanner manufacturer and model, thereby blinding readers to the number of scanner slices, and studies that used a mixture of scanners led to papers being excluded.
Abdominal sepsis was chosen because it is a mainstream application of CT and therefore relevant to common practice. In this retrospective study, all patients who received CT scans after presenting locally with symptoms of potential abdominal sepsis were considered for inclusion. Abdominal sepsis categories considered in the study were those coded and classified by the ICD10 system. We began by considering all patients who underwent CT of the peritoneal cavity (CT abdomen) or scans that extended further to include the pelvis (CT abdomen–pelvis), along with patients whose scans covered the chest, abdomen and pelvis (CT chest–abdomen–pelvis). Those who underwent CT-guided drainage were also included. Trained administrative staff helped to interrogate the electronic databases to recall the relevant referrals. The study period was 12 months from August 2006 to July 2007. Data on patients were collected on a customized supplementary data sheet. There was no advance collection of data or modification of examinations. The chairman of the local ethics committee confirmed that this was a service improvement survey that did not require local ethical review. Extensive use was made of the Picture Archiving and Communications System (PACS), Radiology Information System (RIS) and Computerised Radiology Information System (CRIS), together with patient notes.
Patients were included into the study when positive clinical indications or queries for abdominal sepsis had been recorded during patient assessment. Patient assessment was performed by nursing and clinical staff, and the data recorded were later transferred to electronic databases by administrative staff. Physical indicators included lower abdominal pain, tenderness, patient guarding and distended abdomen. Patient condition indicators included gastrointestinal symptoms (nausea, vomiting, diarrhoea, tenesmus), fever, infection, leukocytosis and previous surgery (within the past 4 weeks). Inflammation and inflammatory markers were also used as positive clinical indications. Trained administrative staff helped to interrogate the databases to recall these data, the CT protocols used and the patient identifiers. Classification of conditions into the ICD10 categories and exclusion of patients with staging or disease treatment response scans were later performed by a clinical scientist and academic radiographer. Queries and results were overseen by a consultant radiologist. Patients with other primary conditions or surgery without any initial indications of infection or sepsis were also excluded at this stage.
A 16-slice GE LightSpeed 16 MDCT scanner (GE Healthcare Medical Systems, Slough, UK and Milwaukee, WI) was used throughout the study, with automatic exposure control using tube current modulation and a noise index of 12.26. The CT protocols used a 0.7 s rotation time, collimation of 5 mm, slice width 5 mm, pitch 1.375 and standard tube potential of 120 kV. The standard peak of 380 mA was increased to 440 mA for large patients at the discretion of the radiographer. The field of view could be varied with patient size. Patients were scanned in the supine position and asked to suspend ventilation during the scan.
All scan impressions (reported by consultant radiologists), laboratory test results and final diagnoses (determined by lead physicians and clinicians) were recorded by a clinical scientist and academic radiographer making use of the electronic patient records and patient notes. Final diagnoses were recorded using patient notes, discharge letters (hardcopy and electronic), electronic reports and follow-up scan reports. CT scan findings were correlated with final clinical diagnoses either at the end of treatment or at the follow-up stage. Patients with normal findings were defined as those who remained free of complications for at least 4 weeks after the initial investigation of potential symptoms of sepsis. Other conditions unrelated to abdominal sepsis were also recorded.
Hospital stay related to the time in hospital between admission and departure. This could be determined from electronic records of admittance and departure. Where patients were admitted for other healthcare problems or for longer periods, hospital stay was taken as the time from onset or worsening of symptoms suggestive of sepsis that led to a CT referral (cross-referenced with patient notes and CT referrals). Patients who died within the period of the study were also recorded.
For each scanner visit, the dose survey parameters recorded were the dose–length product (DLP) and volume CT dose index (CTDI), which were satisfactory values for dose audits, as the scanner was suitably calibrated and maintained [8, 9]. For each patient, dose survey parameters were recorded from their initial CT. Where patients received additional partial scans/repeats during their initial scanner visit, the summed DLP and summed CTDI data were recorded for the attendance. However, when patients had further follow-up scans within the period of the study for ongoing investigation, the cumulative DLP was also recorded. Cumulative DLP was the linear sum of DLPs received by the patient at each scanner visit.
Using patient images from PACS, patient size was estimated using the cross-sectional area at the level of the middle of vertebra L3 (S Meeson, C M Alvey, S J Golding; unpublished), taking the CT scout image as a guide [10–12] and using eFilm software (Merge Healthcare; Milwaukee, WI) to estimate the cross-sectional area with an ellipse draw and analysis tool. A typical customized ellipse has been drawn on the CT image in Figure 1; the cross-sectional area calculated by the draw and analysis tool is indicated on the image.
The leading objective of the literature search was to carry out a meta-analysis of the published literature to determine the updated clinical role of MDCT. Receiver operating characteristic curve analysis was then used as a graphical representation of the trade-off between sensitivity and specificity. 10 studies were required for meaningful results.
In the dose survey, diagnostic accuracy was characterized by the prevalence, accuracy, sensitivity and specificity of the test. The positive predictive value (PPV) and the negative predictive value (NPV) data are also presented, based on the results of a dichotomous test.
To assess the uncertainty and reproducibility of the patient cross-sectional area measurement, a more challenging cross-sectional image was assessed eight times across the period of the study by two image readers. The uncertainty in the measurement was taken as the standard deviation of these area measurements.
The χ2 test, which is used to test whether the number of individuals in different categories fit a null hypothesis, was used to determine if a group of clinical indicators were more suggestive of abdominal sepsis conditions than non-sepsis conditions.
The main finding was that for the purposes of this study there were insufficient published data to perform a statistically significant meta-analysis. 167 potentially relevant articles were identified from literature searches and 164 of these were excluded from final analysis. Table 1 shows the reasons why these papers were excluded and their frequencies during the two stages of the literature review.
Only three papers fitted the inclusion parameters. Abdominal sepsis covers a group of diseases but these three studies quoted data for more specific conditions. Werner et al  reported on data for both abscess (sensitivity, specificity and accuracy of 1.00, 0.97 and 0.98, respectively) and acute diverticulitis (sensitivity, specificity and accuracy of 0.97, 0.98 and 0.98, respectively). In the study by Tack et al , the sensitivity data related to the use of CT as an indicator for diverticulitis (sensitivity and specificity of 0.44 and 0.94, respectively). Johansson et al  studied all types of appendicitis (sensitivity, specificity and accuracy of 0.91, 0.94 and 0.94, respectively).
We find that the literature on investigating suspected abdominal sepsis has not been significantly updated since the publications on conventional CT. Many of the papers identified by the searches were not relevant to our application of CT, and many papers were also ruled out owing to the lack of data on the imaging technique and technology used. Many of the excluded papers had more than one reason for being excluded. It was not possible to establish a basis for clinical audit from the limited published data.
94 patients were suitable for inclusion; 53 of these were male. The majority of patients were over 45 years old and the mean (± standard error of the mean) ages for men (58.9±3.0 years) and women (59.3±2.9 years) were equivalent.
In order to evaluate justification of the exposure fully, it is necessary to determine whether examinations had been successful in detecting or excluding sepsis accurately. Clinical indications for CT referral for all patients in the study are shown in Table 2. The combinations of indicators were different for each patient, usually with more than one indicator per patient. Known or suspected fluid collection, lower abdominal pain and inflammation were among the most frequently recorded clinical indicators. Table 3 contains a shortened list of abdominal sepsis conditions considered in the study, along with their ICD10 codes, and the frequency at which these were encountered during the period of study. All scan impressions were recorded, including those of other conditions unrelated to sepsis. Abscesses and ascites were the forms of abdominal sepsis identified most frequently. More than one clinical finding was often recorded per case. 20 patients died within the period of the study.
The contingency matrix in Table 4 shows the diagnostic accuracy results of the study. Although there were no false-negative or -positive scan results in this study, there were six inconclusive scan results. Half of the inconclusive scan results were positive and half were negative. For the detection of abdominal sepsis by MDCT, the prevalence was 0.64, sensitivity 0.95, specificity 0.91 and accuracy 0.94. The PPV was 0.95 and the NPV was 0.91.
The reference standard for the clinical diagnoses was based on surgical and histopathological findings, treatment outcome and, where available, follow-up scans. Combinations of these three factors applied in each of the 94 cases. 61% of the diagnoses included a surgical reference factor, 43% a treatment outcome factor and 63% of cases had a definitive follow-up scan. As shown in Table 4, there were 57 true positive cases of sepsis; 77% of the true positives were based on surgery, 11% on both treatment outcome and follow-up scans, 5% on follow-up scans and 4% on treatment outcome.
The clinical indications for CT referral have been subdivided into true positives and true negatives in Table 2. Although fever, leukocytosis, infection, previous surgery, distended abdomen and fluid/leak were recorded more frequently in sepsis cases, similar occurrences of lower abdominal pain, tenderness and gastrointestinal symptoms in the two groups indicate the non-specific nature of many of the clinical indicators. Combinations of indicators were different for each patient and no distinct group of clinical indications were suggestive of abdominal sepsis (p = 0.35, χ2 test).
Repeat exposures in patients take two forms. Patients may receive additional partial scans/repeats during their initial investigation. However, they may also have additional referrals when a follow-up CT is required for ongoing monitoring. Repeat exposures were a feature of this study and Table 5 shows the frequencies with which patients had between one and five scanner visits. 62% of patients had a single scanner visit.
For an increasing number of scanner visits, the mean hospital stay, mean DLP (number of scanner visits = 1) and mean cumulative DLP are included in Table 5. As the number of scanner visits per patient increased from one to three, there were increases in both the mean cumulative DLP and mean hospital residences.
The spectrum of exposures during initial scanner visits was measured for the main causes of abdominal sepsis. Table 3 contains a shortened list of abdominal sepsis conditions and these have been abbreviated to abscess, ascites, peritonitis, appendicitis, diverticulitis and pancreatitis in Tables 6 and 7. Each subgroup covers a number of conditions; for example, diverticulitis corresponds to diverticular disease with perforation and abscess of small/large/both/unspecified intestine (K57.0/K57.2/K57.4/K57.8). These six groups, along with “normals”, were selected for further analysis, with the mean DLP, mean CTDI and mean patient cross-sectional area being quoted in Table 6. Mean CTDIs and mean cross-sectional areas of patients with abscesses and ascites were equivalent, but more data are required (particularly for abscesses) to determine whether there is a difference in mean DLP. Although it is a small sample, peritonitis patients had the largest mean cross-sectional area and, concomitantly, the highest mean DLP. Furthermore, when looking at repeat scanner visits, it can be seen from Table 7 that patients with abscesses and acute pancreatitis had the highest number of scanner visits for further investigation of symptoms and treatment. Patients with diverticular disease had the lowest number of scanner visits, lowest cumulative DLP and shortest hospital stays. As expected, normals had the lowest figures for all parameters.
Where drainage of fluids was monitored repeatedly, the cumulative DLP rose rapidly, particularly for larger patients. This is illustrated by comparing two patients who each received three scans using the same CT protocol; they had cross-sectional areas of 670 cm2 and 1080 cm2 and cumulative DLPs of 2172 mGy cm and 3086 mGy cm, respectively. Cumulative DLP was affected by scan length, number of scans and patient size. Assessment of the reproducibility of the cross-sectional area measurements using the eFilm ellipse draw and analysis tool gave an uncertainty in the measurements of ±5 cm2.
There is a need to justify the level and frequency of radiation exposure associated with MDCT but there appear to be insufficient published data on modern MDCT scanners to determine the current performance of abdominal sepsis detection with CT. Although only PubMed and Google Scholar were utilized in the literature search, our results confirm that the literature on investigating suspected abdominal sepsis has not been significantly updated since the time of conventional CT. The lack of acceptable papers and the findings of articles reviewed support the need for further studies to confirm the diagnostic accuracy and use of modern MDCT scanners in what is an established and routinely used technique .
Our dose survey shows that MDCT is a clinically reliable procedure. The detection of abdominal sepsis by MDCT had a sensitivity of 0.95 with a specificity of 0.91. Clinical diagnoses were based on surgical and histopathological findings, treatment outcome and, where available, suitable follow-up scans. In a straightforward study, there would simply be one reference standard. However, as abdominal sepsis is not a single condition (rather it covers a range of conditions in the abdomen) and the non-specific nature of the clinical indicators meant that many non-sepsis “true negative” conditions were included in the study, a group of reference standards was required for this study. This complication is demonstrated by the reference standards for normals. Normal cases were classified as true negative. Surgery cannot be used to confirm scan findings of no abnormality seen in normals. The absence of surgical intervention in normals meant that there was not the same surgical and histopathalogical definitive evidence as in the majority of the sepsis cases, but the normals remained free of complications for at least 4 weeks after initial investigation and it is unlikely that unidentified sepsis or abscess in a patient would resolve on its own.
Our study results suggest that modern MDCT scanners are as good as, or better than, earlier 4-slice MDCT scanners for detecting abdominal sepsis, highlighting that the clinical role of CT has not changed with the development of new technology and remains a suitable modality for imaging causes of abdominal sepsis. The grouping of conditions under the heading “abdominal sepsis” is the reason for the slightly lower specificity in our study than in the studies reported upon. By subdividing the abdominal sepsis study, diagnostic accuracy data for ascites and abscess groups could be considered. All patients who presented with abscess or ascites had this identified correctly as a result of using CT (sensitivity of 1.00).
To avoid any ambiguity in what was covered by abdominal sepsis, the conditions included in the study were those classified by the ICD10 system. This meant that ascites was included in the study (ICD10 code R18), and represented fluid/collection that was not described as infective. At first sight, it is surprising that in a year-long study period, only nine patients had pancreatitis and two had appendicitis. However, this is a result of the conditions included and excluded from the abdominal sepsis list and the shortened group titles adopted in the results tables. Abdominal sepsis only includes acute appendicitis with either generalized peritonitis (K35.0) or peritoneal abscess (K35.1). None of the other more common forms of appendicitis are included. Similarly, the only forms of pancreatitis (K85) included are acute and necrotic. The six groups of conditions used in the results section were formed for each headline condition, such as appendicitis and pancreatitis, to increase the group samples and simplify the presentation of data. A sample of 94 patients was typical of the department throughput, based purely on cases of sepsis excluding scans for malignancy and other pathology.
Clinical indications of abdominal sepsis were non-specific, and there was no distinct group that was suggestive of sepsis. Ascites (R18) and abscesses (K65, T81.4) were the conditions identified most frequently. The non-specific nature of clinical indicators meant that other conditions unrelated to abdominal sepsis were frequently identified during the study.
Repeated exposures — whereby the patient returned to the scanner for further imaging — were a feature of our study and have not been extensively documented before. Although the mean DLP for a single abdominal scan was ∼750 mGy cm, there was a noticeable variation between patients, and cumulative DLP was larger the greater the number of follow-up examinations and the longer the scan length used.
Patients with diverticular disease had the lowest number of scanner visits, lowest cumulative DLP and shortest stays in hospital. Therefore, these patients had better in-patient experiences and on average a lower radiation exposure than patients with other conditions. Single and cumulative DLPs were also noticeably higher for larger patients. The range of cumulative DLP exposures recorded in this study was wide and reflected the scan length, number of scans and patient size.
This study suggests that there is a need for further dose constraint whilst maintaining image sensitivity. Abdominal sepsis can involve several peritoneal elements; therefore, patients presenting for initial diagnosis should receive a comprehensive examination of the peritoneal cavity to search for sites of infection. If a patient is not responding to, or getting worse after, catheterization, a repeat full diagnostic CT is required to check catheter location and search for further sites of infection. However, patients who have responded clinically to treatment should not need a follow-up CT. Emphasis should be placed on minimizing radiation exposure and using ultrasound when appropriate for follow-up scans. Ultrasound has a limited use at the initial assessment phase, but should be considered for drainage follow-up where appropriate. If doubts remain, low-dose and region-specific CT should be used for catheter removal and confirmation that the collection has resolved.
The scanner used in this work (GE LightSpeed 16) was a 16-slice scanner. Newer scanners are appearing with more detector rows and new associated applications. 64-slice MDCT is becoming standard, particularly with the benefits for angiography. Although the results of this study with abdominal CT apply to this new generation of scanners, studies on 64-slice scanners may be valuable to determine what exposure changes result from faster scanning and better image resolution where achievable.
In our study, doses were quoted using DLP. DLP was introduced to estimate fully the radiation delivered in the examination . The quantity itself accounts for the total number of photons received by the patient, and DLP can be converted to effective dose using the results of Monte Carlo simulations. In our study, DLP was used for relative comparisons between patients; absorbed dose remains the favoured measure to compare examinations and the exposure outputs of different CT scanners .
Summary and recommendations
Our study shows that MDCT using 16 slices remains a suitable modality for imaging in cases of suspected abdominal sepsis. The clinical role of CT appears not to have changed with the development of technology using MDCT with 16 or more slices. Exposures vary between patients, indicate what is used in clinical practice and act as the baseline for potential dose constraint in follow-up studies. Repeated exposures, where the patient returned to the scanner for further imaging, were a feature of our study and cumulative DLP increased noticeably with a higher number of follow-up examinations and a longer scan length.
Sepsis in the abdomen sometimes involves several peritoneal channels and initial diagnostic studies need to be comprehensive. Therefore we recommend:
- Patients presenting for initial diagnosis, or examination for failed response to treatment, should receive CT of the entire peritoneal cavity.
- Follow-up CT should preferably be avoided if patients have responded clinically to treatment.
- Low-dose and region-specific scans should be used for the specific indications of catheter removal and confirmation that the collection has resolved fully, if required clinically.
- Emphasis should be placed on minimizing radiation exposure and using ultrasound when appropriate for follow-up scans.
Summary of literature search terms
Table A1 details the search terms used during the literature searches. Combinations of the search terms and parameters were applied to the study period of 2000 to 2007. Search terms and parameters were combined in a logic sense. Different combinations of terms and parameters were used extensively. In general, search terms were combined using AND terms and parameters were combined in groups using OR terms. To illustrate this, the PubMed string below is an example of one of the more substantial searches used in this work. Searches were limited to papers on human subjects. Manuscripts required primarily a minimum of 16 slices for MDCT studies, but down to 4 were accepted. Studies needed to include at least 10 subjects. The search was restricted to full papers of new and original research and papers in English. Standard exclusions included contraindications, transplant, cancer and HIV.
PubMed string: (“abdominal sepsis” OR (abdominal abscess) OR “abdominal infection” OR peritonitis OR peritonism OR “peritoneal abscess” OR “peritoneal infection” OR “retroperitoneal abscess” OR “retroperitoneal infection” OR (subdiaphragmatic abscess) OR “subdiaphragmatic infection” OR “intra-abdominal abscess” OR “intra-abdominal infection” OR “mesenteric abscess” OR (mesenteric infection) OR “omentum abscess” OR (omentum infection) OR (retrocaecal abscess) OR (retrocaecal infection) OR (perforation AND abscess AND abdomen) OR “pancreatic abscess” OR “pancreatic infection” OR (kidney infection) OR “kidney abscess” OR (pelvic infection) OR “pelvic abscess” OR “subphrenic abscess” OR “subphrenic infection” OR “subhepatic abscess” OR “subhepatic infection” OR ascites OR leukocytosis OR leukocytes OR “white blood count” OR “full blood count” OR CRP OR neutrophils OR bacteroides) AND (computed tomography OR “CT”[Title/Abstract] OR ultrasonography OR ultrasound OR echo OR echography OR endoscopy OR laparoscopy OR “MRI” OR (magnetic resonance imaging) OR “PET” OR (positron emission tomography) OR “SPECT” OR “single-photon emission tomography”) AND (“diagnostic accuracy” OR sensitivity OR specificity OR “true positive” OR “true negative” OR “false positive” OR “false negative” OR “true-positive” OR “true-negative” OR “false-positive” OR “false-negative” OR “true positives” OR “true negatives” OR “false positives” OR “false negatives” OR likelihood OR predictive OR prognostic OR (receiver operating characteristics) OR “ROC” OR (reference value) OR (normal value) OR (confidence interval) OR “probability distribution”) NOT((contraindications) OR transplant OR gene OR (DNA) OR (AIDS) OR (HIV) OR seropositive OR (SARS) OR diabetes OR cancer) NOT(Case Reports[pt] OR Editorial[pt] OR Letter[pt] OR Comment[pt] OR Legal Cases[pt] OR review[pt]) AND english[la] AND “2000”:”2006/09/30”[dp].
Work was financially supported by the EC-EURATOM 6 Framework Programme (2002–2007) and forms part of the CT Safety & Efficacy (Safety and Efficacy of Computed Tomography (CT): a broad perspective) project, contract FP6/002388.
1 Saber AA, LaRaja RD, Bjerke HS, Talavera F, Grosso MA, Zamboni P et al, editors. Abdominal abscess. eMedicine 2007. Available from: www.emedicine.com/med/topic2702.htm [Accessed 14 October 2008].
2 Peralta R, Genuit T, Napolitano LM, Guzofski S. Jacocks A, Talavera F et al, editors. Peritonitis and abdominal sepsis. eMedicine 2007. Available from: www.emedicine.com/med/topic2737.htm [Accessed 14 October 2008].
3 Khoury G, Deeba SS. Naradzay JFX, Talavera F, Hardin E, Halamka J et al, editors. Pancreatitis. eMedicine 2007. Available from: www.emedicine.com/emerg/topic354.htm [Accessed 14 October 2008].
4 American College of Radiology. Acute abdominal pain and fever or suspected abdominal abscess. Available from: www.guideline.gov [Accessed 14 October 2008].
5 Verdun FR, Gutierrez D, Schnyder P, Aroua A, Bochud F, Gudinchet F. CT dose optimization when changing to CT multi-detector row technology. Curr Probl Diagn Radiol 2007;36:176–84.
6 Hart D, Wall BF. Radiation exposure of the UK population from medical and dental x-ray examinations. Chilton, UK: NRPB-W4, 2002, Available from: http://www.hpa.org.uk/radiation/ [Accessed 17 July 2008].
7 Brenner D, Hall E, Computed tomography — an increasing source of radiation exposure. N Engl J Med 2007;357:2277–84.
8 Shrimpton PC, Hillier MC, Lewis MA, Dunn M. Doses from computed tomography (CT) examinations in the UK – 2003 review. Chilton, UK: NRPB-W67, 2005. Available from: http://www.hpa.org.uk/radiation/ [Accessed 17 July 2008].
9 Keat N. CT scanner automatic exposure control systems. MHRA Evaluation Report 05016 (February 2005). London, UK: Medicines and Healthcare products Regulatory Agency, 2005. Available from: http://www.pasa.nhs.uk/pasaweb/nhsprocurement/cep/outputs/imaging.htm [Accessed 17 July 2008].
10 Althén JN. Automatic tube-current modulation in CT — a comparison between different solutions. Radiat Prot Dosimetry 2005;114:308–12.
11 Kalra MK, Maher MM, Toth TL, Schmidt B, Westerman BL, Morgan HT, et al. Techniques and applications of automatic tube current modulation for CT. Radiology 2004;233:649–57.
12 Toth T. Auto mA Theory. Milwaukee, WI: GE Healthcare Publication, 2006.
13 Werner A, Diehl SJ, Farag-Soliman M, Duber C. Multi-slice spiral CT in routine diagnosis of suspected acute left-sided colonic diverticulitis: a prospective study of 120 patients. Eur Radiol 2003;13:2596–603.
14 Tack D, Bohy P, Perlot I, De Maertelaer V, Alkeilani O, Sourtzis S, et al. Suspected acute colon diverticulitis: imaging with low-dose unenhanced multi-detector row CT. Radiology 2005;237:189–96.
15 Johansson EP, Rydh A, Riklund KA. Ultrasound, computed tomography, and laboratory findings in the diagnosis of appendicitis. Acta Radiol 2007;48:267–73.
16 Bearcroft PW, Miles KA. Leucocyte scintigraphy or computed tomography for the febrile post-operative patient? Eur J Radiol 1996;23:126–9.
17 Brenner DJ. Is it time to retire the CTDI for CT quality assurance and dose optimization? Med Phys 2005;32:3225–6.