| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ||||||||||||||||||||||||||||||||||||||||||||||||
Editorial |
The selection has been made by the current Honorary Editors and Deputy Editors but we would also like to acknowledge the hard work, carried out quietly, unassumingly and efficiently, by David MacVicar (Honorary Editor – Medical) and Victor Barley (Deputy Editor – Oncology) who both stood down at the end of September.
It would be presumptuous to suggest that we have necessarily chosen the "best" papers during the year. However, we believe the selection illustrates the multidisciplinary nature of the Journal, contains material of interest for all our readers and draws attention to many points of general importance.
If the "Review" is judged a success, our intention is that it will become a regular feature. Therefore this is something of a trial run and we shall welcome feed-back on the merits or otherwise of this initiative.
Radiology
The choice of a paper on a technique that to many radiologists has already been consigned to historical interest may seem surprising, but the paper from McLean et al [1] has perhaps a greater significance than may first be apparent. They highlight the need for critical appraisal of published papers when the results suggest that either new tests or a change in practice should be implemented locally.
McLean et al [1] reviewed four methods of reporting lung scintigraphy and then prospectively assessed their use on a cohort of 238 patients undergoing lung scintigraphy for suspected pulmonary embolic disease in their own hospital. The PIOPED criteria – locally modified [2], McMaster clinical criteria [3], PisaPED scintigraphic grading [4], and Miettinen logistic regression analysis [5] were all applied to the 238 patients and the results compared. The frequencies for differing levels of probability of pulmonary embolism varied widely between the different algorithms, with cross-referencing revealing correlation only between the modified PIOPED criteria and Miettinen logistic regression analysis. This led the authors to conclude that new diagnostic algorithms and tests are not automatically transferable to new environments.
They suggest that this is not an infrequent finding when new technologies, either diagnostic or therapeutic, are introduced locally, with the results appearing less impressive than the published data. There are clearly many reasons for this, not least the local "learning curve" set against the expertise of the published authors. Other reasons such as referral bias during the study, and underlying prevalence of disease – often high in the study population, will also have an effect, and it is these factors that may to some extent result in the local radiologists' inability to reproduce the published data.
This paper may "strengthen the hands" of those who already use one of the two methods of reporting lung scintigraphy that correlated. More generally it should serve as a lesson to all radiologists that it is essential to critically appraise published data and compare performance with already established techniques when a procedure is implemented locally.
Radiation protection in radiology
Not surprisingly, papers relating to doses to patients dominated these contributions to the Journal in 2004 and scarcely an X-ray technique or a part of the body escaped attention.
Barium examinations of the gastrointestinal system make up about 16% of the collective dose from radiological examinations in the UK [6] and Martin [7] has reported on a review undertaken to assess the impact of dose reduction facilities incorporated into fluoroscopic X-ray equipment in the last 5 years.
Substantial reductions in entrance doses were recorded. For barium enemas monitored in 11 hospitals, the median dose–area product (DAP) was 16.3 Gy cm2 (range 4.3–30.4 Gy cm2) compared with 26 Gy cm2 (range 9.3–48.9 Gy cm2) in 1994. However, mean screening times were the same (to within 1%) in the two studies and fewer images had been taken in 1994. Therefore some of the dose reduction could be linked to new X-ray equipment that has been installed in nine of the 11 hospitals in the intervening period with dose-saving facilities such as low dose pulsed fluoroscopy and digital imaging.
Additional copper filtration (0.2 mm) reduced DAP values substantially (about 44–50%) but it is important to remember that effective dose is reduced by much less (12–22%) because the beam has been substantially hardened before entering the body. Decubitus films contribute a substantial proportion of the dose for barium enema examinations and the ability to use the image intensifier in C-arm units gives a substantial dose advantage.
Another technique in which there have been significant technological changes is CT with the introduction of multidetector, multislice scanners. Between 1999 and 2002, nine of the 14 scanners assessed in the East Anglian region had been replaced by multislice technology. Carrying out dose audit to a strictly defined protocol allowed meaningful comparison to be made between the 2002 audit and previous audits in 1996 and 1999 [8].
Averaged across a broad spread of 10 routine examinations, regional mean effective doses had increased by 34% between 1999 and 2002. On average the multislice scanners gave 35% more effective dose than the single slice scanners and as would be expected from simple consideration of geometric efficiencies, the difference between single slice and multislice effective doses was greatest for examinations using narrow slice widths, where the effect of beam collimation in multislice CT is most severe.
These findings, combined with the known rapid increase in the frequency of CT examinations, highlights the importance of dose optimization, both in respect of technical improvements such as automatic tube current modulation and examination protocols.
Both of the preceding papers addressed the question of diagnostic reference levels (DRLs) and these were also the subject of an Editorial by Walker in the May issue [9].
DRLs are now a very clear, legally backed requirement placed on each radiology department in the UK to take ownership of patient dose and to set, monitor and use their own DRLs as an aid to patient dose optimization. Therefore the publication of guidance providing a pragmatic methodology for setting local DRLs for procedures carried out in radiology departments is to be welcomed [10]. The report also explains the difference between the setting and use of local DRLs, as opposed to national DRLs, outlines the actions necessary when local DRLs are consistently exceeded and provides a summary of data that can be used for setting national DRLs.
There is considerable benefit to be gained from a unified, national approach to both the setting up of DRLs and their use to optimize patient dose.
Teaching and image interpretation
Extended roles for both radiographers and nurses are an integral part of the UK Government's plans to create more flexibility in the National Health Service (NHS) and break down professional boundaries [11]. One of the areas in which non-medical healthcare professionals have had an extended role has been in the radiographic interpretation of trauma radiographs.
Hardy and Barrett [12] have looked critically at the differences in reporting arrangements in Accident and Emergency (A & E) Departments across a range of NHS hospitals in the UK. They found marked differences in the education provided for radiographers and nurses to support radiographic interpretation and in the range of examinations each staff group was permitted to interpret. A worrying finding was that, whereas the clinical gold standard for radiographer reporting was reporting by consultant radiologists, nurses have tended to compare their ability to interpret radiographic images with that of SHOs.
Overall, the authors concluded that a review of current service delivery strategies was required to ensure that reporting of radiology films in A & E departments is optimized.
Also in this section we note a thought-provoking paper by Manning et al [13] on the reasons for missing lung cancer from the posteroanterior (PA) chest radiograph. Basically the aim of the work was to determine whether observer error in reporting early lung cancer from PA chest radiographs is due to failure of detection or failure of interpretation.
Using a set of digital images containing a variety of pulmonary nodules, a group of radiologists as expert observers, eye tracking procedures to analyse dwell time and alternate free response operating characteristic (AFROC) techniques [14], the authors found that 65% of the false negative decisions were due to failures of interpretation, not failures of detection. As a consequence they question the widely-held assumption that technological changes to improve contrast, image resolution and detective quantum efficiency will naturally and always result in improved diagnostic outcome.
Such a finding is important because, if confirmed, it may add a new dimension to the way image interpretation is taught and may influence the way in which artificial intelligence should be used to assist the radiologist.
Oncology
For the radiation oncologist reading the BJR, this has been the year of intensity-modulated radiation therapy (IMRT). We have had a major, two-part, review article [15, 16]; a review of an alternative approach using helical tomotherapy, discussed in the next section [17]; an account of radiographers' reactions to the introduction of IMRT [18]; an account of the practicalities of setting up IMRT from scratch [19]. Several papers dealt with the comparisons between IMRT and more conventional techniques: in medulloblastoma [20]; in prostate cancer [21, 22]; and in oesophageal cancer [23]. A summary of the conclusions might be: IMRT is more demanding, in terms of both expertise and time, than more conventional techniques, but may have some minor advantages. These dosimetric advantages have not yet been translated into clinical results.
The other strand running through the year's publications has been the question of concomitant dose in patients treated with radiotherapy. Concomitant dose is that dose, additional to the prescribed dose, that is delivered as a result of the simulation and verification procedures performed during a course of radiotherapy [24–26]. The issue is important, not least because it arises when IR(ME)R is applied to clinical radiotherapy [27]. We might ask whether or not radiotherapy requires its own set of IR(ME)R regulations.
The risks posed by concomitant exposures may pale into insignificance, however, when compared with some of the problems raised by IMRT. IMRT uses multiple beams and there may be problems with exit dose [20], and also, compared with conventional treatments, with the increased volume of tissue treated to low dose. This has consequences both for bystander effects and radiation-induced genomic instability, including the induction of second cancers. In a short, but chilling, paragraph Guerrero and Nutting [16] mention that the whole body equivalent dose is nearly 2000 mSv for IMRT and less than 250 mSv for conventional treatment. This could translate into an eightfold increase in radiation-induced cancers. For advanced, otherwise incurable, tumours this might be a risk worth taking, but there is a serious concern for adjuvant radiotherapy where, as in breast cancer, for example, the majority of patients are already cured before they start their radiotherapy.
These are important issues and the BJR, with its multidisciplinary portfolio of authors and readers (physicists, epidemiologists, statisticians, radiation biologists, radiologists, oncologists) is in a unique position to address them. Our Britishness also helps. The NHS insulates us from many of the commercial pressures that, in other countries, may influence the evaluation and adoption of new technologies. We should be the masters of technology, not its servants. We need to know whether, in every sense, IMRT is worth the cost. Publications in the BJR have made major contributions to this debate, and will continue to do so.
Imaging in radiotherapy
Several papers in BJR in 2004 have highlighted the rapid developments in conformal radiotherapy and the integrated imaging systems which are essential to their success. In particular, Beavis [17] reviewed the emergence of helical tomotherapy [28], a radiotherapy technique which draws equally on both linear accelerator and CT technology to produce a design which moves significantly away from conventional linear accelerator concepts. A 6 MV linac waveguide and multileaf collimator produce a fan beam which is used to build up the required conformal dose distribution as the patient is translated through the gantry aperture. A bank of xenon ion chamber CT detectors mounted on the gantry intercepts the transmitted beam and allows volumetric CT imaging of the patient during treatment delivery. This makes possible modification of target volumes and the avoidance of critical structures on a fraction-by-fraction basis, so called adaptive radiotherapy. Moreover, margins to allow for patient movement should be reduced, leading to higher target doses. Helical tomotherapy is not uniquely required for this strategy and other systems have been proposed [29–31], but the integration of dose planning, image fusion facilities, treatment delivery and volumetric imaging within the same system may indicate the way ahead.
Increasing conformance of the dose distribution to a target volume, made possible by these recent developments in radiotherapy, raises the question of how the target volume should be defined in the first place, and what confidence should be ascribed to current imaging techniques in identifying it. Given the known difficulties in defining the planning target volume (PTV) [32, 33], we cannot assume that better in vitro dose distributions automatically guarantee better results in vivo.
It seems highly likely that the developing field of molecular imaging will be crucial to the understanding of the molecular basis of disease, including cancer. It was therefore timely to see the publication of a BJR Special Issue devoted to these developments [34], including MRI, MRS, PET, SPECT, ultrasound and optical methodologies. It may not be too long before the integration of molecular imaging modalities with radiotherapy planning and delivery is as commonplace as conventional CT is today, making radiotherapy a biologically, as well as geometrically, targeted technique.
Radiobiology/cancer biology
The role of the vasculature in tumour and normal tissue response to therapy has been a significant theme of articles published in the BJR recently.
It has long been suspected that damage to the vasculature is a major contributor to late injury to normal tissues following radiotherapy [35, 36], but direct evidence for this relationship has been lacking. A study published this year in the BJR offers compelling evidence for this in the central nervous system of rats: the use of a radioprotector that does not cross the blood–brain barrier, predictably inhibited vascular damage, but it also completely prevented the brain necrosis seen in non drug-treated animals [37]. While these data might suggest a means of protecting brain tissue against late radiation damage, it may also have the undesirable consequence of protecting against vascular-mediated damage to brain tumours. Recent evidence suggests, however, that tumours release anti-apoptotic factors that inhibit death of their supplying vasculature such that tumour endothelium may be protected anyway [38].
There is much recent evidence for the potential efficacy of agents directed against aspects of tumour vasculature, including direct, specific toxicity against the tumour endothelium and anti-angiogenesis [39–42]. Specific measurement in vivo of vascular changes associated with this therapy has been made possible by recent developments in imaging technology. The Special Issue, Angiogenesis Imaging, edited by Dr Padhani, contained articles addressing the key issues of how we might identify the activity of both anti-angiogenic and anti-vascular therapies in the clinic using a variety of modalities. Of particular note were the articles summarizing recommendations for appropriate methodology and endpoints for MRI studies [43, 44]. This information should help to maximize the quality of data obtainable from clinical trials using such agents.
Technical developments
An important technical contribution to the Journal was the Commentary by Visvikis et al on current trends and future developments in PET technology [45]. Bearing in mind the interests of many BJR readers, the article focused on applications in oncology. In this area the goals for developments in PET technology will continue to be improved (faster) patient throughput coupled with improved image quality and quantification. All of these advancements would be assisted by higher sensitivity devices and the authors reviewed some of the ways in which these might be achieved.
High quality PET imaging places stringent demands on the scintillation detectors – high density and atomic number for efficient capture of high energy photons; high yield of scintillation photons to improve positional resolution and scatter rejection; short scintillation decay times to improve dead time characteristics. Two new scintillators, lutetium oxyorthosilicate and gadolinium oxyorthosilicate, have been used in clinical PET scanners with significant reduction in the detection of scatter and random coincidences by comparison with the standard scintillation detectors in PET scanners (bismuth germanate). Further along the imaging chain there is active research to improve the photodetector system and overall photodetector design. For example a continuous light guide coupling the individual crystal elements to the photomultiplier tubes can minimize variations in light collection and improve energy resolution.
Scatter coincidences and random coincidences seriously degrade the information contained in the true coincidences detected and their effect on image quality must be minimized. For large fields of view, where scatter is high, this has generally been achieved by "collimating" the detectors using septa in the imaging field of view (so called two-dimensional PET). Uncollimated systems (three-dimensional (3D) PET) have difficulty handling the large amount of spurious data collected. However, "collimation" leads to a serious loss in sensitivity so work is continuing to try to refine the inherently more sensitive 3D design.
Finally the poor resolution of PET (4–5 mm) has stimulated work on image fusion to combine images from different techniques. This also provides an opportunity to combine anatomical and functional imaging. Since the use of non-linear registration techniques for sets of data acquired in separate systems is inherently more flexible than a combined system (see below) and considerably less expensive, further important developments can be expected in this area.
Notwithstanding, combined PET/CT systems are probably one of the fastest growing hardware developments in imaging technology. Apart from the obvious benefits of rapid and accurate image registration, the CT component provides a simple, quick facility for generating a body-tissue attenuation map that is essential for accurate quantitation of the PET image. Future developments of PET/CT systems will include the use of state-of-the-art multislice CT scanners and increasing use of combined systems in radiotherapy treatment planning. The potential for developing combined PET/MRI devices also must not be overlooked.
References
This article has been cited by other articles:
 |
|
 |
|
  BJR Review of the Year - 2005. Br. J. Radiol., March 1, 2006; 79(939): 183 - 187. [Full Text] [PDF]
|
|
|
|
|
|
E J Hall and C-S Wuu BJR Review of the Year 2004 Br. J. Radiol., July 1, 2005; 78(931): 672 - 673. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| BJR | DMFR | IMAGING | ALL BIR JOURNALS |
