Physica Medica 71 (2020) 176–177

Contents lists available at ScienceDirect

Physica Medica
journal homepage: www.elsevier.com/locate/ejmp

Is it possible to kill the radiation risk issue in computed tomography? T

Computed tomography (CT) has continued to attract public media must be performed, there is a need to analyze what can be done to help
attention from time to time [1]. Initially, during 2001–2007, it was the these patients as radiation risk creates concern. The need for further
potential cancer risks in children with continuing reports inter- development of appropriateness criteria has already been covered by
mittently; then unexpected overexposures in 2008–2009 that are un- paper by Rehani et al [7] and a recent Editorial [8] and is outside the
likely to occur again; reports of over-utilization in 2007–2009 and re- scope of the current paper. This paper deals with the need expressed on
ports of inappropriate use that keep appearing from time to time [1]. CT machines as the above papers conclude that there is an urgent need
These reports during the past nearly two decades have heightened in- for the industry to develop CT scanners and protocols with sub-mSv
terest in making CT safer for patients. There have been enormous ac- radiation dose, a goal that has been lingering [4,5,7].
tions by all stakeholders; manufacturers of CT systems, medical physi- Is it doable within foreseeable future? The answer appears to be
cists, researchers and users in hospitals and there have been national yes, very likely as indicated in a recent paper [9]. Essentially there are
and international campaigns and programs. While the realization six important factors of the CT system that can contribute to decrease in
started growing that the CT imaging has become much safer due to radiation dose to the patient.
significant reduction in radiation dose achieved [2,3], the recent stu- First, using more powerful X-ray tubes and thicker prefilters, in a
dies are turning the table once again on the side of risk to a large filter changer with a large variety of thicknesses will selectively remove
number of patients [4]. This time it is the risks from multiple CT exams low energy radiation from the beam that do not make it through the
some patients undergo with high cumulative radiation dose. patient and thus add unnecessary radiation dose. The X-rays will be-
Two large-scale studies just published have shown that nearly 1 out come more monochromatic. Currently, only one vendor provides
of 100 patients (0.64% to 3.4% with an overall average of 1.33%) who changeable filter thicknesses (0, 0.4 mm or 0.7 mm Sn). Second, use of
undergo a CT exam may receive high cumulative effective doses (CED) tunable tube voltage settings e.g. from 60 kV to 160 kV in steps of
of 100 mSv or more [4,5]. The minimum time to accrue 100 mSv was a 10 kV, can allow the X-ray spectrum to be individually tailored to the
single day at all four institutions, an unreported finding to date [4]. patient to meet the imaging task [10]. Third, automatic exposure con-
They argued that, although CED is not the perfect means to quantify trol would not only include tube current modulation but also tube
partial and intermittent body exposures by recurrent CTs, nor is it that voltage and filter thickness modulation. Apart from significant dose
100 mSv is the threshold for radiation effects, but values of CED more reduction [10] a positive outcome would be reduced beam hardening
than 100 mSv make it rather likely that various organs may receive and metal artifacts. Fourth, adaptive bow-tie filters, which are currently
absorbed organ doses of more than 100 mGy. They also provide organ under investigation and which aim at shaping the X-ray cone’s fluence
dose data for a sub-set of the cohort which showed several organs with field as a function of the patient anatomy have the potential to further
doses over 100 mGy. An Editorial that followed the above publications reduce the X-ray dose. For pediatric CT angiography, dose reduction up
delves on potential radiation effects at the radiation levels presented in to 70% for head, 77% for thorax, and 34% for abdomen/pelvis has been
above two papers and concurs with concerns [6]. The authors also es- reported [9]. Fifth, the expected introduction of photon counting de-
timated that nearly 0.9 million patients are likely getting into this group tector technology will boost CT image quality and dose efficiency.
every year globally [4,5]. This is totally unexpected finding in terms of Those directly converting detectors have no electronic noise, they offer
magnitude. The new finding comes in wake of tracking of previous favorable photon statistics due to counting instead of integrating, they
exposures that has become possible with modern technology in current provide improved iodine contrast, their spatial resolution is higher and,
decade. Just on click of a button, it is now possible to know all previous last but not least, they intrinsically acquire spectral information [11].
radiological exams and radiation doses, at least within the network of a For example, the combined effect of smaller detector pixels and of
health service provider. photon counting versus energy integrating can yield a 40% dose re-
What should be done? The traditional approach has been to blame duction [12]. All these features can be used to provide improved CT
overuse of CT. For the first time, it comes out that the above findings of images or to significantly lower the X-ray dose [9].
sizable number of patients with 100 mSv+ include data from a hospital What about the software? We all know that deep learning has
where there is good control mechanism in place and where imaging found its way into medical sciences and in particular in imaging. Sixth,
request goes through the clinical decision support (CDS) system [7]. obviously, it will also be used to improve CT image formation and most
Also, the hospital has adequate control on optimization of technique likely soon replace iterative reconstruction techniques [13]. The main
such that the typical CT doses are much below the national benchmark advantage of the neural network-based algorithms is that they do not
[4]. Additionally, the researchers manually studied imaging appro- rely on handcrafted features or prior knowledge, as do the iterative
priateness of a small subset in above high dose cohort of patients with reconstruction techniques, but that they rather learn very complex
recurrent CT exams and found high level of appropriateness as assessed features from the data themselves. Some vendors already provide deep
by clinicians [7]. This creates a unique situation not witnessed before. learning technology to remove noise from the reconstructed images
Obviously, when patients do need the imaging justifiably and the exam [14]. As of today, however, no clinical studies to prove or quantify dose

https://doi.org/10.1016/j.ejmp.2020.02.017
Received 28 January 2020; Received in revised form 16 February 2020; Accepted 21 February 2020
Available online 09 March 2020
1120-1797/ © 2020 Associazione Italiana di Fisica Medica. Published by Elsevier Ltd. All rights reserved.
 Physica Medica 71 (2020) 176–177

reduction levels are available yet. [4] Rehani MM, Yang K, Melick ER, Heil J, Šalát D, Sensakovic WF, et al. Patients
Conservatively assuming that each of these anticipated technolo- undergoing recurrent CT scans: assessing the magnitude. Eur Radiol 2019. https://
doi.org/10.1007/s00330-019-06523-y.
gical improvements, improved spectra, improved detectors and im- [5] Brambilla M, Vassileva J, Kuchcinska A, Rehani MM. Multinational data on cu-
proved reconstruction software, will allow to reduce X-ray dose to, say, mulative radiation exposure of patients from recurrent radiological procedures: call
one third, we will see a dose reduction of 70% in total. While sub-mSv for action. Eur Radiol 2019. https://doi.org/10.1007/s00330-019-06528-7.
[6] Ruehm W, Harisson RM. High CT doses return to the agenda. Radiat Environ
scanning today is possible for certain premium CT devices and selected Biophys 2020;59:3–7. https://doi.org/10.1007/s00411-019-00827-9.
protocols it is foreseeable that in the future not only premium systems [7] Rehani MM, Melick ER, Alvi RM, Khera RD, Batool-Anwar S, Neilan TG, et al.
will operate in this low dose domain and that they will do this on a Patients undergoing recurrent CT exams: assessment of patients with non-malignant
diseases, reasons for imaging and imaging appropriateness. Eur Radiol 2019.
routine basis for nearly all patients and protocols. It is unfortunate that https://doi.org/10.1007/s00330-019-06551-8.
the cumulative radiation doses to patients had not been assessed despite [8] Remedios D. Cumulative radiation dose from multiple CT examinations: stronger
the fact that the technology has been around to track patient doses for justification, fewer repeats, or dose reduction technology needed? Eur Radiol 2020.
https://doi.org/10.1007/s00330-019-06624-8.
the last over decade at some institutions. The latest results should create
[9] Lell MM, Kachelriess M. Recent and upcoming technological developments in
compelling need for the industry to push for sub-mSv CTs. computed tomography: high speed, low dose, deep learning, multienergy. Invest
It is hoped that CT imaging industry will create an agenda such that Radiol 2020;55(1):8–19. https://doi.org/10.1097/RLI.0000000000000601.
100 mSv+ doses will not be seen anymore. [10] Leyendecker P, Faucher V, Labani A, Noblet V, Lefebvre F, Magotteaux P, et al.
Prospective evaluation of ultra-low-dose contrast-enhanced 100-kV abdominal
While the intent remains that users are provided with CT scanners computed tomography with tin filter: effect on radiation dose reduction and image
that can reduce the worries on radiation risks for patients who are quality with a third-generation dual-source CT system. Eur Radiol
falling on the higher end of multiple CT exams, the need for rational 2019;29:2107–16.
[11] Taguchi K, Iwanczyk JS. Vision 20/20: single photon counting x-ray detectors in
practice, which even though is always needed, will become more ser- medical imaging. Med Phys 2013;40(10):100901.
ious when cumulative doses are within a few tens of mSv. One can [12] Klein L, Dorn S, Amato C, Heinze S, Uhrig M, Schlemmer H-P, Kachelriess M, Sawall
never have a diet that can substitute calories-consciousness or a security S. Effects of detector sampling on noise reduction in a clinical photon counting
whole-body CT. Invest Radiol 2020;55(2):111–9. https://doi.org/10.1097/RLI.
system that can make everyone feel 100% safe and thus continued at- 0000000000000616.
tention will be needed for reducing radiation risks to those who may [13] Wang G, Ye JC, Mueller K, Fessler JA. Image reconstruction is a new frontier of
still fall in the range of few tens of mSv. machine learning. IEEE Trans Med Imaging 2018;37(6):1289–96.
[14] Akagi M, Nakamura Y, Higaki T, Narita K, Honda Y, Zhou J, et al. Deep learning
reconstruction improves image quality of abdominal ultra-high-resolution CT. Eur
References Radiol 2019;29(11):6163–71. https://doi.org/10.1007/s00330-019-06170-3.

[1] Rehani MM. What makes and keeps radiation risks associated with CT a hot topic? Marc Kachelrießa, Madan M. Rehanib,
⁎

AJR Am J Roentgenol 2015;204(3):W234–5. https://doi.org/10.2214/AJR.14. a

German Cancer Research Center (DKFZ), Heidelberg, Germany
12860. b
[2] McCollough CH. Computed tomography technology-and dose-in the 21st century. Massachusetts General Hospital, Boston, MA 02114, USA
Health Phys 2019;116(2):157–62. https://doi.org/10.1097/HP. E-mail addresses: [email protected],
0000000000000997.
[email protected] (M.M. Rehani).
[3] Rehani MM, Szczykutowicz TP, Zaidi H. CT is still not a low-dose imaging modality.
Med Phys 2019. https://doi.org/10.1002/mp.14000.

⁎
Corresponding author at: Radiology Department, Massachusetts General Hospital, 175 Cambridge Str., Suite 244, Boston, MA 02114, USA.

177