Keywords
CT - QA/QC - screening - quality assurance - lung cancer screening - constancy test
Status quo of lung cancer screening
Status quo of lung cancer screening
With the Lung Cancer Screening Ordinance (LuKrFrühErkV) [1] and the decision by the Federal Joint Committee (G-BA) of June 18, 2025 [2], lung cancer screening is now approved as a new service covered by Germany’s statutory
health insurance funds. Starting in April 2026, lung cancer screening in Germany is
expected to begin on a large scale. For this reason, it is absolutely essential to
establish quality assurance measures.
Currently, the scope of physico-technical quality assurance for CT devices in radiology
is mostly limited to the acceptance and constancy tests formally required by DIN EN
IEC 61223–3-5 [3]. This scope is perfectly adequate for the primary purpose for which the devices
are used, i.e. clinical diagnostics. However, when conducting quantitative evaluations
of low-dose studies (especially to determine the volume doubling time), it is important
to re-evaluate the permissible tolerance ranges for the variability of the technical
parameters. This position paper addresses when and in what form additional quality
assurance should be conducted for low-dose CT protocols.
In Germany, a broad spectrum of devices and software versions are used in lung cancer
screening. The influence of arbitrary, non-standardized examination protocols on volume
determination for pulmonary nodules and the growth rate determined for lung cancer
tumors is the subject of long-standing and ongoing discussion [4]. The variability of a multitude of factors and the complex parameter range, including
the device used, radiation exposure, slice thickness, type of reconstruction algorithm
and kernel, object properties (e.g. size, shape, location), respiration process, and
software, can significantly influence the measured values [5]
[6]
[7]
[8].
As a result, manufacturers are required to provide their users with corresponding
recommendations in order to minimize variability as far as possible when selecting
the right parameters for CT acquisition protocols. These recommendations should follow
the appendix to Sec. 4 Para. 1 Sentence 1 of LuKrFrühErkV and the clarifications set
forth in this paper (see Appendix). There are currently no European guidelines for
the range of options for protocol parameters and the resulting differences in image
quality for the various device types. In other words, physico-technical quality assurance
in lung cancer screening has yet to be implemented [9].
The Physics and Technology Working Group of the German Radiological Society has therefore
addressed this pressing issue and formulated relevant recommendations for physico-technical
quality assurance (see the info box) in line with the position paper by Hahn et al.
on requirements for quality assurance of AI models for early detection of lung cancer
[10].
For information on the abbreviations used here, refer to List of abbreviations.
Key points of the position paper
-
For lung cancer screening, the established acceptance and constancy tests based on
DIN EN IEC 61223–3-5 should continue to be used. The tolerance limit of ±1 mGy should
not be used when testing low-dose protocols. Additionally, a dose measurement performed
annually is recommended for the low-dose protocols in use.
-
Daily air calibration is recommended to ensure consistent image quality in terms of
homogeneity and freedom from artifacts.
-
Device manufacturers should provide suitable screening protocols for their users.
-
Medical physics experts ensure that suitable protocols are available. Use of a dose
management system to monitor correct application is recommended.
-
It is strongly recommended to establish reference centers to provide centralized,
nationwide registration, documentation, evaluation, and quality assurance for screening
programs.
Literature review and current challenges for physico-technical quality assurance in
lung cancer screening
Literature review and current challenges for physico-technical quality assurance in
lung cancer screening
Currently, no empirical data is available regarding variability or stability/linearity
in the low-dose range for CT. Constancy tests are performed with a significantly higher
dose compared to a suitable screening protocol and are therefore not representative.
With regard to reproducibility of exposure for CT scans with a CTDIvol value in the 1 mGy range, the authors are unaware of any publication that considers
inter- or intra-device dependent dosimetric variability.
Years of experience have led to a high level of confidence in the stability and reproducibility
of dose and image quality, and the testing intervals have therefore recently been
extended [11]
[12].
Despite the increasing implementation of lung cancer screening programs, a significant
information gap remains regarding physico-technical quality assurance. Konrad et al.
address this issue and the authors highlight the need for a systematic survey and
standardization of CT acquisition protocols currently in use, in order to ensure the
quality and sustainability of screening over the long term [13].
In the UK, a CT protocol was established as part of a screening study, which was adapted
explicitly in line with the recommendations of the Quantitative Imaging Biomarkers
Alliance (QIBA). Lung-specific phantoms were used in the study to objectively verify
image quality and calculate doses [14]. The protocol established by Iball et al. is largely based on the QIBA Small Lung
Nodule (SLN) Profile. The publication emphasizes the need to implement quality assurance
in the detection and monitoring of pulmonary nodules when using quantitative imaging
[15].
In its technical recommendations for lung cancer screening, the European Society of
Thoracic Imaging (ESTI) sets forth detailed requirements for CT acquisition and image
reconstruction [16]. However, quality assurance is not addressed here. For applications in Germany,
the recommendations only provide European comparative values. However, these have
to be adapted considerably in practice to ensure regulatory compliance with specific
national laws and the LuKrFrühErkV.
The American Association of Physicians in Medicine (AAPM) has published device-specific
protocol recommendations for lung cancer screening [17]. There is no explicit reference to physico-technical quality assurance in terms
of European or German requirements.
The systematic review by Guedes Pinto et al. clearly demonstrates that the data on
the factors influencing the volumetry of pulmonary nodules is currently insufficient.
In particular, the clinical relevance of many study results cannot yet be fully assessed
[7].
While the influence of evaluation software and algorithms on volume determination
for pulmonary nodules undoubtedly plays an important role in the overall sensitivity
and specificity of lung cancer screening, this position paper focuses specifically
on the physico-technical aspects of quality assurance for CT imaging. The evaluation
software used may also impose further requirements on the parameters for the technological
implementation and must itself be subject to quality assurance. This cannot currently
be evaluated further without a structured lung cancer screening program. Particularly
for small tumors, there can be significant differences in volumetric assessment [18]. For a comprehensive analysis of the challenges and standards in the field of evaluation
software, readers should refer to the current overview by Hahn et al. [10], which deals with this aspect in detail.
In addition to the studies mentioned, the review by Vonder et al. provides a well-organized
overview of the international literature on CT acquisition protocols in lung cancer
screening, and it underscores the need for further standardization [19].
I. Are previous acceptance and constancy tests applicable for the low-dose range (especially
CT series with CTDIvol ≤ 1.3 mGy)?
I. Are previous acceptance and constancy tests applicable for the low-dose range (especially
CT series with CTDIvol ≤ 1.3 mGy)?
Quality assurance for technology according to previous DIN standards:
The established acceptance and constancy tests based on the DIN EN IEC 61223–3-5 standard
[3] are currently considered sufficient for the early detection of lung cancer [20]
[21]. However, these should be re-evaluated continually, building on the findings obtained
scientifically by monitoring screenings (see Section VI).
According to a report from the Federal Office for Radiation Protection (BfS) (BfS-34/21,
2021, point 3.9.1.1 [22]), it is particularly important to provide technical quality assurance for the equipment,
devices, and appliances. This includes verifying the physico-technical parameters
through acceptance, constancy, and expert tests with regard to the quality objective
in Sec. 14 Para. 1 Sentence 5a of StrlSchG.
The constancy test for the dose must be conducted
annually as well as after a major maintenance or repair activity
(e.g. replacing a tube or detector, etc.).
After consultation with the manufacturers of CT equipment who are members of ZVEI,
or the Central Association of the Electrical and Electronics Industry, it is assumed
that the CTDIvol dose value of 1.3 mGy is maintained, with regard to the expected accuracy based on
the standard above [3]. This is confirmed accordingly by the manufacturers in the accompanying documents.
According to point 5.4.6.1 of the standard [3], a deviation of the CTDIvol of ± 20% is tolerated. The second tolerance criterion in the standard of ±1 mGy for the CTDIvol value is not, however, applicable, as such a high deviation is unacceptable at the current low dose level.
II. What should an additional constancy test for a lung cancer screening CT protocol
include?
II. What should an additional constancy test for a lung cancer screening CT protocol
include?
Expanding the annual constancy test (after adapting the protocol)
The test parameters used in the current DIN EN IEC 61223–3-5 standard only provide
for dose measurement, under typical operating conditions, of an adultʼs body and head,
or a child’s body and head. The measured dose values are typically 20 mGy and higher.
Therefore, the results are not representative for the low-dose range. Nevertheless,
manufacturers guarantee that the dose displayed is within a tolerance of approximately
±20% (with higher pre-filtration, e.g. with tin or silver, discrepancies of up to
± 30% are possible).
The recommended protocols for lung cancer screening are no exception. Taking into
account the reasons cited above, the Physics and Technology Working Group considers
it useful to monitor the CTDIvol value of the screening protocols used in the annual constancy test.
Daily quality assurance measures:
Although this is a high-contrast issue, which is often considered less critical in
terms of image quality, the extent to which the low dose and the resulting higher
noise can lead to uncertainties in volume determination remains unclear. Specifically,
modern CT devices are highly sensitive, which can result in ring artifacts in low-dose
scans. This can occur when detector units are not properly calibrated to each other
or when the calibration is outdated [23]. To reduce the number of artifacts, daily air calibration is recommended, in order
to ensure consistent image quality in terms of uniformity and freedom from errors.
Quality assurance using anthropomorphic phantoms
Current studies dealing with quality assurance in lung cancer screening point primarily
to problems related to the scan protocol or the procedure itself. The working group
of Rydzak et al. identifies avoidable technological problems. These include, among
other things, a faulty FOV, patient movement, or stripe, ring, and noise artifacts
in the image. Rydzak et al. therefore recommend complying with the relevant national
regulations and the manufacturer’s guidelines, maintenance, and routine tests to ensure
sufficient image quality for lung cancer screening. The use of anthropomorphic phantoms
is not considered necessary [9]. Gierada et al. also point to reductions in image quality in CT lung cancer screening
caused by faulty scan parameters. No recommendation is made for quality assurance
using anthropomorphic phantoms [24]. Already in 2015, sufficient sensitivity for the automatic detection of nodules
>12 mm was demonstrated for low-dose CT [25]. For artificially created lesions with a diameter of 5 mm, the error in volume determination
for scans with 1.54 mGy was less than 30%, even for soft tissue kernels [8].
If this preliminary assessment changes during the screening evaluation, either existing
commercial solutions can be used or newly developed anthropomorphic phantoms can be
employed. In the event of such a reassessment, recommendations will be made regarding
the phantom (design, material, etc.). A good example of this is the work of Hernandez-Giron
et al., which demonstrated that 3D-printed lung phantoms can generate clinically relevant
and reproducible readings in CT [26].
Existing quality assurance measures for technology must be reviewed for their suitability
for low-dose screening programs. This includes, in particular, the regular determination
of dosage-related uncertainties such as CTDIvol deviations (especially after hardware update/maintenance), as well as suitability
testing for the filter kernels and reconstruction algorithms used in the screening
protocols (especially after software updates), since these affect resolution and volumetry.
III. What tasks should device manufacturers take on?
III. What tasks should device manufacturers take on?
Recommendations for scan parameters
Manufacturers should provide the following information for their users:
-
An overview of suitable protocols for lung cancer screening.
-
Instructions explaining how to adapt existing protocols in line with the specifications
in the appendix, taking into account the specifications mentioned on page 8.
-
Instructions for performing annual dose measurements in the low-dose range.
-
Specify any limitations for the respective device types/software versions.
-
If applicable, specify limitations for the reconstruction kernels to be used with
regard to spatial resolution.
IV. What tasks should medical physics take on?
IV. What tasks should medical physics take on?
Creating, releasing, and monitoring the use of specific protocols
The Physics and Technology Working Group notes that, according to the provisions of
the Radiation Protection Ordinance (Sec. 131 Sentence 2 Point 3), an MPE must be consulted
for CT examinations. In addition, according to Sec. 132 Point 1, MPEs must be involved
in physico-technical quality assurance activities.
EURATOM Directive 2013/59 (e.g. Article 83) already clearly tasks the MPE, in particular,
with ensuring the proper management of radiation exposure [27].
The Physics and Technology Working Group thus recommends creating dedicated, clearly
named examination protocols to be carried out with the involvement of the MPE designated
for the CT scan. The initial approval for clinical use is granted by the experts in
radiology and medical physics responsible for the device. Any changes to the initial
protocols must be made in consultation with or approved by the responsible persons.
Additionally, it must be ensured that only the approved protocols are used on the
person examined as part of the screening program. Monitoring of the protocols used
can be accomplished, for example, with the aid of a dose management system.
As part of a “local” quality assurance concept, the selected scan parameters should
be taken into account in conjunction with the software used for computer-assisted
detection. The stability of the relevant physico-technical parameters should be checked
regularly (a monthly interval is recommended).
The Physics and Technology Working Group will develop and provide a best-practice
guide for implementing this requirement and any additional tasks that may arise.
V. Are reference centers necessary for a successful screening program?
V. Are reference centers necessary for a successful screening program?
Re-evaluation of quality assurance
Although reference centers are not strictly necessary for the launch of the screening
program, it is essential to establish them in the long term to ensure the process
quality and long-term success of the screening program. Without a central and structured
evaluation of the data collected, it will hardly be possible to scientifically evaluate
the program. In order to generate valid information regarding the sensitivity and
specificity of the screening program (detectability, false positive or false negative
values, mortality, and morbidity), we currently see no alternative to establishing
reference centers similar to those already being used successfully in mammography
screening. The structured recording and analysis of patient data sets generated in
the framework of lung cancer screening appears particularly relevant with regard to
CT image quality and protocol parameters. This was already set out in the previous
BfS report (BfS-34/21, 2021, item 3.9.1.3 and A.1.2.7 [22]).
These reference centers are intended, among other things, to issue recommendations
on the following:
-
suitable grayscale value (HU) window widths for diagnosis,
-
possible field-of-view limitations,
-
CTDIvol lower limits depending on body size
-
suitable reconstruction algorithms
and regular updates to these recommendations. This will enable practitioners to formulate
a reliable, screening-based indication regarding the detectability of nodules depending
on their size (volume in mm³) and density (attenuation in HU).
The permissible tolerances or fluctuations in dose and image quality needed to support
valid volume determination are still in the process of being defined.
Outlook
In the framework of lung cancer early detection, little or no research has been conducted
so far to determine the quality assurance measures that are meaningful in relation
to technology.
Recently published studies showed that the deviations in volume determination for
artificially created lesions were less than 30%. Therefore, the use of a special phantom
for quality assurance does not appear necessary at present [8].
In the supplementary material to the position paper published jointly by the German
Radiological Society, the German Society for Pneumology and Respiratory Medicine,
and the German Society for Thoracic Surgery, a diameter of 4 mm (corresponding to
34 mm³ or 4–5 pixels) is considered relevant for determining volume doubling times
(see the section “Modified Lung-RADS classification with volume doubling time”) [28].
Due to the current lack of data on physico-technical quality assurance in the low-dose
range, it is considered necessary to scientifically monitor screenings. This includes,
for example, the influence of reconstruction methods and filter kernels, image noise,
and evaluation software used on volume determination as an essential parameter for
Lung-RADS classification. This is especially true for small lesions. Only when relevant
data are available can the quality assurance processes be evaluated and the specifications
adjusted, as necessary.
It should be possible in the near future to develop a quality assurance package specifically
adapted to the requirements of early lung cancer detection; in this package, physical
parameters such as spatial resolution, noise, and contrast of the scan protocol could
be regularly checked, for example, by using a phantom (possibly also in the form of
an end-to-end test).
Future studies need to clarify how to deal with uncertainties arising from the variability
of CT devices, protocols, and evaluation software in follow-up examinations. Standardized,
manufacturer-independent protocols are not feasible due to very different approaches
regarding dose and tube voltage automation, reconstruction methods used, etc. This
position paper also refers to the obligation of manufacturers to provide – for all
usable devices – suitable protocols that comply with the requirements of the ordinance.
Appendix
How should the restrictions on the device parameters to be selected from the appendix
to Sec. 4 Para. 1 Sentence 1 of LuKrFrühErkV be interpreted?
The appendix to LuKrFrühErkV lists device parameters that must be adhered to when
used in lung cancer screening.
The Physics and Technology Working Group has classified some of the parameters listed
here, or their value ranges, as open to interpretation or even misleading, and has
therefore – after consulting with the manufacturers – added some important specifications:
Pitch 0.9 to 1.2
Using this predefined range of values, high-pitch spirals are no longer possible with
dual-source devices, even though they are more dose-efficient. This goes against the
idea of optimization. It makes sense to give recommendations for selecting the pitch
for individual devices; however, a general restriction to the values mentioned here
does not appear to make sense.
Automatic voltage control
The general requirement to use what is known as an automatic tube voltage control
system excludes manufacturers, and the working group does not consider such a requirement
meaningful.
Adapting the protocols to the body mass index (BMI) of the person being examined is
sufficient, as this ensures an almost constant image quality, regardless of the individual
patient, and reduces radiation exposure accordingly. kV-modulation or tube voltage
automation can potentially influence the image texture and thus affect the results
of the volumetric analysis. Therefore, manufacturers should provide suitable examination
protocols for lung cancer screening for the respective device type and the use of
existing automated systems should be approved, also taking into account higher pre-filters
(e.g. tin or silver).
Automatic exposure control/automatic tube current modulation
Adapting the protocols to the BMI of the person being examined is sufficient. The
working group recommends adjusting the tube voltage (e.g. by means of automatic tube
voltage control) and/or the tube current (e.g. by means of tube current modulation)
to the body size.
Lateral resolution (x,y)
The remarks mention that an isotropic spatial resolution of 0.8 to 1.0 millimeters is required for contrasts of 150 Hounsfield
units and higher. This could be changed to “nearly isotropic spatial resolution”, as a change in the field of view also entails a change in the resolution in (x,y).
Example: For an obese patient with a resulting FOV of 52 cm, the resolution in the
(x,y) plane would be 520 mm/512 = 1.02 mm.
Dose value of 1.3 mGy
According to Sec. 1 of LuKrFrühErkV (Sec.1 (1), Paras. 1 and 2), “low-dose computed
tomography [...] of the thorax is an application in which a volumetric computed tomography
dose index (CTDIvol) of 1.3 milligray is not exceeded or a higher volumetric computed tomography dose
index than 1.3 milligray is necessary in individual cases due to the body size of
the person being examined.” Following the announcement in the Federal Gazette of the
scientific assessment by the Federal Office for Radiation Protection regarding early
detection of lung cancer dated December 6, 2021 (BfS-34/21, 2021, point 3.6.1 Table
19), it should be noted that the 1.3 mGy value refers to a patient group with a BMI
≤ 26 kg/m2 and that state-of-the-art equipment allows for significantly lower values
[22].
It would therefore make sense to have recommended values for CTDIvol for several different body types, adjusted for BMI (e.g. based on the protocol recommendations
for low-dose HRCT published by the Working Group on Diagnostic Radiology of Occupational
and Environmental Diseases [21]).
In the HANSE study, CTDIvol values of 0.8 to 1.6 mGy were observed, although only one device from one manufacturer
was used [20].
List of abbreviations
BMI:
body mass index
BfS:
Federal Office for Radiation Protection [Bundesamt für Strahlenschutz]
CTDIvol
:
volumetric computed tomography dose index
FOV:
Field-of-View
LuKrFrühErkV:
Lung Cancer Screening Ordinance [Lungenkrebsfrüherkennungsverordnung]
mGy:
milligray
MPE:
Medical physics expert
StrlSchG:
German Radiation Protection Act [Strahlenschutzgesetz]
StrlSchV:
German Medical Radiation Protection Ordinance [Strahlenschutzverordnung]
