Radiol Oncol 2023; 57(1): 86-94. doi: 10.2478/raon-2022-0048
86
research article
The spine and carina as a surrogate for 
target registration in cone-beam CT imaging 
verification in locally advanced lung cancer 
radiotherapy
Jasna But-Hadzic1,2, Karmen Strljic1, Valerija Zager Marcius1,3
1 Department of Radiotherapy, Institute of Oncology Ljubljana, Ljubljana, Slovenia
2 Faculty of Medicine, University of Ljubljana, Ljubljana, Slovenia
3 Faculty of Health Sciences, University of Ljubljana, Department of Medical Imaging and Radiotherapy, Ljubljana, Slovenia
Radiol Oncol 2023; 57(1): 86-94.
Received 8 May 2022
Accepted 19 October 2022
Correspondence to: Valerija Žager Marciuš, Ph.D., Institute of Oncology Ljubljana, Zaloška 2, SI-100 Ljubljana, Slovenia 
E-mail: valerija.zager@zf.uni-lj.si; vzager@onko-i.si
Disclosure: No potential conflicts of interest were disclosed. 
Th is is an open access article distributed under the terms of the CC-BY license (https://creativecommons.org/licenses/by/4.0/).
Background. The aim of the study was to evaluate the accuracy of volumetric lung image guidance using the spine 
or carina as a surrogate to target for image registration, as the best approach is not established.
Patients and methods. Cone beam computed tomography images from the 1st, 10th, 15th, and 20th fraction in 40 
lung cancer patients treated with radical radiotherapy were retrospectively registered to planning CT, using three 
approaches. The spine and carina alignment set-up deviations from a reference (tumour/lymph nodes) registration 
in the lateral (LAT), longitudinal (LONG) and vertical (VRT) directions were analysed and compared. Tumour location 
and nodal stage influence on registration accuracy were explored. 
Results. The spine and carina mean set-up deviation from reference were largest in the LONG, with the best match 
in the VRT and LAT, respectively. Both strategies were more accurate in central tumours, with the carina being more 
precise in 50% LAT and 66% LONG mean deviations. For all measurements in all patients a carina vs. spine registration 
comparison showed improved carina accuracy in LAT and LONG. In comparative subgroup analysis the carina was 
superior compared to spine in LAT and LONG in centrally located tumours, N2 and N3. Both strategies were compa-
rable for peripheral tumours and N0.
Conclusions. Carina registration shows greater accuracy compared to spine in the LAT and LONG directions and is 
superior in central tumours, N2 and N3. The spine and carina surrogates are equally accurate for peripheral tumours 
and N0. We propose the carina as a surrogate to target for CBCT image registration in locally advanced lung cancer.
Key words: locally advanced lung cancer; volumetric image verification; tumour registration; carina registration; 
spine registration; adaptive radiotherapy.
Introduction
Recent advances in systemic and radiotherapy 
treatment have resulted in improved survival for 
patients with inoperable locally advanced lung 
cancer.1 But still, nearly half of the patients will ex-
perience locoregional relapse.2 The accuracy of dif-
ferent steps in radiotherapy treatment preparation 
and execution have a strong impact on local con-
trol.3 With the introduction of computed tomogra-
phy (CT), positron emission computed tomogra-
phy (PET CT) and four-dimensional (4D) CT simu-
lation, the target delineation accuracy increased.4 
Modern radiation techniques have enabled a more 
conformal dose delivery, the dose to normal tis-
sue was reduced and the radical treatment of 
Radiol Oncol 2023; 57(1): 86-94.
But-Hadzic J et al. / Accuracy of volumetric lung image guidance 87
more advanced N3 disease and/or dose escalation 
has become possible.5 Because of the steeper dose 
gradient and smaller safety margins, the accuracy 
of treatment delivery became increasingly impor-
tant. For daily treatment position verification, cone 
beam CT (CBCT) largely replaced electronic portal 
imaging because of better soft-tissue visibility. The 
alignment of the treatment image with the plan-
ning CT (pCT) is usually performed by radiation 
therapists (RTTs), who are not trained in target de-
termination. The fast interpretation of CBCT is also 
challenging due to the lower quality of CBCT im-
ages (no i.v. contrast) and changes with or adjacent 
to the tumour during treatment.6 Manual tumour 
matching is subjected to strong inter-observer var-
iability even among radiation oncologists.7 The op-
timal surrogate structure for image matching has 
not yet been established. Spine alignment is feasi-
ble and reproducible, but shows poor correlation 
with tumour position.8 Recently, the carina surro-
gate alignment was explored and showed superior 
reproducibility compared to spine alignment.7
We performed a retrospective study to deter-
mine the accuracy of the carina vs. spine regis-
tration compared to target (primary tumour and 
lymph nodes) registration as a reference. To avoid 
interobserver variability, reference registration 
was performed by individual thoracic radiation 
oncologist. To consider tumour and normal struc-
ture variations during the course of treatment and 
their possible impact on image registration, the 1st, 
10th, 15th, and 20th fraction CBCT were included in 
the registration analysis.
Materials and methods
Patient selection
We retrospectively included 40 consecutive lung 
cancer patients treated with on-line cone-beam 
computer tomography (CBCT) image guided radi-
cal radiotherapy from September 2018 to February 
2019. All the patients had a visible tumour and/or 
lymph nodes on CT. They were treated with con-
ventional, or hypofractionated volumetric modu-
lated arc therapy on the Elekta Synergy linear 
accelerator (Elekta Synergy, Stockholm, Sweden). 
Clinical and treatment details were retrieved from 
medical records.
Simulation and planning
The planning CT scan was performed on the Big 
Bore CT simulator (Philips N.V., Eindhoven, NL), 
the Somatom Definition AS CT simulator (Siemens, 
Erlangen, D) and, in 6 patients, on Siemens 
Biograph mCT 40 (Siemens, Erlangen, D). The pa-
tients were immobilised on a Posirest-2 (Civco, 
Coralville, USA) with the arms abducted above 
the head (36 patients) or with a long thermoplastic 
mask (4 patients). The gross tumour volume (GTV) 
was delineated as the visible tumour and patho-
logic lymph nodes on free breathing pCT (with i.v. 
contrast). Additionally, 4D pCT was used for inter-
nal target volume (ITV) delineation in 10 tumours. 
Planning was performed on the Monaco treatment 
planning system using the Monte Carlo calculation 
algorithm. Conventional fractionation (1.8–2 Gy 
daily dose) was used in 37 patients, and hypofrac-
tionation in 3 (2.2, 2.2 and 2.75 Gy daily dose). The 
plan, pCT images and the delineated (target) con-
tours were exported to the Elekta Synergy X-ray 
Volumetric Imaging (XVI) System.
Imaging
Kilovoltage gantry mounted CBCT systems were 
used for daily on-line CBCT treatment verification. 
According to physician instructions, localisation 
was based on automatic spine or carina matching 
between CBCT and pCT with additional manual 
translation correction by RTT. All set-up errors 
were corrected before treatment delivery.
Study procedure
Retrospective rigid image registration was done in 
the Elekta XVI System. We used the first treatment 
verification CBCT image from the 1st, 10th, 15th, and 
20th fraction.
The CBCT image was retrospectively registered 
with pCT based on three different strategies: (a) 
bony registration on the spine, (b) soft tissue regis-
tration on the carina and (c) target (tumour/lymph 
node) matching on GTV/ITV. For 40 patients, we 
analysed 160 registration images and recorded 
1440 corrections in the x (lateral – LAT), y (longi-
tudinal – LONG) and z (vertical – VRT) directions. 
First two retrospective registrations on the 
spine and carina were performed by RTT and were 
based on automatic registration using a clip box 
(Figure 1).
Residual translation errors were corrected 
manually, if necessary. Next, reference registra-
tion on target (tumour/lymph node) matching was 
performed by an experienced thoracic radiation 
oncologist. Following automatic bone registra-
tion, translational misalignments were manually 
Radiol Oncol 2023; 57(1): 86-94.
But-Hadzic J et al. / Accuracy of volumetric lung image guidance88
corrected based on the visual adjustment of the 
GTV (ITV) contour on the CBCT image, to provide 
the best match for all known gross disease. If the 
lymph nodes were not visible, close anatomical 
surrogates were used. Translation corrections for 
target matching were recorded and used for the 
reference position. Spine and carina corrections 
were compared to the reference position and de-
viations in LAT, LONG and VRT measurements 
were analysed.
Statistics
Microsoft Excel 2010 and the Statistical Package for 
the Social Sciences, version 25.0 (SPSS Inc., Chicago, 
IL, USA) were used. General data were presented 
with descriptive statistics. Kolmogorov-Smirnov 
and Shapiro-Wilk tests rejected normal data distri-
bution. The Nonparametric Mann Whitney U test 
(MW), Wilcoxon Signed Ranks test and nonpara-
metric Kruskal-Wally’s test (KW) were used for 
the analysis of the set-up deviation. The Wilcoxon 
Signed Ranks Test for dependent samples was 
used for comparison (pairwise). The KW and MW 
were used when we analysed the differences in a 
certain measurement according in two (MW) or 
into multiple groups (KW). A p value ≤ 0.05 was 
considered statistically significant.
Results 
Patients, tumour and treatment 
characteristics
The patients, disease and treatment characteris-
tics are summarised in Table 1. Radiation was the 
primary treatment of “de novo” lung cancer in 
80% of patients. One patient was treated for local 
progression of epidermal growth factor receptor 
(EGFR)+ adenocarcinoma and one had postop-
erative regional recurrence. Five patients received 
reirradiation for local/regional recurrence. There 
were three plan adaptations after the 21st or 22nd 
fraction due to atelectasis (developing, resolution 
and worsening).
 Spine to target registration set-up 
deviation analysis
The set-up deviation in the LAT, LONG and VRT 
directions for the spine according to target regis-
tration for the 1st, 10th, 15th, and 20th fractions were 
analysed (Figure 2A). The best registration match 
was in the VRT direction, with a mean set-up devi-
ation between 1.2 and 1.68 mm. The biggest devia-
tion was detected in the LONG direction (2.03–2.73 
mm). Deviation differences from the 1st through 
20th fractions were not statistically significant in 
any direction. There was no time trend detected. 
Comparison of deviations between directions on 
the 20th fraction showed a significant set-up dif-
ference in deviation between LAT vs. LONG (p = 
0.002) and VRT vs. LONG (p = 0.000). The mean de-
viation for all set-up measurements was 1.39 mm 
in LAT (SD 0.9, range 0–4.0 mm), 2.44 mm in LONG 
(SD 1.6, range 0.5–8.0 mm) and 1.36 mm in VRT 
direction (SD 1.04, range 0–4.75 mm).
Carina to target registration set-up 
deviation analysis
Analysis of the set-up carina registration devia-
tions from the target set-up measurements for the 
1st, 10th, 15th, and 20th fractions were also analysed 
(Figure 2B). The smallest set-up difference was in 
the LAT direction (from 0.9 to 1.8 mm). The larg-
est discrepancies were detected in the LONG di-
rection (from 1.53 to 2.15). The difference in de-
viations for different fractions was not significant 
in any direction. There was no time trend in the 
deviations. Calculated from all the measurements, 
the mean deviation for the carina set-up deviations 
from the reference was 1.03 mm in LAT (SD 0.75, 
FIGURE 1. Automatic clip box carina registration with manual alignment check. 
Target contours (gross tumour volume [GTV] inner contour, planning target 
volume [PTV] outer contour) are imported for target registration.
Radiol Oncol 2023; 57(1): 86-94.
But-Hadzic J et al. / Accuracy of volumetric lung image guidance 89
range 0–3.75mm), 1.78 mm in LONG (SD 1.5, range 
0–6.75mm) and 1.33 mm in the VRT direction (SD 
1.25, range 0–5.25).
Comparison of spine to target and carina 
to target set-up deviation differences
The differences in the spine/target and carina/tar-
get mean set-up deviations were compared indi-
vidually for the different fractions and the mean 
for all measurements in all directions (Table 2). 
The registration on carina was more accurate ac-
cording to the reference in all measurements ex-
cept in the vertical direction on the 1st and 20th frac-
tions. The only significant difference was found on 
the 10th fraction VRT direction, with the smallest 
deviation for carina registration.
For all the measurements, registration on the ca-
rina was significantly closer to the reference regis-
tration in the LAT and LONG direction (p = 0.003 
and p = 0.002, respectively). We found no differ-
ence in the set-up deviation in the VRT direction 
for all measurements.
The impact of tumour location (central/
peripheral) and N stage on the spine/
target and carina/target registration 
deviation
Analysis of the spine registration deviations from 
the reference showed a significantly better regis-
tration match for centrally located tumours in LAT 
on the 1st and 10th fractions, in LONG on the 1st 
fraction and in VRT on the 10th fraction (Table 3).
The carina/target registration comparison 
showed significantly smaller differences for cen-
tral tumours in 50% LAT measurements (10th 
and 15th fraction), in 2/3 LONG measurements 
(10th –20th fraction), but not in the VRT direction 
(Table 4). We found no impact of the node stage 
on the spine/target registration deviations. When 
the carina was used for alignment, the differences 
were significantly smaller for N2 and N3 in the 
LAT 15th, LAT 20th, LONG 15th and VRT 20th frac-
tion (p = 0.034, 0.028, 0.025 and 0.034, respectively) 
(Supplementary Tables S1–S2). Possible time trend 
for spine/target LAT deviation difference was de-
tected for central tumours (Table 3).
TABLE 1. Patients, disease, and treatment characteristics
N = 40
Gender
Female 16 (40 %)
Male 24 (60 %)
Age (years) Median (range) 67 (53–81)
Tumour location*
RUL 15 (37.5%)
RML 3 (7.5%)
RLL 7 (17.5%)
LUL 8 (20%)
LLL 8 (20%)
Central (C)/
peripheral (P) 
tumour location**
C 17 (42.5%)
P 22 (55%)
Histology
NSCLC 31 (77.5%)
SCLC 9 (22.5%)
Disease treated
De novo lung cancer 32 (80%)
Local progression 1 (2.5%)
Local recurrence reirradiation 2 (5%)
Regional recurrence 1 (2.5%)
Locoregional recurrence reirradiation 4 (10%)
Systemic treatment
Concurrent chemotherapy 13 (32.5%)
Sequential chemotherapy 16 (40%)
Target therapy 1 (2.5%)
None 10 (25%)
Tumour (T) stage
T0 1 (2.5%)
T1 4 (10%)
T2 14 (35%)
T3 8 (20%)
T4 13 (32.5%)
Lymph nodes (N) 
stage
N0 9 (22.5%)
N1 1 (2.5%)
N2 14 (35%)
N3 16 (40%)
Fractionation
Conventional 37 (92.5%)
Hypofractionation 3 (7.5%)
Radiation 
technique VMAT 40 (100%)
* In one patient, two synchronous tumours were treated (RUL and LUL). 
**In one patient, only the lymph nodes were treated (regional recurrence).
LLL = left lower lobe; LUL = left upper lobe; NSCLC = non-small cell cancer; RML = right middle 
lobe; RLL = right lower lobe; RUL = right upper lobe; SCLC = small cell lung cancer; VMAT = 
volumetric modulated arc therapy
Radiol Oncol 2023; 57(1): 86-94.
But-Hadzic J et al. / Accuracy of volumetric lung image guidance90
FIGURE 2. Spine to target (A) and 
carina to target (B) registration 
set-up deviation in the lateral 
(LAT), longitudinal (LONG) and 
vertical (VRT) directions.
A
B
Radiol Oncol 2023; 57(1): 86-94.
But-Hadzic J et al. / Accuracy of volumetric lung image guidance 91
The impact of tumour location (central/
peripheral) and N stage on the spine/
target vs. carina/target deviation 
differences
The deviation differences analysis between the 
spine/target and carina/target registration accord-
ing to tumour location and N stage is shown in 
Supplementary tables S3–S5. For peripherally lo-
cated tumours, the carina registration deviation 
was only smaller compared to the spine set-up de-
viation in the 1st fraction LONG with a deviation 
difference of 1.91mm (p = 0.034). In centrally located 
tumours, the registration on carina was found to be 
significantly more accurate on the 15th and 20th frac-
tions LAT, (p = 0.012 and 0.048, respectively), the 
15th and 20th fractions LONG (p = 0.010 and 0.011, 
respectively) and for all LAT and LONG measure-
ments (p = 0.003). The deviation differences were 
0.76, 0.94, 1.59, 1.35, 0.5 and 0.71mm, respectively. 
There was no difference in the set-up deviation in 
the VRT direction regardless of tumour location.
Comparison of the deviation differences ac-
cording to the N stage showed a better correlation 
between the carina and target registration for N2 
and N3 disease. We found deviation differences 
TABLE 2. Mean set-up deviation difference (DD) comparison according to spine/target vs. carina/target registration
Fraction
LAT deviation 
mean (mm) p 
value
LONG deviation 
mean (mm) p 
value
VRT deviation 
mean (mm) p
valueSpine/ 
target
Carina/ 
target DD
Spine/ 
target
Carina/ 
target DD
Spine/ 
target
Carina/ 
target DD
1st 1.28 0.90 0.38 NS 2.65 1.80 0.85 NS 1.20 1.43 -0.23 NS
10th 1.30 1.08 0.22 NS 2.03 1.53 0.50 NS 1.68 1.10 0.58 0.04
15th 1.48 1.08 0.4 NS 2.38 1.65 0.73 0.05 1.38 1.38 0 NS
20th 1.53 1.05 0.48 NS 2.73 2.15 0.58 NS 1.2 1.43 -0.23 NS
All measurements 1.39 1.03 0.37 0.003 2.44 1.78 0.66 0.002 1.36 1.33 0.03 NS
 DD = deviation difference; LAT = lateral; LONG = longitudinal; NS = non-significant (p>0.05); target = primary tumour and lymph nodes; VRT = vertical
TABLE 3. Spine/target registration deviation according to tumour location
Fraction
LAT mean
(mm) p
value
LONG mean
(mm) p
value
VERT mean
(mm) p
value
Central Peripheral Central Peripheral Central Peripheral
1st 0.71 1.77 0.048 1.18 3.91 0.002 1.12 1.32 NS
10th 0.76 1.77 0.008 1.53 2.41 NS 1.41 1.86 0.017
15th 1.47 1.55 NS 2.59 2.32 NS 1.35 1.45 NS
20th 1.94 1.27 NS 2.29 3.14 NS 1.29 1.09 NS
LAT = lateral; LONG = longitudinal; NS = non significant (p>0.05); target = primary tumour and lymph nodes; VRT = vertical
TABLE 4. Carina/target registration deviation according to tumour location
Fraction
LAT mean
(mm) p
value
LONG mean
(mm) p
value
VERT mean
(mm) p
value
Central Peripheral Central Peripheral Central Peripheral
1st 0.65 1.14 NS 1.65 2.00 NS 1.06 1.77 NS
10th 0.53 1.55 0.002 1.18 1.86 0.041 0.82 1.32 NS
15th 0.71 1.41 0.049 1.00 2.23 0.027 0.88 1.77 NS
20th 1.00 1.14 NS 0.94 3.14 0.019 0.88 1.82 NS
LAT = lateral, LONG = longitudinal, NS = non significant (p>0.05); VRT = vertical; target = primary tumour and lymph nodes
Radiol Oncol 2023; 57(1): 86-94.
But-Hadzic J et al. / Accuracy of volumetric lung image guidance92
of 0.63mm N2 and 1.17mm N3 in LONG, 0.48mm 
N3 in LAT for all measurements and 0.94mm N3 
10th fraction LONG (p = 0.013, 0.015, 0.002 and 0.007, 
respectively). A small but significant 0.5mm differ-
ence in favour of the carina registration was found 
for N0 in all measurements LAT (p = 0.034). Because 
only 1 patient had the N1 stage, we excluded his 
measurements from this analysis (Supplementary 
Table S5).
Discussion
Since the introduction of CBCT into treatment 
verification, new insights to target position uncer-
tainties have evolved6, suggesting that image guid-
ance could currently be the weakest link in the ra-
diotherapy procedure.3,9 Only a few studies have 
explored the carina as a surrogate in volumetric 
lung image guidance7,10,11 and there is a lack of 
knowledge in this field. Here we present new data 
on the accuracy of the carina vs. spine surrogate 
compared to the target alignment as a reference in 
image registration.
A separate analysis of the spine and carina reg-
istration set-up differences compared to the ref-
erence (target) registration showed the smallest 
differences in the VRT and LAT directions, with 
the largest deviation in the LONG direction for 
both strategies (Fig. 2). The differences were not 
significantly different in time and no time trend 
was detected for the cohort, so we evaluated all the 
measurements together and again showed a good 
registration match in the VRT and LAT directions, 
but in LONG direction, the range of misalignment 
reached above 5 mm in both registration strategies. 
The greatest area of uncertainty in the LONG di-
rection was also shown in the study by Ottosson et 
al. where they compared the spine and target reg-
istration in free-breathing and breath-hold CBCT 
in locally advanced lung cancer patients. The 
largest intra- and inter-fractional misalignments 
were found in the LONG direction, independent 
of the registration method.12 Although we found 
a smaller mean LONG misalignment in the carina 
vs. spine registration (1.78 vs. 2.44 mm), uncertain-
ties in image registration should be considered for 
both strategies, with the enlargement of the PTV in 
the craniocaudal (CC) direction.
To determine the best registration match com-
pared to the reference, we compared differences 
in the set-up deviations for the spine and carina 
alignment. Both strategies were equally accurate in 
the VRT direction, but the carina was more accu-
rate in the LONG and LAT directions in all meas-
urements. The accuracy of the spine vs. carina reg-
istration was also tested in a study by a Canadian 
group, where four independent observers auto-
matically and manually aligned the first fraction 
CBCT with the pCT in 30 lung cancer patients.7 
They used spine, carina and tumour registration 
strategies. Automatic spine and carina registra-
tions provided similar tumour coverage, with the 
tumour inside the ITV in 60% of observations. The 
same group, in a second study, verified the tumour 
(T) and lymph node (LN) coverage following spine 
and carina registration for initial, middle, and final 
fraction CBCT. Both strategies improved the com-
bined target coverage throughout the treatment 
course compared to tattoo alignment. Carina bet-
ter improved the combined coverage and showed 
significantly superior nodal coverage compared to 
the spine, without compromising primary tumour 
coverage.10 With a significantly better registration 
match in the LONG and LAT directions, our data 
supports the suggestion from the Canadian group 
that the carina may be superior to the spine in the 
image guidance of locally advanced lung cancer.
The second Canadian study showed similar pri-
mary tumour coverage, regardless of the registra-
tion strategy.10 But in our study, the registration ac-
curacy was influenced by tumour location. Both the 
spine and carina alignment showed smaller set-up 
differences for centrally located tumours with the 
difference being more significant in the carina reg-
istration LONG (66% fractions) and LAT (50% frac-
tions). For peripheral tumours, we found no differ-
ence in the accuracy of spine vs. carina alignment, 
but for centrally located tumours, the carina was 
more accurate in the LAT and LONG directions. 
Our data suggests that the carina is a better sur-
rogate for centrally located tumours. Importantly, 
we also showed that the carina is as accurate as 
the spine for registration in peripherally located 
tumours and can be proposed as a registration sur-
rogate regardless of the tumour location.
According to our data, carina matching can also 
be used regardless of the nodal stage. We found no 
differences in the spine, but a better match for ca-
rina alignment in N2 and N3 disease. In advanced 
nodal disease, the carina showed superior registra-
tion vs. the spine in LAT and LONG. The finding 
is in concordance with the second Canadian study, 
where carina matching improved the node cover-
age compared to spine registration.10 Importantly, 
we found no set-up difference for N0 disease sug-
gesting that the carina and spine can equally be 
used as a surrogate in this stage. As there was 
Radiol Oncol 2023; 57(1): 86-94.
But-Hadzic J et al. / Accuracy of volumetric lung image guidance 93
only a single case of early nodal disease, we can-
not make any conclusions for N1. The reliability of 
spine matching for LN coverage was investigated 
in the study by Mohammed et al., where an equal 
geographical target miss was found for spine vs. 
combined target matching, but inferior LN cover-
age in the case of tumour matching, based on a 
weekly 4D CBCT registration.9 Because the same 
geographical miss was shown for hilar and medi-
astinal lymph nodes, we can hypothesize that ca-
rina matching could safely be used in N1 as well.
Target alignment should be “the ground truth” 
and have an impact on the therapeutic ratio,3,12 
but is rather difficult to implement clinically. Due 
to the poor soft tissue contrast on CBCT and tu-
mour changes during the course of treatment, 
only half of the tumours can clearly be contoured 
using CBCT.13 For these reasons, we used con-
toured-based registration for reference. A similar 
approach was used for target coverage and geo-
graphical miss assessments studies9,10 and target 
registration in the study by Ottosson et al.12 Direct 
tumour registration is also impractical because 
it is time-consuming and unreliable due to high 
intra-observer variability.7 In contrast, the carina 
is clearly visible on the CBCT and automatic ca-
rina alignment shows high reproducibility, supe-
rior even to spine alignment.7 Although the carina 
shows some respiratory movement, especially in 
the CC direction, it is still an excellent surrogate 
for directly adjacent mediastinal lymph nodes. 
Preliminary data also suggests a good correlation 
between the carina and GTV motion.7,11
We acknowledge the limitations of our study, 
which was retrospective and relatively small. We 
also present a heterogeneous group of patients, 
though they represent real everyday radiothera-
pists’ lung cancer patients. In our study, CBCT im-
age changes were not systematically determined. 
Because the majority of changes appear in the first 
five weeks of radiotherapy,14 we analysed CBCT 
images from the 1st, 10th, 15th and 20th fractions. 
Maybe the 25th and 30th fraction CBCT should have 
been included since we found cases for plan modi-
fication in the 6th week of treatment. Nevertheless, 
we detected no significant set-up differences in 
time. The rate of plan adaptation was low (7.5%), 
suggesting the need for a “traffic light” protocol 
implementation.3
In conclusion, carina registration was shown to 
be feasible, fast, and reproducible. Our data shows 
that, compared to target registration, carina is 
equally as accurate as spine registration, with su-
perior accuracy in the LONG and LAT directions. 
The carina is a better surrogate than the spine for 
alignment in centrally located tumours, N2 and 
N3 disease. Although spine alignment can equally 
be used in N0 and peripheral tumours, for simplic-
ity we propose that the carina should be the pri-
mary surrogate for the target in image guidance in 
locally advanced lung cancer. The next step should 
be to incorporate carina registration uncertainties 
into the PTV margin.
Acknowledgement
This research was supported by Slovenian re-
search programme for comprehensive cancer con-
trol SLORApro (P3-0429). 
Reference
1. Spigel DR, Faivre-Finn C, Gray JE, Vicente D, Planchard D, Paz-Ares LG, et al. 
Five-year survival outcomes with durvalumab after chemoradiotherapy in 
unresectable stage III NSCLC: an update from the PACIFIC trial. J Clin Oncol 
2021; 39: 8511-11. doi: 10.1200/JCO..39.15_suppl.8511
2. Vrankar M, Stanic K, Jelercic S, Ciric E, Vodusek AL, But-Hadzic J.  Clinical 
outcomes in stage III non-small cell lung cancer patients treated with 
durvalumab after sequential or concurrent platinum-based chemoradio-
therapy - single institute experience. Radiol Oncol 2021; 55: 482-90. doi: 
10.2478/raon-2021-0044
3. Tvilum M, Khalil AA, Møller DS, Hoffmann L, Knap MM. Clinical outcome of 
image-guided adaptive radiotherapy in the treatment of lung cancer patients. 
Acta Oncol (Madr) 2015; 54: 1430-7. doi: 10.3109/0284186X.2015.1062544
4. Mercieca S, Belderbos JSA, van Herk M. Challenges in the target volume 
definition of lung cancer radiotherapy. Transl Lung Cancer Res 2021; 10: 
1983-98. doi: 10.21037/tlcr-20-627
5. Chan C, Lang S, Rowbottom C, Guckenberger M, Faivre-Finn C. 
Intensity-modulated radiotherapy for lung cancer: current status and 
future developments. J Thorac Oncol 2014; 9: 1598-608. doi: 10.1097/
JTO.0000000000000346
6. Clarke E, Curtis J, Brada M. Incidence and evolution of imaging changes 
on cone-beam CT during and after radical radiotherapy for non-small 
cell lung cancer. Radiother Oncol 2019; 132: 121-6. doi: 10.1016/j.ra-
donc.2018.12.009
7. Higgins J, Bezjak A, Franks K, Le LW, Cho BC, Payne D, et al. Comparison of 
spine, carina, and tumor as registration landmarks for volumetric image-
guided lung radiotherapy. Int J Radiat Oncol Biol Phys 2009; 73: 1404-13. 
doi: 10.1016/j.ijrobp.2008.06.1926
8. Purdie TG, Bissonnette J-P, Franks K, Bezjak A, Payne D, Sie F, et al. Cone-
beam computed tomography for on-line image guidance of lung ste-
reotactic radiotherapy: localization, verification, and intrafraction tumor 
position. Int J Radiat Oncol Biol Phys 2007; 68: 243-52. doi: 10.1016/j.
ijrobp.2006.12.022
9. Mohammed N, Kestin L, Grills I, Shah C, Glide-Hurst C, Yan D, et al. 
Comparison of IGRT registration strategies for optimal coverage of primary 
lung tumors and involved nodes based on multiple four-dimensional CT 
scans obtained throughout the radiotherapy course. Int J Radiat Oncol Biol 
Phys 2012; 82: 1541-8. doi: 10.1016/j.ijrobp.2011.04.025
10. Lavoie C, Higgins J, Bissonnette J-P, Le LW, Sun A, Brade A, et al. Volumetric 
image guidance using carina vs spine as registration landmarks for conven-
tionally fractionated lung radiotherapy. Int J Radiat Oncol Biol Phys 2012; 84: 
1086-92. doi: 10.1016/j.ijrobp.2012.02.012
Radiol Oncol 2023; 57(1): 86-94.
But-Hadzic J et al. / Accuracy of volumetric lung image guidance94
11. van der Weide L, van Sörnsen de Koste JR, Lagerwaard FJ, Vincent A, van 
Triest B, Slotman BJ, et al. Analysis of carina position as surrogate marker for 
delivering phase-gated radiotherapy. Int J Radiat Oncol Biol Phys 2008; 71: 
1111-7. doi: 10.1016/j.ijrobp.2007.11.027
12. Ottosson W, Rahma F, Sjöström D, Behrens CF, Sibolt P.  The advantage of 
deep-inspiration breath-hold and cone-beam CT based soft-tissue registra-
tion for locally advanced lung cancer radiotherapy. Radiother Oncol 2016; 
119: 432-7. doi: 10.1016/j.radonc.2016.03.012
13. Lim G, Bezjak A, Higgins J, Moseley D, Hope AJ, Sun A, et al. Tumor 
regression and positional changes in non-small cell lung cancer dur-
ing radical radiotherapy. J Thorac Oncol 2011; 6: 531-6. doi: 10.1097/
JTO.0b013e31820b8a52
14. Elsayad K, Kriz J, Reinartz G, Scobioala S, Ernst I, Haverkamp U, et al. Cone-
beam CT-guided radiotherapy in the management of lung cancer: diagnostic 
and therapeutic value. Strahlenther Onkol 2016; 192: 83-91. doi: 10.1007/
s00066-015-0927-y