Professional Documents
Culture Documents
Abstract
Introduction: The aim of this study was to compare the use of CBCT imaging may be a valuable tool for the evaluation of healing of endodontic
assessment of healing after endodontic microsurgery us- surgery. (J Endod 2017;43:1072–1079)
ing 2-dimensional (2D) periapical films versus 3-
dimensional (3D) cone-beam computed tomographic Key Words
(CBCT) imaging. Methods: The healing of 51 teeth Apicoectomy, cone-beam computed tomography, endodontic, healing, microsurgery,
from 44 patients was evaluated using Molven’s criteria outcome, root-end surgery, success
(2D) and modified PENN 3D criteria. The absolute area
(2D) and volume (3D) changes of apical lesions preoper-
atively and at follow-up were calculated by segmenta-
tion using OsiriX software (Pixmeo, Bernex,
E ndodontic microsurgery
uses high-magnification,
ultrasonic root-end prepa-
Significance
CBCT evaluation allowed for a precise volumetric
Switzerland) and ITK-Snap (free software). Results: analysis of preoperative periapical lesions and the
ration and biocompatible
There was a significant difference between the mean assessment of healing after endodontic microsur-
root-end filling materials.
preoperative lesion volumes of 95.34 mm3 (n = 51, Success rates in the range gery. Healing classification in 3D (CBCT) analysis
standard deviation [SD] 196.28 mm3) versus of 90% have been reported was significantly different from 2D (periapical radi-
6.48 mm3 (n = 51, SD 17.70 mm3) at follow-up for endodontic lesions ography) analysis.
(P < .05). The mean volume reduction was 83.7%. Pre- (1, 2). Most original
operatively, mean lesion areas on periapical films were studies used Molven’s criteria for the assessment of healing after endodontic
13.55 mm2 (n = 51, SD 18.80 mm2) and 1.83 mm2 surgery, including potential clinical symptoms and radiographic healing based on
(n = 51, SD .68 mm2) at follow-up (P < .05). Accord- periapical radiographs.
ing to Molven’s criteria, 40 teeth were classified as com- Cone-beam computed tomographic (CBCT) imaging is a widely accepted tool for
plete healing, 7 as incomplete healing, and 4 as diagnostic evaluation in dentistry. However, its main limitation is radiation exposure.
uncertain healing. Based on the modified PENN 3D Indications in endodontics include the detection of periapical lesions, fractures, or per-
criteria, 33 teeth were classified as complete healing, forations; the evaluation of complex root anatomy, existing root fillings, and the location
14 as limited healing, 1 as uncertain healing, and 3 as of separated instruments; surgical treatment planning; and the diagnosis of traumatic
unsatisfactory healing. The variation in the distribution injuries to teeth or the alveolar bone (3, 4). In surgical treatment planning, CBCT
of the 2D and 3D healing classifications was significantly imaging is helpful to assess the extent and location of apical periodontitis; the bone
different (P < .05). Periapical healing statuses incom- thickness over pathologic defects; and the proximity to anatomic structures such as
plete healing or uncertain healing according to Molven’s the mental nerve, sinus cavity, or adjacent teeth.
criteria could be clearly classified using 3D criteria. Con- Studies have shown that CBCT imaging is superior for the detection of apical
clusions: CBCT analysis allowed a more precise evalua- periodontitis when compared with periapical radiographs (5–7). The risk-benefit
tion of periapical lesions and healing of endodontic ratio in terms of radiation exposure outweighs the use of CBCT imaging for regular
microsurgery than periapical films. Significant differ- follow-ups after endodontic procedures unless the stage of healing is difficult to
ences existed between the 2 methods. Over the observa- discern. Few studies compared 2-dimensional (2D) and 3-dimensional (3D) healing
tion period, the mean periapical lesion sizes significantly for primary endodontic treatment (8, 9) or endodontic surgery (10–12). No
decreased in volume. Given the correct indications, the investigation compared the outcome assessment for endodontic microsurgery
From the *Private Practice, Nuremberg, Germany; †Department of Endodontics, School of Dental Medicine, University of Pennsylvania, Philadelphia, Pennsylvania;
and ‡Poliklinik fuer Zahnerhaltung, Parodontologie und Endodontologie, Heinrich-Heine-Universitaet Duesseldorf, Duesseldorf, Germany.
Address requests for reprints to Dr Frank C. Setzer, School of Dental Medicine, University of Pennsylvania, 240 S 40th Street, Philadelphia, PA 19104. E-mail address:
fsetzer@upenn.edu
0099-2399/$ - see front matter
Copyright ª 2017 American Association of Endodontists.
http://dx.doi.org/10.1016/j.joen.2017.02.007
JOE — Volume 43, Number 7, July 2017 2D and 3D Healing Assessment after Endodontic Surgery 1073
Clinical Research
randomized in ITK-SNAP using a semiautomatic segmentation algorithm allowing for additions and/or subtractions as small as individual voxels
sequence (18). Briefly, ITK-SNAP allowed for a simultaneous represen- (eg, in areas where foreign body materials prevented the automatic seg-
tation of the sagittal, axial, and coronal views. The periradicular defects mentation from working correctly). The Neuroimaging Informatics
were selected as subregions using a grayscale value range selection tool Technology Initiative allowed for direct expression of the number of vox-
after reading the grayscale values of the defects. These defect and els and conversion into mm3. If a multirooted tooth had more than 1
volume-specific grayscale values were individually selected to allow periapical lesion, the individual defect volumes were added together.
for semiautomatic defect volume recognition and calculation using auto- All cases included in the study had teeth with an original defect at least
mated spherical fillers (‘‘bubbles’’) that populated the defect in repeated twice the width of the periodontal ligament (6, 7, 19–21). For follow-up,
runs until the defect volume was completely filled. If necessary, a manual any radiolucent volumes with less than twice the width of the periodontal
correction was applied at this stage of the defect volume measurements, ligament were counted as a 0-mm3 defect size and considered healed.
Figure 2. Manual 2-dimensional segmentation of a periapical defect on a maxillary left first molar. (A) A radiograph imported into ITK-SNAP software. (B) Eval-
uation of lesion area by manual drawing.
Statistical Analysis patterns were part of these scans. CBCT images of 33 teeth were ac-
The nonparametric Wilcoxon signed rank test, paired and un- quired to rule out any odontogenic involvement in which symptoms
paired t tests, and analysis of variance tests were applied using SPSS associated with sinusitis, temporomandibular or occlusal dysfunctions,
v.20 (SPSS/IBM, Armonk, NY). or marginal gingivitis had been diagnosed.
Of 44 patients, 18 were men and 26 were women with a mean age
of 45.9 years. A total of 80 roots underwent surgery. There were 23 teeth
Results in the mandible and 28 in the maxilla, including 11 anteriors, 11 pre-
Between 2011 and 2013, a total of 385 patients received endodon- molars, and 29 molars. Twenty periapical lesions were located on a sin-
tic microsurgery by the first author. Of these, a subpopulation of 44 pa- gle root, 12 teeth had separate lesions on 2 roots, and 19 teeth had
tients had CBCT scans taken at a follow-up appointment, including 51 combined lesions affecting more than 1 root. Twenty-three teeth had
teeth that complied with the inclusion/exclusion criteria. For 13 teeth, received surgery on 1 root, 27 on 2 roots, and 1 tooth on 3 roots.
the scans were taken by the first author (T.S.) to ascertain the stage of The mean follow-up period was 23.7 months, ranging from 12 to
periapical healing. An additional 5 teeth with conclusive healing 37 months.
JOE — Volume 43, Number 7, July 2017 2D and 3D Healing Assessment after Endodontic Surgery 1075
Clinical Research
Healing Assessment with 2D Radiographs (n = 51, mean = 1.83 mm2) at follow-up. This reduction in lesion
Randomized blinded evaluation resulted in 40 teeth being classified size was statistically significant (P < .05, paired t test).
as complete healing, 7 teeth as incomplete healing, and 4 teeth as uncer-
tain healing (47/51 [92.2%] successful healing) according to Molven’s Healing Assessment with 3D CBCT Volumes
criteria. The kappa value for the initial agreement was 90.2%; 5 cases The outcome assessment based on 3D analysis resulted in 33 teeth
were resolved by discussion (F.S. and M.K.). The mean area size being classified as complete healing, 14 teeth as limited healing, 1 tooth
measured on the periapical radiographs was 13.55 mm2 (n = 51, stan- as uncertain healing, and 3 teeth as unsatisfactory healing according to
dard deviation [SD] 18.80 mm2) preoperatively and 1.83 mm2 the modified PENN 3D criteria (47/51 [92.2%] successful healing). The
kappa value for the initial agreement was 88.2%; 6 cases were resolved 2D periapical films and the 3D CBCT images after at least 1 year of
by discussion (F.S. and T.S.). Preoperatively, the mean absolute lesion follow-up was rejected.
volume was 95.34 mm3 (n = 51, SD 196.28 mm3) and 6.48 mm3
(n = 51, SD 17.70 mm3) at follow-up. This change in the mean vol-
ume was statistically significant (P < .05, paired t test). Multiple lesions 3D Evaluation of Nonconclusive Healing Patterns
per tooth or combined lesions were evaluated as a single volume. Pre- The initial healing evaluation was performed by using periapical
operatively, the absolute largest volume was 654.1 mm3; the smallest radiographs (T.S.); however, for 13 teeth, the healing pattern was
was 1.5 mm3. The mean reduction was 83.7%. For 38 teeth, the volume inconclusive, and CBCT volumes were taken. The retrospective 2D eval-
reduction was 100%; for 7 cases, it was >90%; for 2 cases, it was <90%, uation for this investigation using Molven’s criteria (F.S. and M.K.) clas-
and for 4 cases, it was <0% (increase of lesion volume). There was a sified 2 of these teeth as complete healing, 7 as incomplete healing, and
statistical correlation between the 3D criteria and volume reduction 4 as uncertain healing. CBCT analysis allowed for all of these teeth to be
(P < .05, analysis of variance; uncertain healing and unsatisfactory heal- placed into the 3D healing categories of the modified PENN 3D criteria
ing were grouped because of the small sample size). (T.S.and F.S.). One tooth each was classified as 1A, 1D, and 3A; 2 teeth
each as 2A, 2D, and 4; and 4 teeth as 2B. Based on these results, the
second hypothesis that if healing assessment was inconclusive with a
Comparison of 2D and 3D Healing periapical radiograph, it would also be inconclusive using CBCT imag-
Scores were assigned to the individual 2D and 3D healing cate- ing was rejected.
gories (1 for complete healing [2D/3D], 2 for incomplete healing
[2D] and limited healing [3D], 3 for uncertain healing [2D/3D], and
4 for unsatisfactory healing [2D/3D]) and compared using the paired Discussion
nonparametric Wilcoxon signed rank test. The overall mean score for Endodontic surgery requires regular follow-up to evaluate the
2D healing was 1.29 (n = 51, SD 0.61) and 1.49 (n = 51, healing process. A 1-year follow-up is commonly accepted as a good
SD 0.81) for 3D healing; when dichotomized for success and failure, indication for the final outcome (22, 23). Thus, a mean follow-up
it was 1.15 (2D) (n = 47, SD 0.36) and 1.30 (3D) (n = 47, period of 23.7 months was deemed sufficient for an adequate analysis
SD 0.46) and 3.0 (2D) (n = 4, SD 0.0) and 3.75 (3D) (n = 4, of 2D versus 3D assessment.
SD 0.5), respectively. The distribution of the healing categories be- CBCT volumes were taken with the Veraviewepocs 3De CBCT ma-
tween 2D and 3D analysis was significantly different (z = 1.8, chine, providing a voxel size of 125 mm in the limited field of view mode
P < .05). The mean lesion reduction on periapical radiographs versus at a setting providing 40% less irradiation compared with standard pro-
that assessed on CBCT volumes was compared indirectly. The square gramming. This resulted in an effective dose of 19–44 mSv compared
roots of the preoperative to follow-up area reduction (2D) (n = 51, with single digital radiographs in which it may range between 0.6 and
mean = 2.33 mm, SD 2.77 mm) were compared with the cubic roots 5.0 mSv (24). The increased radiation exposure of CBCT imaging
of the preoperative to follow-up volume reduction (3D) (n = 51, mean limited the available sample size compared with the overall number
= 3.14 mm, SD 2.40 mm), which was significantly different (P < .05, of patients who had received surgery between 2011 and 2013. Thus,
paired t test). Therefore, the first hypothesis that there would be no dif- the overall distribution of successful versus failed cases may not be
ferences in the outcome classifications derived from the assessment of representative for the overall population, and as a result the
JOE — Volume 43, Number 7, July 2017 2D and 3D Healing Assessment after Endodontic Surgery 1077
Clinical Research
dichotomized ratio of successful versus failed cases was not reported as The difference in assessment between 2D and 3D evaluation was
an outcome rate. statistically significant. Aggregating the scores assigned for statistical
The primary reasons for CBCT acquisition were assessment of evaluation, complete healing (1), incomplete/limited healing (2), un-
inconclusive healing and evaluation of odontogenic versus nonodonto- certain healing (3), and unsatisfactory healing (4) resulted in a sum
genic symptoms. Although possibly not representative for the overall of 66 for the 2D criteria and 76 for the 3D criteria evaluation, demon-
population, this subpopulation allowed for the investigation of the 2 strating a trend for a worse outcome with 3D evaluation. Together with
central hypotheses (ie, that there were no differences in outcome clas- the statistically significant differences in lesion area (2D) versus lesion
sifications based on 2D periapical films versus 3D CBCT images and that volume (3D) reduction, this led to the rejection of the first hypothesis.
inconclusive healing assessment would not be different between 2D and This is in agreement with von Arx et al (12), who described worse sit-
3D imaging). According to Friedman (25), incomplete healing versus uations for about one third of cases evaluated, but in disagreement with
uncertain healing could be easily misinterpreted using Molven’s Christiansen et al (10), and Tanomaru-Filho et al (11), who found no
criteria. Indeed, 11 teeth included in the initial sample of 13 CBCT im- differences in radiographic healing assessment. However, these studies
ages taken for inconclusive healing patterns fell into the categories compared planar changes of lesion sizes on periapical radiographs with
incomplete healing or uncertain healing based on the 2D Molven clas- selected sagittal and/or coronal views in CBCT volumes and not volu-
sification for this study. Furthermore, 3D evaluation revealed that most metric changes.
difficulties in assessment were related to the summation effect of oral The use of square roots of lesion areas and cubic roots of lesion
radiographs, including overprojection of the incisive or mental volumes allowed for an objective comparison of the assessment of
foramina, zygomatic arch, or linea mylohyoidea, which is in agreement lesion changes. Clinically, a 2D lesion area addresses the mesiodistal
with previous studies (21, 26–28). CBCT evaluation allowed for a and apicocoronal extent of the defect. The result that mean cubic roots
precise evaluation of all related cases, which is in accordance with derived from the CBCT volumes were significantly larger than mean
findings that showed good repeatability and reproducibility for CBCT square roots derived from the periapical radiographs demonstrated
evaluation of healing after apical surgery (29). that buccolingual directions on average had the largest extent of a peri-
The quality of the 3D segmentation depended on various factors, radicular defect. This finding implies that, when clinically relying on
such as device-specific parameters, settings and stability, artifacts, and periapical radiographs, the buccolingual extent of a defect is not only
homogeneity of the density of the object. The precision of CBCT 3D seg- the unknown dimension but also the largest of the 3 dimensions.
mentation was verified by in vitro investigation of artificial periapical As of today, increased radiation doses administered by CBCT im-
bone defects (30). However, there is a difference between artificial, aging limit its use for regular follow-up for endodontic surgery. It
clearly limited defects and random borders of an in vivo pathologic should only be considered if appropriate. However, if radiation intensity
bone defect, which explains the need for manual corrections can be minimized by improved hardware and/or software algorithms,
throughout the segmentation process. Although CBCT imaging may this study may suggest its use for follow-up of endodontic microsurgery
have limitations in regard to accurately representing smaller defects, because of its superior accuracy compared with periapical radiographs.
digital periapical radiography was shown to be even less accurate
(31). Because most studies define a periapical defect to be at least twice Conclusion
the width of a healthy periodontal ligament space, it can be assumed that CBCT imaging allowed a more precise evaluation of periapical le-
periapical defects were correctly identified in this study (6, 7, 19–21). sions and healing than periapical films. Significant differences existed
Because the study was retrospective, no standardized radiographs were between the 2 methods. Over the observation period, the periapical le-
taken. This may lead to slight angulation changes; however, all sions significantly decreased in volume. Given the correct indications,
radiographs were taken with the same sensor, software, and film the use of CBCT imaging is recommended as a valuable tool for end-
holder by the same operator, limiting the variability. odontic microsurgery.
Preoperative radiographs and volumes were compared with follow-
up periapical films and CBCT images. No CBCT volumes were taken
immediately after the surgical intervention because of radiation con- Acknowledgments
cerns, so the results of this study reflect the absolute changes from before The authors deny any conflicts of interest related to this study.
endodontic surgery until follow-up, with a mean reduction in lesion vol-
ume from 95.34 mm3 (preoperatively) to 6.48 mm3 (at follow-up). The References
percentage of volume reduction calculated after segmentation showed a
1. Setzer FC, Shah SB, Kohli MR, et al. Outcome of endodontic surgery: a meta-analysis
statistical correlation with the 3D criteria. Analogue to the Molven of the literature–part 1: comparison of traditional root-end surgery and endodontic
criteria, the worst score of any root was representative for the tooth microsurgery. J Endod 2010;36:1757–65.
for 3D analysis. Although 38 cases showed a 100% volume reduction, 2. Setzer FC, Kohli MR, Shah SB, et al. Outcome of endodontic surgery: a meta-analysis
only 33 teeth were classified as complete healing in 3D analysis. In total, of the literature–part 2: comparison of endodontic microsurgical techniques with
and without the use of higher magnification. J Endod 2012;38:1–10.
an equal number of cases were classified as successful or failed after 3. Patel S, Durack C, Abella F, et al. European Society of Endodontology position state-
blinded and randomized evaluation for both 2D and 3D imaging because ment: the use of CBCT in endodontics. Int Endod J 2014;47:502–4.
of the dichotomized grouping of complete healing (1) and incomplete/ 4. Schulze R, Deppe H, Betz W et al. Dentale digitale volumentomographie. S2k-
limited healing (2) for success and uncertain healing (3) and unsatis- Leitlinie. Available at: http://www.awmf.org/uploads/tx_szleitlinien/083-005l_
factory healing (4) for failure. However, some cases evaluated as com- S2k_Dentale_Volumentomographie_2013-10.pdf. Accessed March 30, 2017.
5. De Paula-Silva FW, Wu MK, Leonardo MR, et al. Accuracy of periapical radiography
plete healing (1) based on Molven’s criteria were reclassified as limited and cone-beam computed tomography scans in diagnosing apical periodontitis us-
healing (2) using CBCT imaging, and uncertain healing (3) cases based ing histopathological findings as a gold standard. J Endod 2009;35:1009–12.
on Molven’s criteria were changed to unsatisfactory healing (4) using 3D 6. Patel S, Wilson R, Dawood A, Mannocci F. The detection of periapical pathosis using
criteria. For example, 5 maxillary molars that healed with intact lamina periapical radiography and cone beam computed tomography - part 1: pre-
operative status. Int Endod J 2012;45:702–10.
dura and periodontal ligament covering the resected root surface but 7. Patel S, Wilson R, Dawood A, et al. The detection of periapical pathosis using digital
demonstrated a physiologic fenestration or a thin, undetectable buccal periapical radiography and cone beam computed tomography - part 2: a 1-year
plate received the 3D criteria classification ‘‘2D.’’ post-treatment follow-up. Int Endod J 2012;45:711–23.
JOE — Volume 43, Number 7, July 2017 2D and 3D Healing Assessment after Endodontic Surgery 1079