H. J. Naoum, R. M. Love, N. P. Chandler & P. Herbison
Departments of Oral Rehabilitation, Stomatology, Preventive and Social Medicine, University of Otago, Dunedin, New Zealand.

Introduction.
Successful root canal treatment requires effective chemo-mechanical debridement to eradicate intraradicular infection and proper canal shaping to facilitate obturation. The root canal geometry may have a direct impact on the thoroughness and extent of debridement and canal shaping (Peters et al. 2001). Many methods have been used to investigate root canal anatomy in vitro, each has its own advantages and disadvantages with variable degrees of accuracy and procedural complexity. However, with the exception of radiography, these methods cannot be applied in vivo. Usually, the processes either destroy the specimen rat best preclude the unimpeded entry of instruments into the root canal.
Careful preoperative assessment of root canal anatomy obtained from a diagnostic radiograph is a key prerequisite for thorough canal preparation and, ultimately, successful therapy. The technique of choice for endodontic radiography is the long cone paralleling technique, which creates an accurate radiograph with minimal distortion and a high level of reproducibility (Fava & Dummer 1997). Intentional alteration of the X-ray beam angulation may provide additional information not readily available from the orthoradial view. The recommended horizontal beam angulation for identification of two canals in one root is dependent on the amount of separation and divergence between canals and is reported to lie between 20 and 408 (Walton 1973, Klein et al. 1997, Martinez-Lozano et al. 1999). In assessing the canal configuration of mandibular first molar teeth, a 308mesial horizontal beam angulation was suggested to provide additional detail not provided by 0 or 208 angulation (Walton1973).
Contrast media are radiopaque substances that can be introduced into various parts of the body to artificially alter subject contrast (Whaites1992).The use of radiopaque materials to enhance the contrast of a root canal system is not new. Barrett (1925) presented radiographs from clinical cases where radiopaque material had been forced into empty root canals to demonstrate the presence of lateral branches. Other investigators have used a variety of radiopaque materials to study canal anatomy in extracted teeth. Similar principles of removing the pulp tissue and introducing radiopaque material with the aid of centrifuging or vacuum were used (Barker et al.1969, Lowman et al.1973, Hession 1977, Mayo et al. 1986, Thomas et al.1993). Although this method reveals the complex anatomical characteristics of the root canal system, it cannot be applied in vivo. More recently, Scarfe et al. (1995) passively injected contrast medium into the root canal to determine the presence of lateral and secondary canals. Similarly, to improve radiographic interpretation, Shearer et al. (1996) injected iohexol contrast medium passively into the root canals of extracted human teeth. Contrast medium may have many applications in endodontic radiography, e.g. to improve radiographic interpretation of root canal systems in teeth with aberrant anatomy, for the detection of root perforations and ledges and in the diagnosis of internal resorption. Shearer et al. (1996) concluded that contrast media could be a valuable aid in radiographic interpretation of root canal systems and suggested the need for further clinical studies to assess the efficacy of this technique. The aims of this study were to compare the accuracy of evaluator interpretation of the root canal anatomy on radiographs taken at 0 or 308mesial angulation without or with radiopaque contrast medium introduced into the root canal by incremental passive injection or infusion under vacuum. Furthermore, the effectiveness of the passive injection technique was assessed by comparing it to the laboratory-accepted method to introduce the contrast medium into the root canal system.

Materials and methods.
Twenty extracted human mandibular first molar teeth with sound roots and mature apices were collected. Unimpeded coronal access to canals in the mesial and distal roots was gained and canal patency was ensured using a size 8 K-file (Caulk, Dentsply, Milford, DE, USA). In order to avoid cutting and modifying the canal anatomy and plugging the openings of small lateral canals by dentinal debris, pulp tissue from each canal was removed with 20 mL 2.5% NaOCl irrigation using a 31- gauge needle (Endo-Eze, Ultradent Products, UT, USA) and mechanical agitation with blunted size 8 K-files. Each canal was then irrigated with 10 mL of water and dried with paper points.
The body of a dried human mandible was split in half, longitudinally, using a rotating saw with the plane of section extending mesiodistally. The teeth and cancellous bone were removed, leaving the buccal and lingual cortical plates intact. To simulate cancellous bone, a silicone laboratory putty (Bayer Dental, Dormagen, Germany) was mixed with sawdust and the roots of each sample were embedded in the putty mixture between the cortical plates such that each tooth was custom-fitted into the bone. The cortical plates were held together on a wax base on the rotating table of a jig attached to a film holder and beam-aiming device. Angulations were scribed onto the rotating part of the jig to indicate the horizontal angle of the X-ray beam. A soft tissue equivalent to15-mmplasticine was attached with in the beam alignment ring (Fig. 1). The sample and dental X-ray unit were positioned to obtain a 26-cm focal spot-object distance and a 2-cm object-film distance. Conventional E-speed radiographic film (Kodak Co., Rochester, NY, USA) was exposed ata 0or 308mesialhorizontal angulation (Walton 1973) with the X-ray unit operating at 65 kVp, 8 mA, for 0.44 s (Gendex, Milan, Italy). These settings were determined in a pilot study to produce the best quality radiographs.

Figure 1. Jig for standardized radiographic images:
a, soft tissue equivalent in aiming ring;
b, section of mandible with tooth in position;
c, film holder;
d, angle indicator.

Iohexol is a tri-iodated, nonionic, water-soluble contrast medium with low osmotic pressure and a molecular weight of 821.14 (iodine content 46.4%). The product has been tested in both in vitro and in vivo and has been shown to be safe (Aakhus et al.1980, Mutzel et al.1980). To aid visual reference of contrast medium in the root canal, one drop of 0.1% methylene blue dye was mixed with 1mL of iohexol radiopaque contrast medium (Omnipaque1, Nycomed, Birmingham, UK). Increments of the mixture were injected passively into the canals with a 31-gauge needle placed to the most apical level possible. A blunt size 8 K-file was introduced1mmshort of the apical foramen after the first increment and manipulated in each canal for 1min. File insertion was repeated for another minute after the second increment to enhance penetration of the contrast medium and expel air bubbles. A second pair of radiographs was taken at the previous angles and settings. The teeth were then removed from their moulds and immersed in iohexol and subjected to vacuum (24 in. mmHg_1) for 2min. The vacuum was reapplied for further 2 min until air bubbles ceased escaping from the access cavity. Samples were carefully cleaned without disturbing the canals and replaced into their moulds for repositioning in the mandible and a third pair of radiographs exposed. All films were developed in an automatic processor (All-Pro 2000M, Hicksville, NY, USA) using Kodak GBZ developer and fixer solutions (Kodak Co., Rochester, NY, USA).
The mesial root on all images (Fig. 2) was examined and assessed by three evaluators, a specialist endodontist and two experienced general practitioners. The assessment included the number of visible canals, whether the canal(s) was visible along the entire length of the root, the location of the canal terminus in relation to the radiographic apex, configuration of canal(s) based on Pineda (1973), level of multiple canals merging and the presence of lateral canals. The aim of the experiment was explained to the evaluators who were given verbal instructions and schematic diagrams to assist them in their answers. The evaluators were first allowed to examine and assess images not included in the study by answering the assessment questionnaire. Points of confusion were clarified using the schematic drawings of the possible findings as re-enforcement of assessment criteria may improve evaluator agreement (Reit 1987, Saunders et al.1999). The images were examined with a x2 magnifier on a light viewing box (Magni Viewer III, Proyrex, Tokyo, Japan) with peripheral light excluded. No time limit was set for assessments and the image sequence was randomized for the evaluators. The viewing was conducted in two sessions no less than 1week apart. Finally, to accurately determine the root canal anatomy, each tooth was rendered transparent (Fig. 3) with a clearing technique (Kelsen et al.1999), and examined using a stereomicroscope (Olympus, Tokyo, Japan) at x10 magnification and the same assessment criteria. Evaluation of the cleared teeth represented the gold standard.

Figure 2. Representative images of the 0 and 308 radiographs without and with contrast medium.
A and B; 0 and 308 radiographs without contrast medium,
C and D; 0 and 308 radiographs with contrast medium introduced by passive injection,
E and F; 0 and 308 radiographs with contrast medium introduced by vacuum.

The experimental data were compared to the gold standard using logistic regression with the standard errors adjusted for clustering on samples. The Kappa test was applied to determine the interevaluator reliability. To determine the intraevaluator reliability, radiographs of four samples were randomly selected and reassessed after 3 months by the three evaluators. Data were analysed using the Kappa test and values were defined as: 0.81-1.00 represented very good agreement, 0.61-0.80 represented good agreement and 0.41-0.60 represented moderate agreement (Landis & Koch1977).

Figure 3. The mesial root of the specimen in Fig. 2 following clearing.

Effect of X-ray beam angulation.

Images without contrast medium.
Results presented in Table 1 show that evaluators found the 08 radiograph to be significantly better than the 308 image (P < 0.05) for correctly detecting the number of canals, canal visibility, location of canal terminus and the level at which the two canals merged. There was no significant difference between the two angles in determining the canal configuration and detection of lateral canals.
Images with contrast medium For images contrasted by passive injection, there was a significant difference (P < 0.05) between the 0 and 308 images in determining the canal visibility and canal terminus, with the 08 radiographs being better for predicting the correct anatomy. There was no significant difference between the two angles for the other criteria (Table 2). With the use of vacuum to introduce the contrast medium into the root canal system, the 08 radiographs were significantly better (P < 0.05) for assessing canal visibility whilst the 308 radiographs were better (P < 0.05) for determining the canal configuration (Table 2).
Effect of radiopaque contrast medium Results presented in Table 3 demonstrate the effect of using radiopaque contrast medium in assessing canal anatomy at the two angles. In 08 radiographs, contrast medium introduced passively into the root canal system significantly improved canal visibility (P < 0.05).When introduced by vacuum, the contrast medium did not significantly aid in assessing the root canal system on the 08 radiograph (P > 0.05). However, when the 308 radiographs were evaluated, passive injection of the contrast medium solution significantly improved (P < 0.05) the assessment of the number of canals, canal visibility and canal configuration. When introduced by vacuum, contrast medium significantly improved (P < 0.05) the assessment of the number of canals, canals visibility, canal terminus and canal configuration. Although there were some individual differences between passive and vacuum contrast images, overall there was no significant difference (P > 0.05) in evaluation of the root canal anatomy in images contrasted by passive injection or vacuum.
The results of the Kappa test showed ‘very good’ intraevaluator reliability for detecting the number of canals and canal visibility. The intraevaluator Kappa values for the canal terminus, canal configuration and the level of canals merging were within the ‘good’ range. A Kappa value could not be obtained for the lateral canals because of the extremely high prevalence of one of the outcomes. The Kappa test showed ‘good’ agreement between evaluators for number of canals, canal visibility, canal configuration and the level of canals merging. For canal termination there was ‘moderate’ agreement between evaluators. NoKappa value could be obtained for lateral canals due to the high prevalence of one of the outcomes (Table 4).

Table 1. Comparison between 0 and 308 noncontrasted radiographs.



Table 2. Comparison between 08 and 308 contrasted radiographs.



Table 3. Comparison between noncontrasted, passive or vacuum contrasted images.



Table 4. Intra-  and interevaluator reliability.

Discussion.
A radiograph is a two-dimensional image of a three dimensional object. A shift of the X-ray beam from the orthoradial position may provide additional information compared to the 08 projection. The results of this study show that when noncontrasted images taken at 0 and 308 angulations were compared, the 08 radiographs were significantly better than 308 projections for detecting the number of canals, canal visibility along the entire root length, determining the level of canal termination and the level of merging of two canals (Table 1). This is in contrast to other studies that suggest that 20-408 is the optimum beam angulation for detecting the number of canals in one root (Walton1973, Klein et al.1997, Martinez-Lozano et al. 1999). Klein et al. (1997) evaluated the optimum horizontal beam angulation for detecting two canals in lower incisors embedded in a silicone impression material and a plastic tray. Martinez-Lozano et al. (1999) radiographed maxillary and mandibular premolars without the effect of the surrounding bone and soft tissue. The wider buccolingual dimension of the mesial root of a mandibular first molar, and therefore potential separation of two canals compared to the corresponding dimension in a lower incisor, combined with the differences in the laboratory techniques may contribute to these conflicting results.
Altering beam geometry may separate structures on a radiographic film; however, it simultaneously produces an inherent amount of image distortion. Image distortion at the 308 horizontal angulation was evident in this study (Fig. 2B) and may have contributed to the difficulty in determining some characteristics of the root canal system on the radiograph. Moreover, in some of the 08 images it was obvious that the two canals and other features of the root canal system could be clearly visualized. This may have been due to the position of the two canals in different anatomical planes, or most probably the 08 angulation was not true in some samples but was at a slight rotation due to anatomical and experimental variables, and this may have aided interpretation of the 08 radiographs. Clinically, intended 08 projections may show the presence of two canals in the mesial roots of mandibular first molars. This is likely to be due to the difficulty in precisely projecting the two canals in one plane. In fact, a slight angulation of the incident beam may be beneficial in this respect.
When the two angles were compared in passively contrasted images,08 radiographs were significantly better than 308 radiographs for canal visibility and determination of canal terminus only (Table 2). Comparing the two angles in images contrasted by vacuum revealed that the 08 radiograph was still better for canal visibility; however, the 308 radiograph was better for determining canal configuration (Table 2).These results were mainly due to increased diagnostic yield of the contrasted 308 radiographs which approached or exceeded the amount of information obtained from noncontrasted 08 and 308 radiographs. In general, introducing the radiopaque contrast medium by passive injection improved radiographic detection of four of the assessment criteria in 08 radiographs (Tables 1and 2) but only the canal visibility reached the significance level (Table 3). Similarly, when 308 radiographs were considered, passive injection of iohexol improved detection of five of the assessment criteria (Tables 1 and 2) with the number of canals, canal visibility and canal configuration being significantly better than the noncontrasted 308 radiographs (Table 3). When contrast medium was introduced with the aid of vacuum, it significantly improved detecting the number of canals, canal visibility, canal terminus and canal configuration (Table 3). The difference in detecting canal termination between the passive injection and vacuum may be explained by the possible entrapment of air bubbles with passive injection precluding proper visualization of the most apical part of the root canal. Interestingly, in both contrasted and noncontrasted images, the 08 radiograph was always significantly better than the 308 images for canal visibility (Tables 1and 2).
Subject contrast is a result of the variation in X-ray transmission through different parts of a patient’s tissues due to variable tissue thickness, density and composition. Radiopaque materials introduced into parts of the body, including the root canal system, artificially alter radiographic contrast. The results indicate that the use of intraradicular contrast medium primarily improved the diagnostic yield of the 308 radiograph and this was probably related to the increase in root canal contrast compensating the effect of the image distortion (Fig. 2 B-D,F). The increased amount of information obtained from contrasted images supports the results of Shearer et al. (1996), and are in agreement with Scarfe et al. (1995) who found that radiopaque contrast medium passively introduced into the root canal did not improve the detection of lateral canals. In the present study, radiographs were taken from two angulations to potentially improve the detection of lateral canals in the bucco-lingual and mesio-distal planes. Furthermore, no active dentine cutting was involved to avoid possible plugging of accessory canals, whilst optimal viewing conditions were ensured by exclusion of the peripheral light and the use of magnification. Possible explanations that lateral canals were not readily detected are that the passive injection of the contrast medium was insufficient to allow the solution to penetrate into the patent canals and/or the resolution of the radiograph was inadequate to recover the detail of the lateral canals.
In several in vitro studies of root canal anatomy, radiopaque materials were introduced into the root canal system by negative pressure or centrifuging (Barker et al. 1969, Lowman et al. 1973, Hession 1977, Mayo et al. 1986,Thomas et al.1993). The studies demonstrated that this method was reliable and accurate for determining root canal anatomy. It would therefore be of use to know if a clinical technique employed to aid evaluation of root canal anatomy can produce similar results compared to the accepted laboratory techniques. In the present study every attempt was made to follow steps that could be applied in vivo. The method used to remove the pulp tissue with 2.5% sodium hypochlorite irrigation and mechanical agitation with blunted hand files is clinically achievable. Furthermore, the samples were radiographed with simulated cortical plates and cancellous bone and soft tissue to mimic the scatter arising from these tissues in vivo. Our results demonstrated that the technique for introducing iohexol by passive injection and file manipulation provided radiographic detail similar to a radiograph with contrast medium introduced into the root canal system by negative pressure, and suggest that the technique can be applied clinically.
The Kappa test is a widely used and accepted assessment for agreement when dealing with nominal and ordinal scales and it incorporates a correction for chance agreement (Brennan & Silman1992). In general, a value above 0.61 represents a high standard of agreement and an acceptable level of reproducibility. The high level of intraevaluator agreement indicated an understanding of the definitions of the given criteria making it possible to reproduce the answers to the assessment questionnaire. In this case the criteria definitions were probably correlated with the actual and correct definitions, since all the questions for assessment were clarified for evaluators (Saunders et al. 1999). Similarly, agreement between evaluators may also have been attributed to evaluator calibration by prior explanation and re-enforcement of assessment criteria.
Contrast media are usually used in high doses for imaging procedures with no or minimum risk. Rarely, allergic reactions may develop and patients particularly at risk are the elderly or very young children, patients with a history of allergic reaction to the contrast media, diabetics and patients with cardiac or renal failure (Whaites 1992). The amount of contrast media required to fill a root canal space is very small and is incomparable to the doses used in other imaging procedures. In addition, a careful passive injection with file manipulation within the con- fines of the root canal should extrude no or a minimum amount of the radiopaque material. These facts suggest that the technique can be applied in clinical dentistry. It is unknown whether iohexol can be completely washed out of a root canal or if it has any deleterious effect on sealers or filling materials. However, in this experiment and in a pervious study (Shearer et al. 1996), it was noted that the solution could be washed out with copious irrigation.

References.

Aakhus T, Sommerfelt SC, Stormorken H, Dahlstrom K (1980) Tolerance and excretion of iohexol after intravenous injection in healthy volunteers. Acta Radiologica Supplementum 362, 131-4.
Barker BCW, Lockett BC, Parsons KC (1969) The demonstration of root canal anatomy. Australian Dental Journal 18, 37-41.
Barrett MT(1925) The internal anatomy of the teeth with special reference to the pulp with its branches. Dental Cosmos 67, 581-92.
Brennan P, Silman A (1992) Statistical methods for assessing observer variability in clinical measures. British Medical Journal 304,1491-4.
Fava LRG, Dummer PMH (1997) Periapical radiographic techniques during endodontic diagnosis and treatment. International Endodontic Journal 30, 250-61.
Hession RW(1977) Endodontic morphology. Part I. An alternative method of study. Oral Surgery, Oral Medicine and Oral Pathology 44, 456-62.
Kelsen AE, Love RM, Kieser JA, Herbison P (1999) Root canal anatomy of anterior and premolar teeth in Down's syndrome. International Endodontic Journal 32, 211-6.
Klein RMF, Blake SA, Nattress BR, Hirschmann PN (1997) Evaluation of X-ray beam angulation for successful twin canal identification in mandibular incisors. International Endodontic Journal 30,58-63.
Landis JR, Koch GG (1977) The measurement of observer agreement for categorical data. Biometrics 33,159-74.
Lowman JV, Burke RS, Pelleu GB (1973) Patent accessory canals: incidence in molar furcation region. Oral Surgery, Oral Medicine and Oral Pathology 36,580-4.
Martinez-Lozano MA., Forner-Navarro L, Sanchez-Cortes JL (1999) Analysis of radiologic factors in determining premolar root canal systems. Oral Surgery, Oral Medicine, Oral Pathology, Oral Radiology and Endodontics 88,719-22.
Mayo CV, Montgomery S, del Rio C (1986) A computerized method for evaluating root canal morphology. Journal of Endodontics12, 2-7.
MutzelW, Siefert H-M, Speck U (1980) Biochemical-pharmacologic properties of iohexol. Acta Radiologica Supplementum 362,111-5.
Peters OA, LaibA, Gohring TN, Barbakow F (2001) Changes in root canal geometry after preparation assessed by high-resolution computed tomography. Journal of Endodontics 27, 1-6.
Pineda F (1973) Roentgenographic investigation of the mesiobuccal rot of the maxillary first molar. Oral Surgery, Oral Medicine and Oral Pathology 36, 253-60.
Reit C (1987) The influence of observer calibration on radiographic periapical diagnosis. International Endodontic Journal 20,75-81.
Saunders MB, Gulabivala K, Holt R, Kahan RS (1999) Reliability of radiographic observation recorded on a proforma measured using inter- and intraobserver variation: a preliminary study. International Endodontic Journal 33, 272-8.
Scarfe WC, Fana CR, Farman AG (1995) Radiographic detection of accessory/lateral canals: use of RadioVisioGraphy and hypaque. Journal of Endodontics 21,185-90.
Shearer AC, Wasti F, Wilson NHF (1996) The use of a radiopaque contrast medium in endodontic radiography. International Endodontic Journal 29, 95-8.
ThomasRP,MouleAJ, Bryant R(1993) Root canal morphology of maxillary permanent first molar teeth at various ages. International Endodontic Journal 26, 257-67.
Walton RE (1973) Endodontic radiographic technics. Dental Radiography and Photography 46, 51-9.
Whaites E (1992) Specialized investigations and imaging modalities. In: Essentials of Dental Radiography and Radiology, 1st edn. London, UK: Churchill Livingstone,159-67.