Home Print this page Email this page Users Online: 41
Home About us Editorial board Search Ahead of print Current issue Archives Submit article Instructions Subscribe Contacts Login 

 Table of Contents  
ORIGINAL ARTICLE
Year : 2023  |  Volume : 13  |  Issue : 1  |  Page : 44-50

Selection of the best endodontic treatment option using data mining: A decision tree approach


1 Department of Biostatistics, Proteomics Research Center, School of Allied Medical Sciences, Shahid Beheshti University of Medical Sciences, Tehran, Iran
2 Iranian Center for Endodontic Research, Research Institute for Dental Sciences, Shahid Beheshti University of Medical Sciences, Tehran, Iran

Date of Submission24-May-2022
Date of Decision13-Jul-2022
Date of Acceptance15-Jul-2022
Date of Web Publication11-Jan-2023

Correspondence Address:
Prof. Saeed Asgary
Iranian Center for Endodontic Research, Research Institute for Dental Sciences, Shahid Beheshti University of Medical Sciences, Evin, Tehran 1983963113
Iran
Login to access the Email id

Source of Support: None, Conflict of Interest: None


DOI: 10.4103/sej.sej_97_22

Rights and Permissions
  Abstract 

Introduction: The presence of postendodontic pain is an important issue, which can affect the patients' quality of life. Appropriate treatment selection, based on specific characteristics (e.g., clinical test results and patients' demographics), may reduce postendodontic pain. We aimed to evaluate the relationship of data mining algorithms in longitudinal data of postendodontic pain and treatment allocation to predict the best treatment option.
Materials and Methods: The pain data of an original multicenter randomized clinical trial with two study arms, pulpotomy with mineral trioxide aggregate (PMTA) (n = 188) and root canal therapy (RCT) (n = 168), were used. The linear mixed-effects model and predictive algorithms were fitted in accordance with the personal characteristics of patients and diagnostic test results to determine the best treatment option. Using SPSS 23, SAS 9.1, and WEKA 3.6.9, the preferred treatment was identified via comparing the areas below the receiver operating characteristic curves and identifying the most appropriate algorithm. In addition, a decision tree was used to allocate the best type of treatment modality to reduce posttreatment pain.
Results: For <18-year-old patients with irreversible pulpitis (IP) based on cold test and >18-year-old patients whose electrical pulp test (EPT) exhibited IP, the chosen treatment would be RCT (P < 0.05). However, for >18-year-old patients with IP based on cold test and <18-year-old patients whose EPT revealed IP, the recommended treatment would be PMTA (P < 0.05).
Conclusions: The decision tree model seems to be able to predict the reduction of postendodontic pain in ~65% of patients if they receive optimal treatment.

Keywords: Calcium-silicate cements, decision trees, endodontics, medical informatics, mineral trioxide aggregate, pain, pulpotomy


How to cite this article:
Baghban AA, Zayeri F, Eghbal MJ, Parhizkar A, Asgary S. Selection of the best endodontic treatment option using data mining: A decision tree approach. Saudi Endod J 2023;13:44-50

How to cite this URL:
Baghban AA, Zayeri F, Eghbal MJ, Parhizkar A, Asgary S. Selection of the best endodontic treatment option using data mining: A decision tree approach. Saudi Endod J [serial online] 2023 [cited 2023 Feb 3];13:44-50. Available from: https://www.saudiendodj.com/text.asp?2023/13/1/44/367524


  Introduction Top


Endodontic pain is one of the chief causes of patient complaints and is frequently experienced in-between or after root canal treatments; however, primary endodontic pain mainly initiates from untreated dental caries.[1] Owing to the multifactorial nature of postendodontic pain, many researchers have tried to link the pain to predictive factors, i.e., tooth/patient factors and pre/intraoperative settings.[2] Various tooth and individual/personal factors of each patient can affect the process of the needed treatment and cause different patients to respond differently to the same stimuli,[3],[4] e.g., an intervention. Thus, it can be assumed that a material/technique may not be completely appropriate to be applied for all patients.[5] Therefore, type of the treatment (technique/material), symptoms of the disease (before and after the treatment), and effectiveness of the treatment performed on a large number of patients can be used to evaluate various algorithms of patient responses via statistically examining the data related to the disease. Moreover, endodontic pain can significantly affect the patients' quality of life.[2] Hence, determination of a proper treatment option to control pain at different stages of dental pulp pathosis is a key factor that should be invariably considered.[6]

Recently, because of the fast technology outreach and the considerable data/information growth, key progress has been achieved in several fields, including medicine/dentistry.[7] Nonetheless, the analysis/computational methods of data are a main biomedical challenge.[8] Besides, data mining techniques have rapidly developed and an increasing number of data mining algorithms are currently being used to construct/predict/analyze biomedical data.[9]

Models called learning algorithms can be employed to measure the efficacy of treatment modalities.[10] Learning algorithms include (i) linear classifiers (i.e., logistic regression, linear discriminate analysis [LDA], and support vector machine [SVM] with a radial kernel) and (ii) nonlinear classifiers (i.e., decision tree, neural network, nonlinear SVM with a radial kernel, and random forest) to fit the forecast models.[5],[7] All the models can process and classify data; however, they do not fully examine the impact of classification. In other words, the validity of classification methods is not measured in these processes. Therefore, it is important to apply a complete subgroup process and validation of the methods.[5] While subgrouping can be done based on the prediction models, the problem of its appropriateness and which model is best remains unresolved. Therefore, validation procedures have been engendered to select the best prediction model under a marginal regression context.[11]

In the current research, and for the first time in dentistry, a process was used to investigate the validity of these predictive models. In addition, given the possibility of using data mining methods to investigate the effectiveness of two types of endodontic treatment (pulpotomy with mineral trioxide aggregate [PMTA] and root canal therapy [RCT]) on the reduction of postendodontic pain (based on clinical test results and patients' demographics), the present interdisciplinary study aimed to evaluate the relationship of data mining algorithms in longitudinal data of postendodontic pain and treatment allocation to predict the best treatment option.


  Materials and Methods Top


Seven Iranian academic departments of endodontics were selected based on their declaration of readiness to participate in the multicenter study. The obtained data (pain intensity during 7-day follow-ups of 356 participants) were used to correlate postendodontic pain reduction with clinical test results and patients' demographics. The inclusion criteria were carious pulp exposures of vital mature permanent molars with/without irreversible pulpitis (IP)/apical periodontitis. Based on the sign/symptoms and clinical examinations (e.g., pulp sensibility tests and percussion test), the pulp and periapical diagnoses were made. In the original multicenter randomized clinical trial,[2] using a computer-generated system, each patient was randomly assigned to either RCT group (n = 168) or PMTA group (n = 188) and then treated by postgraduate students. In the RCT group, after pulp exposure, an access cavity was prepared and instrumentation was performed using the BioRaCerotary system (FKG Dentaire, Switzerland). Employing a master cone (35–40/0.04) for straight and (35/0.02) for curved canals, the root canals were filled and sealed. In the PMTA group, after pulp exposure, full pulpotomy was performed, and subsequent to hemostasis, canal orifices were covered with ProRoot mineral trioxide aggregate (MTA) (Dentsply, OK, USA). The coronal cavities of all treated teeth were filled using sandwich techniques. The pain intensity index was recorded by patients at the baseline and 10-time points posttreatment (6, 12, 18, 24, and 36 h and 3, 4, 5, 6, and 7 days) with a numerical rating scale. The statistical mixed-effects model was employed because the pain index was recorded in ten time intervals and therefore the quantities were time dependent.[5] The employed model led to the estimation of the correlation between pain intensities at different time intervals for each patient. According to the results, patients were separated into two different groups due to the reduction/nonreduction of the pain irrespective of the received treatment method. The used computation formula for model #1 is as follows:

Yij = β0 + α0i + (β11i) ZiTij2Tij+eij, i = 1,…,n j=1,…,ni,(1)

Where α0i and α1i are the random intercept and slope for subject i, respectively; eij and Tij are random errors and the index of time j, respectively. Zi was chosen as either − 1 or 1 for the treatment assignment (RCT or PMTA) with a probability of 0.5. While β1 represents the average of the treatment effect over the time, α1i allows us to take into account individual differences.

To evaluate the method, we first fragmented the data into two smaller datasets with 70% of them in the training dataset and 30% in the testing. The training dataset was employed to fit model #1. Then binary independent variables (i.e., sex, age, presence of filling, bleeding after the exposure of the pulp, the results of sensitivity/percussion/cold tests and electrical pulp test [EPT], and the presence of apical lesion) were used to estimate α1i, which act as characteristics of the patient. Since we classified the patients into one of the two subgroups in different machine learning algorithms, the desired prediction model was specified as f(xi) = P (ci = 1|xi), where f(.) is a function representing the association between ci and xi. The estimated sum of α1i and β1 coefficients in the model was employed to assess the effect of each treatments assigned to patients. In this step, due to the multiplicity of factors affecting the treatment allocations, based on the expert clinician's opinion, the results of sensitivity/percussion tests, bleeding after the exposure of the pulp, and age of patients were considered important. These factors were used as predictive variables in modeling. Due to the indefinite relationship between these factors and treatment outcomes, linear algorithms including logistic regression, SVM, and LDA and nonlinear algorithms including decision tree, random forest, neural network, and SVM with a radial kernel were employed to fit predictive models.[5]

Then, we employed model #1 to estimate the correlation between pain indices at different time intervals for each patient. The pain index was expected to decrease over time for patients receiving optimal treatments. Therefore, patients whose total sum of coefficients were more than were considered as the population who did not receive appropriate treatment (Group 1). However, patients with an estimated sum of coefficients of <0 were considered as the group receiving appropriate treatment (Group 1).

Finally, the validity between each learning algorithm subgroup and subgroup based on model #1 was measured via an assessment of the testing dataset. Then, the comparison receiver operating characteristic curve and sublevel surface index led to the document and sublevel surface index, which led to the most appropriate predictive algorithm documentation. The model was recognized as the best model in which the area beneath the receiver operating characteristic curve was closer to 1 and kappa coefficient had the highest value compared to other models. Software SPSS 23 (SPSS Inc., Chicago, IL, USA) and SAS 9.1 (SAS Institute Inc., Cary, NC, USA) were used to fit the mixed effects for the measurement of the efficacy of each treatment.WEKA3.6.9 environment was employed to fit data mining model algorithms for the prediction of optimal treatment.[5],[7]


  Results Top


Three hundred and fifty-six participants from 7 Iranian academic departments of endodontics from March 2017 to March 2018 entered the study and were followed-up for 7 days. Pretreatment conditions (e.g., mean pain intensity, endodontically involved teeth, clinical observations [sensitivity test], and radiographic examinations) and intraoperative conditions (e.g., mean size of pulp exposure, type of bleeding, duration of treatment, and restoration methods) were similar in the two treatment groups (P > 0.05). 83.3% of patients were over 18 years of age. The results of percussion test and presence of periapical lesion showed that 6% and 6.8% of patients were positive for the mentioned conditions, respectively. Cold test showed that in 70% of patients, pulp was vital and uninflamed whereas EPT revealed that 24.3% of patients had IP. Severe bleeding was observed after pulp exposure in ~10% of participants [Table 1].
Table 1: Comparison of patients' demographics, preoperative, intraoperative and postoperative factors in the two study groups

Click here to view


By fitting the mixed-effects models to the data set, the effect of randomly assigned treatments to each patient was measured over time. Computing these values and comparing them to the number 1 permitted us to do the initial classification of patients indicates that the treatment was appropriate for the individual). According to the above classification, 256 patients responded positively to each of the received treatments (MTA = 147 (57.42), RCT = 109 (42.58]), ultimately leading to a reduction in postendodontic pain.

Among the fitted algorithms, the random forest model and decision tree were designated as the most appropriate models. The area under the receiver operating characteristic curve and kappa curves were calculated as values “0.76 and 0.67” and “0.425 and 0.393” for the two mentioned models, respectively. Moreover, these two algorithms showed the least errors compared to other models. The decision tree model was employed to allocate the therapy due to its interpretability [Table 2].
Table 2: Comparison of linear and nonlinear predicting algorithms

Click here to view


Based on the results obtained from the decision tree on diagnostic tests of pulp sensitivity and age of patients, it was shown that the patient's age should be initially considered when predicting the allocation of endodontic treatments, i.e., PMTA or RCT. In other words, patients >18 should be tested with cold test; however, if the patient is <18, EPT should be used (P < 0.05). Therefore, for patients >18 who have shown IP based on the results of the cold test, and patients <18 who have revealed IP using EPT, PMTA is recommended (P < 0.05). On the contrary, patients >18 who exhibited IP due to EPT, or patients <18 showing IP using cold test, require RCT (P < 0.05). However, it was shown that if decision-making was based on the sensitivity to percussion and patient's age, the presence of periapical lesions in radiographic examination would not make a significant difference in clinical decision-making (P < 0.05) [Figure 1].
Figure 1: Age-based allocation of treatments using sensitivity and EPT tests. EPT: Electrical pulp test, RCT: Root canal therapy, PMTA: Pulpotomy with mineral trioxide aggregate

Click here to view



  Discussion Top


In the current study, allocation of the selected endodontic treatments with their most favorable outcomes for each patient based on their personal characteristics was investigated. Furthermore, the modality, which could lead to a significant reduction in postendodontic pain index, was determined. Owing to large amount of data analysis that could not be conducted with traditional methods, data mining algorithms were used to extract information from sheer/huge volume of data.[7] Therefore, the results of patients' response to sensitivity tests and their age were used in algorithms of data mining prediction. Numerous studies have focused their research on the prediction of an optimal treatment for all patients; however, they have failed to provide individual treatments as well as validated algorithms, which could help predict such ministrations. Nonetheless, the present study addressed the mentioned concerns.[12],[13]

In the current study, we have proposed a comprehensive method from subgrouping to validation for personalized treatments in longitudinal data. To estimate the efficacy of RCT and the patient's recovery process, the pain index is examined frequently and over time.[14],[15] In fact, longitudinal data can be used to compare treatment modalities and evaluate the patient's responses to treatment over time to find the best practice.[16] The method starts by providing a random-effects linear model. The random effects in the model evaluate individual treatment effects over time, yet the fixed effects still allow researchers to look at the population as a whole. Since the changes in the random effects act as the variation among characteristics of the patients,[17] we employ different classification approaches to form prediction models based on the individual effects and characteristics of the patients;[18] linear/nonlinear classification approaches are considered since the association between the characteristics of the patient and the outcome is unbeknown in practice.

The methods of data mining were evaluated based on area under the receiver operating characteristic curve, including random forest, decision tree, nonlinear SVM with a radial kernel, nonlinear diagnostic analysis, linear SVM with a linear kernel, logistic regression, and LDA. The main reason for advantage of random forest algorithm in predicting the correct allocation of PMTA and RCT for the treatment of pulpal pathosis should be studied in the learning process. This is principally due to the significantly high predictive power of this model; however, the prediction occurs slowly after the training model was learned. The most important weakness of random forest algorithm is the large number of trees, which slows down and makes ineffective predictions in the real world.[19] On the other hand, nonparametric decision tree model, as a powerful tool in data mining, showed high accuracy in the prediction of the accurate treatment allocation to pulpal disease. The decision tree is regarded as a good method for data mining if a high percentage of variables is qualitative.[20] The most significant advantage of the method is high interpretability because of its tree structure.[12],[21] Nonetheless, one of the difficulties with this model is that at each stage of the algorithm implementation, it makes a separation process based on only one decision variable.[21] Among the methods considered for the study, LDA had the weakest predictive results. LDA is a parametric method that is dependent on the acceptance of normality of several variables and is sensitive to scattered points. In spite of the advantages, coefficients of this model are not interpretable, which makes it not suitable for classification purposes.[22] Thus, decision tree was used to predict the appropriate treatment for each individual in the current research.

Dental carious lesions and the consecutive pain is a global public health problem, and as one of the main concerns of oral health, it could affect the quality of life of patients in all age groups.[23],[24],[25] Dental caries is often considered a cause which could lead patients to endodontics treatments.[2] Various modalities have long been used as treatments for pulp and periapical diseases. Nowadays, with the application of new biomaterials/techniques/equipment in dentistry, specifically endodontics, there has been a dramatic increase in the success rate of the treatments.[2] The novel methods, i.e., vital pulp therapy and regenerative endodontics, have gained success in the treatment of the pulp–periapical pathosis.[26],[27],[28],[29],[30] PMTA is an easy and inexpensive approach with successful outcomes and has stood out amongst such procedures.[31] In addition, PMTA means significantly less tissue damage in comparison to routine root canal treatment procedures. It has been shown that PMTA can be successfully performed for the treatment of IP in permanent molars and thus may be a more appropriate treatment than conventional methods. It is an important issue that clinicians as decision-makers would rely on the principles that can help them in real practice to predict well the outcomes of treatment decisions,[32] and this research opens a new approach for such a goal.

In the original study, the postendodontic pain after PMTA, as an alternative to RCT, was evaluated. The results showed that PMTA was very effective in pain relief and showed comparable postoperative pain reduction after the endodontic treatment.[2] Therefore, the high success rate of PMTA (78.1%) for the treatment of IP may shift the management of deep carious lesions in permanent teeth from RCT to vital pulp therapy.[33],[34],[35] Nevertheless, the results of each of these treatments in different individuals can have distinctive outcomes.[5]

The main advantage of the proposed approach in this study was that the methodology (i) measured the efficacy of each treatment modality and (ii) used learning algorithms; making them easier to use and more useful to apply. Nevertheless, owing to the fact that each algorithm differs from the other in terms of analysis of results, time of computation, classification, and availability of statistical software, the appropriate algorithm to be implemented in every study could also vary due to the nature of the relation of variables to the variable of response.[5]

Pulp sensibility testing, i.e., heat/cold/electric pulp tests, has been used generally in dental clinics; however, they do not detect pulpal blood flow as the real index of pulp tissue vitality. These tests have serious limitations such as their reliance on individuals' subjective responses and clinicians' interpretations. While the cold pulp test has indicated high diagnostic accuracy values among all pulp sensibility tests, EPT was more likely to correctly identify vital teeth rather than nonvital teeth.[36]

For more valid and reliable findings, the data mining approaches need huge samples; however, we had no choice other than to examine these limited data. Therefore, we suggest applying these processes with a larger sample size.

Further research would be necessary to apply our suggested longitudinal and predictive linear and nonlinear models with more independent variables to assess their specific effects in choosing the desired treatment.


  Conclusions Top


The decision tree model could correctly predict the appropriate treatment for ~67% of patients. Thus, if the optimal treatment is appropriately allocated to the patient based on individual characteristics, it may lead to further reduction of pain after endodontic treatment.

Acknowledgments

Iranian Center for Endodontic Research has managed/supported the study; the original trial was approved/supported by the Deputy Minister of Research, Iranian Ministry of Health and Medical Education. The authors thank Ms. Zahra Keumarsi who helped with the statistical analysis.

Financial support and sponsorship

Deputy Minister of Research, Iranian Ministry of Health and Medical Education, Tehran, Iran.

Conflicts of interest

There are no conflicts of interest.

 
  References Top

1.
Renton T. Dental (Odontogenic) Pain. Rev Pain 2011;5:2-7.  Back to cited text no. 1
    
2.
Eghbal MJ, Haeri A, Shahravan A, Kazemi A, Moazami F, Mozayeni MA, et al. Postendodontic pain after pulpotomy or root canal treatment in mature teeth with carious pulp exposure: A multicenter randomized controlled trial. Pain Res Manag 2020;2020:5853412.  Back to cited text no. 2
    
3.
Gong J, Lv J, Liu X, Zhang Y, Miao D. Different responses to same stimuli. Neuroreport 2008;19:671-4.  Back to cited text no. 3
    
4.
Nunes GP, Delbem AC, Gomes JM, Lemos CA, Pellizzer EP. Postoperative pain in endodontic retreatment of one visit versus multiple visits: A systematic review and meta-analysis of randomized controlled trials. Clin Oral Investig 2021;25:455-68.  Back to cited text no. 4
    
5.
Andrews N, Cho H. Validating effectiveness of subgroup identification for longitudinal data. Stat Med 2018;37:98-106.  Back to cited text no. 5
    
6.
Rossi MT, de Oliveira MN, Vidigal MT, de Andrade Vieira W, Figueiredo CE, Blumenberg C, et al. Effectiveness of anesthetic solutions for pain control in lower third molar extraction surgeries: A systematic review of randomized clinical trials with network meta-analysis. Clin Oral Investig 2021;25:1-22.  Back to cited text no. 6
    
7.
Çığşar B, Ünal D. Comparison of data mining classification algorithms determining the default risk. Sci Prog 2019;2019:8706505.  Back to cited text no. 7
    
8.
Norouzi M, Souri A, Samad Zamini M. A data mining classification approach for behavioral malware detection. J Comp Net Com 2016;2016:260905.  Back to cited text no. 8
    
9.
Zou Q, Mrozek D, Ma Q, Xu Y. Scalable data mining algorithms in computational biology and biomedicine. BioMed research international. 2017;2017.  Back to cited text no. 9
    
10.
Yeh YF, Chang CP. Applying data mining techniques to identify suitable activities. Math Prob Engin 2015;2015:618061.  Back to cited text no. 10
    
11.
Iyit N, Genc A. A constitution of linear mixed models (LMMs) in the analysis of correlated data: Random intercept model (RIM) for repeated measurements data. J Mod Math Stat 2009;3:60-8.  Back to cited text no. 11
    
12.
Kiang MY. A comparative assessment of classification methods. Decis Support Syst 2003;35:441-54.  Back to cited text no. 12
    
13.
Dreiseitl S, Ohno-Machado L. Logistic regression and artificial neural network classification models: A methodology review. J Biomed Inf 2002;35:352-9.  Back to cited text no. 13
    
14.
Fitzmaurice GM, Laird NM, Ware JH. Applied longitudinal analysis. John Wiley & Sons; 2012.  Back to cited text no. 14
    
15.
Searle SR, Gruber MH. Linear models. John Wiley & Sons; 2016.  Back to cited text no. 15
    
16.
Kanagaratnam S, Schluter PJ. The age of permanent tooth emergence in children of different ethnic origin in the Auckland region: A cross-sectional study. N Z Dent J 2012;108:55-61.  Back to cited text no. 16
    
17.
Diaz FJ. Measuring the individual benefit of a medical or behavioral treatment using generalized linear mixed-effects models. Stat Med 2016;35:4077-92.  Back to cited text no. 17
    
18.
Core Team R. R: A Language and Environment for Statistical Computing. Vienna: R Foundation for Statistical Computing; 2013.  Back to cited text no. 18
    
19.
Liaw A, Wiener M. Classification and regression by randomforest. R News 2002;2:18-22.  Back to cited text no. 19
    
20.
Harper PR. A review and comparison of classification algorithms for medical decision making. Health Policy 2005;71:315-31.  Back to cited text no. 20
    
21.
Tu JV. Advantages and disadvantages of using artificial neural networks versus logistic regression for predicting medical outcomes. J Clin Epidemiol 1996;49:1225-31.  Back to cited text no. 21
    
22.
Pohar M, Blas M, Turk S. Comparison of logistic regression and linear discriminant analysis: A simulation study. Metodoloski Zvezki 2004;1:143.  Back to cited text no. 22
    
23.
Hamedy R, Shakiba B, Fayazi S, Pak JG, White SN. Patient-centered endodontic outcomes: A narrative review. Iran Endod J 2013;8:197-204.  Back to cited text no. 23
    
24.
Alroudhan IE, Ravi J, Magar SS, Alam MK, Alsharari KN, Alsharari FM. Oral health-related quality of life and satisfaction after root canal treatment according to operator expertise: A longitudinal prospective study. Saudi Endod J 2021;11:388.  Back to cited text no. 24
  [Full text]  
25.
Ezzat AT, Nagro AF, Fawzy AT, Bukhari OM. The effect of root canal treatment on oral health-related quality of life: Clinical trial. Saudi Endod J 2021;11:334.  Back to cited text no. 25
  [Full text]  
26.
Kamal EM, Nabih SM, Obeid RF, Abdelhameed MA. The reparative capacity of different bioactive dental materials for direct pulp capping. Dent Med Probl 2018;55:147-52.  Back to cited text no. 26
    
27.
Alghamdi F, Alsulaimani M. Regenerative endodontic treatment: A systematic review of successful clinical cases. Dent Med Probl 2021;58:555-67.  Back to cited text no. 27
    
28.
Al-Habib MA. Outcome of vital pulp therapy, revascularization, and apexification procedures: A retrospective study. Saudi Endod J 2022;12:210.  Back to cited text no. 28
  [Full text]  
29.
Asgary S, Fazlyab M, Sabbagh S, Eghbal MJ. Outcomes of different vital pulp therapy techniques on symptomatic permanent teeth: A case series. Iran Endod J 2014;9:295-300.  Back to cited text no. 29
    
30.
Ghoddusi J, Forghani M, Parisay I. New approaches in vital pulp therapy in permanent teeth. Iran Endod J 2014;9:15-22.  Back to cited text no. 30
    
31.
Bansal P, Kapur S, Ajwani P. Effect of mineral trioxide aggregate as a direct pulp capping agent in cariously exposed permanent teeth. Saudi Endod J 2014;4:135.  Back to cited text no. 31
  [Full text]  
32.
Reit C, Grondahl HG, Engstrom B. Endodontic treatment decisions: A study of the clinical decision-making process. Endod Dent Traumatol 1985;1:102-7.  Back to cited text no. 32
    
33.
Asgary S, Hassanizadeh R, Torabzadeh H, Eghbal MJ. Treatment outcomes of 4 vital pulp therapies in mature molars. J Endod 2018;44:529-35.  Back to cited text no. 33
    
34.
Asgary S, Eghbal MJ, Fazlyab M, Baghban AA, Ghoddusi J. Five-year results of vital pulp therapy in permanent molars with irreversible pulpitis: A non-inferiority multicenter randomized clinical trial. Clin Oral Investig 2015;19:335-41.  Back to cited text no. 34
    
35.
Asgary S, Eghbal MJ, Shahravan A, Saberi E, Baghban AA, Parhizkar A. Outcomes of root canal therapy or full pulpotomy using two endodontic biomaterials in mature permanent teeth: a randomized controlled trial. Clin Oral Investig 2022;26:3287-97.   Back to cited text no. 35
    
36.
Mainkar A, Kim SG. Diagnostic accuracy of 5 dental pulp tests: A systematic review and meta-analysis. J Endod 2018;44:694-702.  Back to cited text no. 36
    


    Figures

  [Figure 1]
 
 
    Tables

  [Table 1], [Table 2]



 

Top
 
 
  Search
 
Similar in PUBMED
   Search Pubmed for
   Search in Google Scholar for
 Related articles
Access Statistics
Email Alert *
Add to My List *
* Registration required (free)

 
  In this article
Abstract
Introduction
Materials and Me...
Results
Discussion
Conclusions
References
Article Figures
Article Tables

 Article Access Statistics
    Viewed262    
    Printed4    
    Emailed0    
    PDF Downloaded34    
    Comments [Add]    

Recommend this journal