AI-Assisted Shade Selection Versus Digital Spectrophotometry in Determining Maxillary Anterior Tooth Color in a Group of Egyptian Patients at Cairo University, Faculty of Dentistry Hospital (Diagnostic Accuracy Study)

Overview

Achieving an accurate shade match is a critical factor in the success of anterior esthetic restorations, directly influencing patient satisfaction, perceived treatment success, and long-term acceptance of restorations. Tooth color is a complex, multidimensional phenomenon influenced by hue, chroma, value, translucency and surface texture, and small discrepancies can be easily perceived in the esthetic zone. Traditionally, shade selection has been performed visually using commercial shade guides such as the VITA Classical or VITA 3D-Master systems. However, visual shade matching is inherently subjective and is significantly affected by examiner experience, training, surrounding environment, light source, observer fatigue, and metamerism. Several studies have shown that visual methods demonstrate only mild-to-moderate reliability and agreement, even among trained clinicians and students. To overcome these limitations, digital spectrophotometers were introduced to provide objective, reproducible, CIELAB-based color measurements of natural teeth and restorations. These devices analyze reflected light within a defined wavelength range and express the tooth shade within established systems such as VITA Classical A1-D4 and VITA 3D- Master. They have been widely used as an instrumental "gold standard" against which visual shade selection is evaluated, consistently demonstrating higher accuracy and better repeatability than conventional visual methods. More recently, artificial intelligence (AI) and machine learning (ML) approaches have been explored for dental shade matching. Deep learning models based on convolutional neural networks and other ML algorithms can analyze standardized intraoral photographs or smartphone images to automatically classify tooth shades according to VITA shade systems, often showing promising accuracy, precision and F1-scores, comparable to or sometimes exceeding experienced clinicians. In vitro studies have started to compare AI-based shade matching applications with spectrophotometers and image-based photometric analysis, suggesting that although spectrophotometers still tend to provide the most accurate color match, AI systems are rapidly improving and may offer clinically acceptable results with advantages in speed, usability, and integration into digital workflows. However, most of these investigations have been conducted using laboratory setups, artificial teeth, or non-Egyptian populations, and there remains a scarcity of in vivo diagnostic-accuracy studies validating AI shade selection systems against an accepted instrumental standard in real clinical settings

Study objectives and hypotheses: The aim of the study includes hypothesis and objectives Aim of the study This study aims to assess the accuracy of Artificial Intelligence-assisted shade selection in determining the shade of maxillary anterior teeth among a group of adult Egyptian patients attending Cairo University Dental Hospital, when compared to a digital spectrophotometer taken as the reference standard. Null Hypothesis No significant difference between AI-assisted shade selection and digital spectrophotometric measurements in determining the correct shade of maxillary anterior teeth in adult Egyptian patients attending Cairo University Dental Hospital. III. Methods: A) Trial design This diagnostic accuracy study assess the accuracy of Artificial Intelligence-assisted shade selection in determining the shade of maxillary anterior teeth among a group of adult Egyptian patients attending Cairo University Dental Hospital, when compared to a digital spectrophotometer taken as the reference standard. Participants: Eligibility criteria: Patient selection will be carried out by (B.B) after diagnosis at the Cairo University dental clinics. All patients falling under the inclusion criteria will be given a chance to be included until desired sample size is reached. 1. Eligibility criteria of participants: Inclusion criteria Exclusion criteria Age: 18-65 years. Male or female. Good oral hygiene Co-operative patients approving to participate in the trial. Have sufficient congnitive ability to understand consent procedure. Patients with orthodontic appliances, or bridge work that might interfere with evaluation. Systematic disease that may affect participation. 2. Eligibility criteria of teeth: Inclusion criteria Exclusion criteria Maxillary anterior teeth No signs of clinical mobility. Teeth with healthy periodontium. Non-vital tooth. Signs of pathological wear. Endodontically treated teeth. Severe periodontal affection or tooth indicated for extraction. Missing anterior teeth 10\. On what basis potentially eligible participants will be identified: The principal investigator will be doing a scranning analysis for shade selection of the patients addenting the outpatient clinc of the Conservative Dentistry Department . 11\. Study setting: i The study will be conducted in the outpatient clinic of Conservative Dentistry Department, Faculty of Dentistry, Cairo University, the operator in charge will be: B.B.. The researcher will bear ultimate responsibility for all activities associated with the conduct of a research project including recruitment of patients, explaining and performing the procedures to them. ii The anticipated and end dates are through the academic year (2026-2027). 12\. Whether participants formed a consecutive, random or convenience series: Patients will be consecutively recruited from the Diagnostic Department at Cairo University's Faculty of Dentistry. Test method: 13\. Index test \& reference standard in sufficient detail to allow replication The study will utilize digital spectrophotometry (VITA Easyshade®, Zahnfabrik H. Rauter GmbH \& Co. KG) as the reference standard, and an Artificial Intelligence-based shade selection system as the index test for determining the shade of maxillary anterior teeth in adult Egyptian patients. All enrolled participants will undergo both tests during the same visit to allow direct comparison between AI shade prediction and spectrophotometric measurement. The reference standard will be applied using a standardized clinical protocol because spectrophotometry is widely regarded as the instrumental "gold standard" for objective tooth color measurement in addition to objective color coordinates when available. Reference standard (digital spectrophotometer): Before each measurement session, the spectrophotometer will be calibrated according to the manufacturer's instructions. The tooth surface will be cleaned from any plaque or stain if present anterior teeth will be kept hydrated to avoid dehydration which may affect the shade matching accuracy, then the measurement area will be standardized which is typically the middle third of the labial surface of each maxillary central incisors selected for evaluation to minimize region-related color variation across cervical-incisal gradients. The probe tip will be positioned perpendicular to the tooth surface with stable contact and minimal angulation errors; 3 consecutive readings per tooth will be obtained, and the final reference shade will be recorded as the most frequent (mode). The device output will be documented in the VITA Classical shade system and stored as the reference value for analysis. Index test (AI-assisted shade selection): The AI system will determine tooth shade by analyzing standardized clinical photographs of the maxillary anterior teeth and predicting the shade according to the VITA Classical shade system. Image acquisition will be standardized to reduce variability from lighting and exposure including fixed illumination source, consistent distance and angulation, fixed camera settings/white balance, and consistent background/retraction). Each tooth will be photographed under the same conditions used for all participants to ensure repeatability. Images will then be uploaded to the AI system/software, and the predicted VITA shade will be recorded as the index-test result for each assessed tooth. Blinding and consistency measures: To reduce bias, the operator performing the AI image capture and AI shade prediction should not have access to the spectrophotometer results at the time of AI assessment, and the operator performing spectrophotometry should not see the AI output until reference measurements are completed. All operators will receive training on the standardized acquisition protocol to maintain consistency across patients. The same clinical setting and standardized acquisition steps will be applied for both methods throughout the study to support reliable comparison. 14\. Rationale for choosing the reference standard: Digital spectrophotometry is widely regarded as the instrumental gold standard for objective tooth color measurement. Devices such as the VITA Easyshade use spectral filters to analyze the re-emission spectrum within the visible range (approximately 400-700 nm) and output corresponding shades in established systems like VITA Classical A1-D4 and VITA 3D-Master, as well as CIELAB coordinates. Multiple clinical and in vitro studies have demonstrated that spectrophotometric methods show higher accuracy and better reproducibility than conventional visual shade selection, and they are frequently used as the benchmark against which new shade-selection methods, including digital 6photography, intraoral scanning, and visual assessments, are evaluated. Therefore, in this study, a digital spectrophotometer such as VITA Easyshade will be used as the reference standard against which the diagnostic performance of AI-assisted shade selection will be compared in determining the shade of maxillary anterior teeth in adult Egyptian patients the target condition is "accurate shade match of maxillary anterior teeth" 15\. Test positivity will be defined as agreement (match) between the AI-predicted shade (index test) and the spectrophotometer-measured shade (reference standard) within the VITA Classical shade system (A1-D4). Prespecified cut-off (primary threshold): Exact match. Index test positive: AI output = the same VITA Classical shade tab as the reference standard for the same tooth. Reference standard positive: VITA Easyshade shade designation (VITA Classical A1-D4) recorded as the benchmark value. Rationale: Both the AI system and VITA Easyshade are planned to output shades using the same established clinical shade system (VITA Classical A1-D4), making exact categorical agreement a clear, clinically interpretable primary threshold for defining "correct shade determination" in the protocol population. Exploratory thresholds (secondary analyses): No specific accepted clinical tolerance threshold for "near-match" versus "mismatch" for this setting, the study will also explore accuracy for additional thresholds after data collection, such as adjacent shade agreement (±1 shade step within VITA Classical ordering). 16a. The operator performing the index test will be blinded to the VITA Easyshade results at the time of image capture and AI shade prediction to avoid bias in the AI reading/recording. Masking is applied to prevent influence of prior results on recording the AI output. The index-test performer will have access only to routine visible clinical information needed to capture standardized images including the tooth to be assessed, retraction and moisture control, but will not have access to any prior shade measurements or the reference standard output during the index test procedure. 16b. The operator performing the reference standard will be blinded to the AI output until all spectrophotometric measurements are completed and recorded for each tooth, to prevent expectation bias. The reference-standard assessor will have access only to the minimal clinical information required to perform spectrophotometric measurement including tooth identification and positioning of the device, but will not be informed about the AI result or any prior shade outcome during the reference measurement session. D) Statistical analysis 17. Statistical analysis: Statistical analysis will be performed using Medcalc software, version 22 for windows (MedCalc Software Ltd, Ostend, Belgium). Continuous data will be presented as mean and standard deviation (SD), while categorical data will be presented as frequencies (n) and percentages (%). Diagnostic accuracy of index tests will be determined, including sensitivity, specificity, overall accuracy, positive and negative predictive values and ROC curve analysis. The significance level was set at P ≤ 0.05. Inter-observer agreement between assessors will be evaluated using Cohen's Kappa test. Chi-square test will be used to test association between index tests and reference test. Spearman's correlation will be used to measure the level of correlation between the index and the reference tests. 18\. How indeterminate index test or reference standard results will be handled: Data analysis will be performed using intention to diagnose aproach to prevent withdrawal bias 19\. How missing data on the index test and reference standard will be handled As there is no need for follow-up in this study, we can ensure that all data collection occurs at a single point in time and so eliminating the risk of missing data due to participant dropout or failure to attend future sessions. 20\. Sample size calculation According to the results of Ebeid et al. 2022 in which accuracy of Vita Easyshade for shade matching was 77.5%. Based on the equation (〖Z\_(1-α/2)〗\^2 P〖(1-P)〗\^ )/d\^2 by Charan et al in 2013, margin of error of 5% (0.05) and by adopting an alpha (α) level of 0.05 (5%), power=80%. The predicted sample size (n) was a total of 268 patients. Sample size was calculated using statistics kingdom sample size calculator (https://www.statskingdom.com/50\_ci\_sample\_size.html IV. Ethics and dissemination 21. Research ethics approval Application forms for Observational studies, checklist, and informed consent of Research Ethics Committee (REC) Faculty of Oral and Dental Medicine, Cairo University will be retrieved and filled, then will be delivered to REC committee for approval, this is done to prevent any ethical problems during the study or any harm for any of the participants. 22\. Protocol amendments If a new protocol is used, a protocol amendment will be submitted containing a copy of the new protocol and a brief explanation about the differences between it and the previous protocols. If there is a change in the existing protocol that affects the safety of subjects, investigation scope, or scientific quality of the trial an amendment containing a brief explanation about the change must be submitted. If a new author is added to accomplish the study an amendment including the investigator's data and qualifications to conduct the investigation will be submitted to prevent ghost authorship. 23\. Informed consent Researchers will discuss the trial with the participants. They will be able to discuss his thoughts with the researcher after a full explanation of the procedure in simple words. Participants who meet the inclusion criteria will sign informed consent. 24\. Confidentiality Name and personal data of the participants will not appear on the protocol form and will be maintained secured after the trial. This is done for the protection of participants' privacy and civil rights. 25\. Declaration of interest Financial and other competing interests for principal investigators for the overall trial and each study site 26. Access to data The principal investigator and the supervisors will be given access to the data sets. All data sets will be protected by passwords. To ensure confidentiality, data will be blinded of any identifying participant information. 27\. Dissemination policy The study results will be accessible through the department's database, with copies of the thesis available in the department library and on the Egyptian Universities Libraries Consortium's official site. The final report will be submitted for publication in international journals, with additional dissemination through conferences, seminars, and expert meetings in Egypt and internationally.