Histogram‐Corrected Intraoral Photographs for Early Gingivitis Screening in Orthodontic Patients

ABSTRACT

Objectives

Gingivitis has a significant impact on periodontal health, both during and after orthodontic treatment. Early detection and management are essential for minimizing tissue damage. This study investigated an early gingivitis detection method using intraoral photographs (IOPs) taken before, at the midpoint (MO), and three‐quarters of the way through orthodontic treatment (TO), as well as immediately after debonding (IDO).

Methods

To accurately measure gingival redness, the red/green (R/G) ratio and grayscale visible‐light value were calculated from IOPs using two methods: histogram‐based and grey value‐based intensity correction (HIC and GIC). The correlations between the gingival index (GI) and these metrics were then analysed.

Results

The GI increased at TO and IDO, with significant variation across gingival locations (p < 0.05). HIC analysis revealed increased R/G ratios in the papillary gingivae at TO and IDO (p < 0.01). GIC analysis showed higher visible‐light values at MO, which decreased at TO and IDO (p < 0.01). The HIC R/G ratio correlated positively with the GI at IDO in both the maxillary (r = 0.581, p < 0.001) and mandibular (r = 0.556, p < 0.001) gingivae. Conversely, GIC values were negatively correlated with the GI at IDO for the maxillary (r = −0.505, p < 0.05) and mandibular (r = −0.407, p < 0.05) gingivae.

Conclusions

This study demonstrates the potential of using HIC IOPs as an early screening tool for gingivitis. By applying image‐correction techniques, this novel method enhances diagnostic precision and paves the way for personalized oral disease management through early intervention and tailored patient care.

1. Introduction

Orthodontic treatment is a widely sought‐after medical intervention aimed at improving dental aesthetics and masticatory function [1, 2]. However, long‐term orthodontic procedures can introduce significant oral health challenges, including gingivitis, dental caries, and tooth root resorption [3]. The complexity of oral hygiene management increases substantially during treatment, particularly due to crowded teeth and the presence of orthodontic devices that impede effective brushing [4, 5].

Gingival inflammation represents a critical concern in orthodontic care with the potential to progress from a reversible condition to irreversible periodontal disease if not promptly detected and managed [6, 7, 8, 9]. Progression can lead to alveolar bone resorption, gingival recession, and the formation of unsightly black triangles between teeth, which are common complaints among orthodontic patients [1, 10]. Consequently, developing reliable and objective tools for early gingivitis detection has become paramount for both patients and orthodontic professionals.

Current diagnostic methods, particularly the Gingival Index (GI), have significant limitations. Traditionally, gingival inflammation has been assessed through indicators such as redness, swelling, and bleeding. However, these methods are inherently operator‐dependent and lack complete objectivity [11, 12]. To address these shortcomings, Kim et al. proposed a method to analyse the gingiva using intraoral photographs (IOPs) [13]. IOPs are typically taken before, during, and upon completion of orthodontic treatment to monitor treatment progress and to facilitate effective patient consultation. Early gingivitis detection using IOPs can help prevent the progression of periodontal disease in a timely manner. However, variations in the photographic conditions—such as brightness, shadows, and flash intensity—can affect the colour tone, brightness, saturation, and contrast of photographs. Hence, more precise and standardized detection methods are needed, especially given the potential for environmental factors to compromise diagnostic accuracy.

The colour of an object in a photograph can be determined based on its light quality [14]. Even when photographing the same patient, image quality can vary. Colorimeters and spectrophotometers have been proposed as tools to enhance colour‐matching precision [15, 16, 17]. Although these devices offer improved accuracy in shade matching and high intra‐device reproducibility (> 96%), their inter‐device consistency is reportedly relatively low (67%–93%) [15, 18]. In contrast, digital photography demonstrates high performance in terms of reproducibility, and digital photographs can be colour‐corrected using the recorded image data [19]. Based on colour matching, the difference between the actual gingival colour and that represented in the photograph may decrease, and the re‐photographing ratio may be reduced [20]. Various algorithm‐based colour‐matching methods have been introduced. The lookup table (LUT) method, based on predefined data, enables fast operation processing and can be extended to multidimensional applications [21]. Specifically, three‐dimensional (3D) LUTs map colour values along each axis of the red, green, blue (RGB) colour space, facilitating a wide range of precise colour transformations. These LUT‐based colour matching techniques are used in various applications, including image processing, video editing, and medical imaging. Furthermore, 3D LUTs use a 3D grid structure to map the input colour values to the output colour values, which enables high‐quality colour correction [22]. These LUTs can be applied in real time by leveraging hardware accelerators such as graphic processing units. However, their reliance on predefined data limits their ability to handle unexpected colour variations [21]. Intensity histogram‐based colour matching is a well‐established method that has demonstrated higher accuracy of object identification than that of grayscale‐based matching [23]. In particular, cumulative histogram‐based colour matching has shown strong performance against lighting variations [24, 25], preserving chromatic details and enabling accurate and consistent colour correction. Furthermore, this approach has demonstrated sufficient reproducibility for images captured with ring‐flash digital cameras, which are commonly used in real‐world oral photography [26]. Given these advantages, histogram‐based colour matching was selected as the core image processing approach in this study.

Therefore, there is a need for a simpler and reliable tool to detect early‐stage gingival inflammation caused by poor oral hygiene. This study aimed to determine the effectiveness of histogram‐based (HIC) and grey value‐based (GIC) intensity correction as tools for processing IOPs to screen for early gingivitis in orthodontic patients.

2. Study Population and Methodology

2.1. Participants

The necessary sample size was determined using power analysis, employing a medium effect size (f = 0.25; α = 0.05; 1‐β = 0.80) using G*Power software (3.1.9.2, Universität Düsseldorf, Germany) [27]. To recruit participants, we contacted dental clinics specializing in orthodontic treatment in Cheongju City via email, informing them of the purpose and methodology of the study and requesting their participation. The recruitment and analysis period was between September and December 2023.

A dental hygienist collected all IOPs. Gingivitis was considered the primary disease in this study. The inclusion criteria were as follows: (1) completion of orthodontic treatment when aged 10–39 years; (2) orthodontic treatment received for ≥ 6 months; (3) treatment received with fixed orthodontic devices; and (4) diagnosis of gingivitis by a dentist during orthodontic treatment.

Individuals were excluded if they: (1) were ≥ 40 years old; (2) had a systemic disease, such as diabetes or hypertension, within the last 6 months and were receiving continuous treatment; (3) had another disease that caused inflamed gingivae, including oral cancer; (4) were using antibiotics for disease treatment; (5) had dental caries on the labial smooth surface of the anterior teeth; (6) had undergone tooth extraction during orthodontic treatment; (7) had noticeable tooth discoloration; (8) had severe melanin pigmentation of the gingiva or gingivae that were dark red without inflammation; or (9) were aged < 40 years with advanced periodontitis or abnormal anatomical structures due to periodontal disease, since advanced periodontitis can cause alveolar bone loss and gingival recession. Gingival size was not considered due to the potential of gingival deformation based on the participant's oral condition. No participants in this study had caries on smooth tooth surfaces.

A data organizer assigned numbers to the IOPs used in the study to prevent exposure of personal information and ensure patient anonymity. Since this was a retrospective study and all data were anonymized, the requirement for informed consent was waived. This study also received an institutional review board approval exemption from the Bioethics Review Committee of OO University (1041107‐202212‐HR‐053‐01).

2.2. Target Gingivae

In total, 816 target gingival samples were included in this study. The selected sites were six areas of the papillary gingivae in the maxillary and mandibular incisors (Fédération Dentaire Internationale dental numbering system: 12, 11, 21, 22, 42, 41, 31, and 32) from 34 participants. Four IOPs were recorded for each patient: before orthodontic treatment (BO), at the midpoint of treatment (MO), three‐quarters of the way through treatment (TO), and immediately after debonding of the orthodontic appliances (IDO) (Figure 1). Figure 2 shows a representative example of a photograph processed using HIC and GIC to improve the quality of the image analysis.

FIGURE 1.

Intraoral photographs (IOPs) showing the target gingivae. The black boxes indicate the papillary gingivae, which were selected to measure gingival redness (A) before, (B) at the midpoint, and (C) three‐quarters of the way through orthodontic treatment, as well as (D) immediately after the debonding of orthodontic appliances.

FIGURE 2.

Improved intraoral photographs following image processing. HIC, histogram‐based intensity correction; GIC, grey value‐based intensity correction.

2.3. Proposed Redness Calculation Framework Using Histogram and Grayscale Visible‐Light Analysis

Figure 3 presents a simplified framework for calculating redness using HIC and GIC in the original IOP. The numbering indicates the order used in the algorithm. Briefly, an image of the papillary gingival area between the maxillary and mandibular incisors was extracted ($C$), which consisted of three channels: red, green, and blue (RGB) (①). The flash used during photography can reduce the accuracy of the measurements due to halo artefacts, which causes errors when calculating the absolute intensity within the RGB channel. By applying the gamma correction [27] and Otsu method [28], which have been used in previous studies [13], the halo‐artefact area was extracted ($L$) (②). The result of its removal was then derived [$C_{\mathit{new}}=C\circ \left(1-L\right)$] (③). Here, the constant values of multiples and indices in the original image were empirically set to 1.0 and 2.2, respectively, with ∘ representing an element‐wise multiplication operator. The mean value of each RGB channel for the derived $C_{\mathit{new}}$ image was calculated using Equation (1) (④):

$${\text{Mean}}_{R,G,B}=\frac{\sum _{i=1}^{\mathit{max}}\sum _{j=1}^{\mathit{max}}{I}_{R,G,B}\left(i,j\right)}{\sum _{i=1}^{\mathit{max}}\sum _{j=1}^{\mathit{max}}\left(1-L\left(i,j\right)\right)}$$ (1)

where $i$ and $j$ represent the positions in the coordinate system.

FIGURE 3.

Simplified scheme of the proposed redness calculation framework with mean red, green, and blue (RGB) values using histogram‐based and grey value‐based intensity corrections on an original intraoral photograph.

The histogram of the input image was analyzed in parallel (⑤). The mean value of the histogram was calculated using Equation (2) (⑥):

$$\begin{array}{c}{\text{Mean}}_{\mathit{\text{histo}}}=\frac{\sum _{m=1}^{\mathit{max}}\text{centerBin}\left(m\right)\circ \text{count}\left(m\right)}{\sum _{m=1}^{\mathit{max}}\text{count}\left(m\right)},\hfill \\ \text{centerBin}=\frac{\mathrm{Bin}\left(1:\mathrm{end}-1\right)+\mathrm{Bin}\left(2:\mathrm{end}\right)}{2}\hfill \end{array}$$ (2)

where $\text{centerBin}$ denotes the center intensity value between the bins of the histogram consisting of the mth bin.

In this study, the acquired RGB images were saved as JPG files, and the 8‐bit images supported 256 colours. Therefore, the number of bins was set to 256 while performing each histogram smoothing, which has been applied in previous studies [28, 29]. The $\mathit{\text{count}}$ value indicates the number of corresponding $\mathit{\text{centerBin}}$ values. After determining the mean value of the histogram for each image, the weight was calculated using Equation (3) by dividing the mean by the average value of the histogram of the entire image (⑦):

$$W=\frac{{\text{Mean}}_{\mathit{\text{histo}}}}{\sum _{n=1}^{\mathit{max}}{\text{Mean}}_{\mathit{\text{histo}}}\left(n\right)}$$ (3)

where $n$ is the total number of IOPs.

Finally, the HIC values (${\mathit{Cal}}_{R,G,B}$) were calculated by multiplying the mean RGB by the weight ($W$). The GIC values (${\mathit{Cal}}_{\mathit{\text{Gray}}}$) were generated by converting the image to grayscale (by assigning a conversion weight to each channel) (⑧). The GIC value was calculated as $0.299\times {\mathit{\text{Mean}}}_R+0.587\times {\mathit{\text{Mean}}}_G+0.114\times {\mathit{\text{Mean}}}_B$. The conversion method was implemented using the MATLAB software (R2023a, MathWorks, Natick, MA, USA) and the built‐in rgb2gray function, which converts a three‐channel image into a single‐channel image. The R/G ratio was calculated using Equation (4):

$$R/G=\frac{{\mathit{Cal}}_R}{{\mathit{Cal}}_G}.$$ (4)

If ${\mathit{Cal}}_G$ was zero, it was replaced with a value of 2⁻⁵², which was calculated using eps, a built‐in function of MATLAB.

2.4. Modified GI

One researcher (H‐N.K.) and a dental hygienist performed the GI analysis. The degree of gingivitis was confirmed by visually inspecting images on the same computer. The assessment criteria for gingivitis were determined based on clinical appearance, including gingival colour, shape, and interdental papillae, without probing. These criteria followed the 2017 classification system for periodontal disease diagnosis [30].

Duplicate analysis of 10% of all photographs confirmed > 95% concordance between results. The degree of gingivitis was scored between 0 and 4 using a modified GI developed by Tobias et al. [29] If the gingiva was pale pink, filled the interdental space sharply, showed no swelling or signs of bleeding, and exhibited stippling, the gingivitis score was recorded as 0. A gingivitis score of 4 was recorded when the papillary gingiva showed severe inflammatory signs, such as intense erythema, pronounced swelling, absence of stippling, and spontaneous bleeding. The GI was determined using the same papillary gingival area as that in the IOPs, corrected using HIC and GIC.

2.5. Statistical Analyses

All data were analysed using SPSS (version 24.0; IBM Corp., Armonk, NY, USA). Descriptive statistical analyses were performed on GI, HIC, and GIC values at four time points during orthodontic treatment. Specifically, the mean and standard deviation of each GI, HIC, and GIC value for each of the six target areas are presented. Additionally, data were treated as repeated measures within participants. The Shapiro–Wilk test indicated a non‐normal distribution, and nonparametric tests were used. The Friedman test was used to assess differences in GI, HIC, and GIC values across the four time points for each tooth region. Given the nonparametric nature of the data, the Wilcoxon signed‐rank test was used for post hoc pairwise comparisons at each time point. To control for the risk of type I error due to multiple comparisons, Bonferroni correction was applied for post hoc analysis (α_adj = 0.05/6 = 0.0083). Kendall's W coefficient of concordance was calculated to evaluate the consistency of changes across time points. Statistical significance was set at p < 0.05. Finally, Pearson's correlation tests were performed to explore the relationships between the modified GI and high HIC and GIC values.

3. Results

3.1. General Characteristics

In total, 45 individuals were recruited for the study; however, only 34 participants were ultimately included. One participant was excluded because they exceeded the study age limit (≥ 40 years old), whereas 10 participants were excluded due to difficulties in determining the gingiva colour in the marginal gingiva area owing to light reflection or poor IOP clarity. A total of 136 IOPs, including 816 target gingivae, were collected from 34 patients (16 male and 18 female) aged between 18 and 39 years who underwent orthodontic treatment. Among the participants, 47.1% and 41.1% were in their 20s and 30s, respectively. The mean duration of orthodontic treatment was 20.4 months. ceramic orthodontic devices were used in most participants (76.5%) (Table 1).

TABLE 1.

General characteristics of the participants.

Characteristic	Item	N (%)
Sex	Male	16 (47.1)
	Female	18 (52.9)
Age (years)	10s	4 (11.8)
	20s	16 (47.1)
	30s	14 (41.1)
Orthodontic device	Metal bracket	8 (23.5)
	Ceramic bracket	26 (76.5)
Average period of orthodontic treatment (mo, mean ± standard deviation)	20.4 ± 14.0

3.2. GI Changes

Table 2 presents the changes in the modified GI scores over the course of orthodontic treatment. Although the degree of change varied across the six gingival locations, GI scores were higher at TO and IDO and lower at BO and MO (p < 0.05).

TABLE 2.

Gingival index (GI) of papillary gingivitis in intraoral photographs.

Orthodontic treatment	Tooth region (maxilla)			Tooth region (mandible)
	#12–#11	#11–#21	#21–#22	#42–#41	#41–#31	#31–#32
BO	0.79 ± 0.70a	0.57 ± 0.65a	0.50 ± 0.65a	0.57 ± 0.85a	0.63 ± 0.91a	0.54 ± 0.94a
MO	0.86 ± 0.89a	0.61 ± 0.51a	0.64 ± 0.74a	0.93 ± 0.83a	1.00 ± 0.88 a	0.93 ± 0.83a
TO	1.50 ± 1.09a	1.50 ± 0.85a b	1.56 ± 0.91a	1.79 ± 1.12a	1.71 ± 0.91b	1.93 ± 1.07b
IDO	1.79 ± 0.97b	1.79 ± 1.05b	1.64 ± 0.63b	1.86 ± 0.86b	2.07 ± 0.91bc	1.85 ± 0.89b
p *	0.003	0.001	0.001	0.001	< 0.001	0.001

Note: Values are presented as mean ± standard deviation of the GI.The P‐valuse's significanc level was 0.05, superscript alphanets's significant levels were described in foot notes, Friedman test, Kendall's W test, Post hoc pairwise comparisons using the Wilcoxon signed‐rank test with Bonferroni correction (α_adj = 0.05/6 = 0.0083) revealed significant increases in GI between BO and TO, BO and IDO, and MO and IDO.

Abbreviations: BO, before orthodontic treatment; IDO, immediately after the debonding of orthodontic appliances; MO, at the midpoint of orthodontic treatment; TO, three‐quarters of the way through orthodontic treatment.

^*Friedman test, Kendall's W test, Post hoc pairwise comparisons using the Wilcoxon signed‐rank test with Bonferroni correction (α_adj = 0.05/6 = 0.0083) revealed significant increases in GI between BO and TO, BO and IDO, and MO and IDO.

3.3. R/G Ratio Obtained Using HIC

Table 3 shows the R/G ratios of the papillary gingivae obtained using HIC from the IOPs at the four treatment time points. The papillary gingivae in both the maxilla and mandible showed increased R/G ratios at TO and IDO (p < 0.01).

TABLE 3.

Red/green ratios of papillary gingivae in histogram‐based intensity‐corrected intra‐oral photographs.

Orthodontic treatment	Tooth region (maxilla)			Tooth region (mandible)
	#12–#11	#11–#21	#21–#22	#42–#41	#41–#31	#31–#32
BO	1.61 ± 0.30a	1.49 ± 0.27a	1.79 ± 0.42 a	1.57 ± 0.28 a	1.51 ± 0.28a	1.55 ± 0.31 a
MO	1.57 ± 0.33a	1.52 ± 0.30a	1.74 ± 0.35 a	1.59 ± 0.36 a	1.53 ± 0.32 a	1.57 ± 0.34 ab
TO	1.75 ± 0.35b	1.65 ± 0.31ab	1.90 ± 0.41a	1.75 ± 0.35 a b	1.69 ± 0.34 a b	1.75 ± 0.35 bc
IDO	1.97 ± 0.32bc	1.86 ± 0.32b	2.22 ± 0.38b	2.02 ± 0.40b	1.92 ± 0.34 b	1.96 ± 0.33c
p *	< 0.001	0.001	0.001	0.001	< 0.001	< 0.001

Note: Values are presented as mean ± standard deviation of the histogram‐based intensity correction. The P‐valuse's significanc level was 0.05, superscript alphanets's significant levels were described in foot notes, Friedman test, Kendall's W test, Post hoc pairwise comparisons using the Wilcoxon signed‐rank test with Bonferroni correction (α_adj = 0.05/6 = 0.0083) revealed significant increases in GI between BO and TO, BO and IDO, and MO and IDO.

Abbreviations: BO, before orthodontic treatment; IDO, immediately after the debonding of orthodontic appliances; MO, at the midpoint of orthodontic treatment; TO, three‐quarters of the way through orthodontic treatment.

^*Friedman test, Kendall's W test, Post hoc pairwise comparisons using the Wilcoxon signed‐rank test with Bonferroni correction (α_adj = 0.05/6 = 0.0083) revealed significant increases in R/G ratio of histogram‐based intensity between BO and TO, BO and IDO, and MO and IDO.

3.4. Grayscale Visible‐Light Values Obtained Using GIC

Table 4 outlines the grayscale visible‐light values obtained using GIC from the IOPs at the four treatment time points. These values were higher at MO and decreased at TO and IDO across all six papillary gingivae (p < 0.01).

TABLE 4.

Grayscale visible‐light values of papillary gingivae in grey value‐based intensity‐corrected intraoral photographs.

Orthodontic treatment	Tooth region (maxilla)			Tooth region (mandible)
	#12–#11	#11–#21	#21–#22	#42–#41	#41–#31	#31–#32
BO	149.62 ± 19.03 a	164.10 ± 20.96 a	151.89 ± 21.05 a	154.57 ± 20.14 a	161.95 ± 20.87 a	156.29 ± 22.01 a
MO	156.84 ± 21.70 a	166.46 ± 20.58 a	156.98 ± 21.21 a	157.04 ± 21.93 a	162.35 ± 22.35 a	158.06 ± 22.44 a
TO	148.55 ± 20.32 a	159.55 ± 19.26 a	151.72 ± 20.64 a	148.49 ± 18.33 ab	153.24 ± 19.96 ab	147.08 ± 19.14 ab
IDO	132.60 ± 17.46b	145.80 ± 16.10 b	136.06 ± 15.48 b	133.62 ± 19.48b	140.72 ± 17.18 b	135.76 ± 16.41 b
p *	0.003	0.001	0.001	0.001	< 0.001	< 0.001

Note: Values are presented as mean ± standard deviation of the grey value‐based intensity correction. The P‐valuse's significanc level was 0.05, superscript alphanets's significant levels were described in foot notes, Friedman test, Kendall's W test, Post hoc pairwise comparisons using the Wilcoxon signed‐rank test with Bonferroni correction (α_adj = 0.05/6 = 0.0083) revealed significant increases in GI between BO and TO, BO and IDO, and MO and IDO.

Abbreviations: BO, before orthodontic treatment; IDO, immediately after the debonding of orthodontic appliances; MO, at the midpoint of orthodontic treatment; TO, three‐quarters of the way through orthodontic treatment.

^*Friedman test, Kendall's W test, Post hoc pairwise comparisons using the Wilcoxon signed‐rank test with Bonferroni correction (α_adj = 0.05/6 = 0.0083) revealed significant increases in Grayscale visible‐light values between BO and TO, BO and IDO, and MO and IDO.

3.5. Correlation Between the GI and HIC R/G Ratio

Table 5 displays the correlations between the GI and R/G ratio from HIC‐processed IOPs. The correlation coefficient between GI_IDO and HIC_IDO was 0.581 (p < 0.001) for the maxillary gingivae and 0.556 (p < 0.001) for the mandibular gingivae, indicating a strong positive correlation.

TABLE 5.

Correlation analysis of the highest GI and HIC red/green ratio for each maxillary and mandibular gingiva.

Location	Orthodontic treatment	Maxilla				Mandible
		HIC _BO	HIC _MO	HIC _TO	HIC _IDO	HIC _BO	HIC _MO	HIC _TO	HIC _IDO
Maxilla	GI_BO	0.135
	GI_MO		0.183
	GI_TO			0.197
	GI_IDO				0.581*
Mandible	GI_BO					0.243
	GI_MO						0.095
	GI_TO							0.134
	GI_IDO								0.556*

Note: Pearson's correlation test using each of the highest values among the three papillary gingivae in the maxilla and mandible.

Abbreviations: BO, before orthodontic treatment; GI, gingival index; HIC, histogram‐based intensity correction; IDO, immediately after the debonding of orthodontic appliances; MO, at the midpoint of orthodontic treatment; TO, three‐quarters of the way through orthodontic treatment.

^* p < 0.001.

3.6. Correlation Between the GI and GIC Grayscale Visible‐Light Values

Table 6 shows the correlations between the GI and the grayscale visible‐light values from the GIC‐processed IOPs. The correlation coefficient between GI_IDO and GIC_IDO was −0.505 (p < 0.05) for the maxillary gingivae and −0.407 (p < 0.05) for the mandibular gingivae.

TABLE 6.

Correlation analysis of the highest GI and GIC Grayscale visible‐light values for each maxillary and mandibular gingiva.

Location	Orthodontic treatment	Maxilla				Mandible
		GIC _BO	GIC _MO	GIC _TO	GIC _IDO	GIC _BO	GIC _MO	GIC _TO	GIC _IDO
Maxilla	GI_BO	−0.331
	GI_MO		−0.011
	GI_TO			−0.306
	GI_IDO				−0.505*
Mandible	GI_BO					−0.252
	GI_MO						0.002
	GI_TO							0.30
	GI_IDO								−0.407*

Note: Pearson's correlation test using each of the highest values among the three papillary gingivae in the maxilla and mandible.

Abbreviations: BO, before orthodontic treatment; GIC, grey‐based intensity correction; IDO, immediately after the debonding of orthodontic appliances; MO, at the midpoint of orthodontic treatment; TO, three‐quarters of the way through orthodontic treatment.

^* p < 0.05.

4. Discussion

Orthodontic treatment requires the placement of various devices in the mouths of patients. Fixed orthodontic devices, such as bands or brackets, are attached to the teeth, whereas additional components, including wires, elastics, and coil springs, are connected to them to facilitate teeth movement [31]. Most candidates for orthodontic treatment do not have straight teeth, which hinders effective dental plaque removal compared to that in individuals with properly aligned teeth. In addition, these complex orthodontic devices promote the accumulation of dental plaque around the device and complicate oral hygiene management, such as brushing and flossing [32]. Moreover, the primary targets of orthodontic treatment are children, who tend to neglect oral hygiene. Consequently, orthodontic treatment may have a significant negative impact on the periodontal tissue [33].

Orthodontic tooth movement is the result of periodontal tissue remodelling and is significantly affected by periodontal conditions before, during, and after treatment [7, 34, 35, 36]. However, orthodontic treatment can cause complications, such as gingivitis, gingival hyperplasia, and gingival recession, due to poor oral hygiene [1]. In this study, GI was evaluated using IOPs at four time points: BO, MO, TO, and IDO. Our findings revealed that GI was higher in the TO and IDO groups, indicating that gingivitis increased as the treatment progressed. To develop an early gingivitis detection tool using IOPs, we processed photographs using two methods: HIC and GIC. To evaluate gingival redness in orthodontic patients, we determined the R/G ratios and grayscale visible‐light values of papillary gingivae in the processed IOPs. Similar to GI, the HIC R/G ratio increased at TO and IDO. Boke et al. [4] also reported that visible plaques, inflammation, and gingival recession significantly increased in patients with fixed orthodontic appliances. Our results, alongside those of previous studies, support the hypothesis that orthodontic treatment contributes to increased gingival inflammation and redness.

Upon analysing the correlations among the GI, R/G ratio, and grayscale visible‐light values calculated from the IOPs, we observed a strong correlation between the R/G ratio and the GI at treatment completion and device removal (IDO), particularly when gingivitis and redness progressed due to poor oral hygiene associated with the attachment of orthodontic appliances. In contrast, no correlation was observed between the R/G ratios of the HIC and GI before or during the early stages of orthodontic treatment. Moreover, in the IOPs processed using GIC, which is one of the methods proposed to make gingivitis screening more efficient, a negative correlation between grayscale visible‐light values and GI at IDO was identified. We assumed that the red colour in these images was processed as a dark shade, specifically dark grey.

HIC technology is widely used in diagnostic medical imaging and has been successful in overcoming various environmental factors, including flashes and shadows [37, 38]. HIC is an appropriate method for simple and accurate IOP analysis that could potentially replace traditional GI evaluation, which is the main hypothesis of this study. Although there are limitations in diagnosing gingivitis or periodontitis in orthodontic patients using IOP alone, we expect that HIC technology will provide improved diagnostic capabilities. Indeed, we demonstrated that the HIC has a stronger correlation with GI values than the GIC. Consequently, we believe that HIC should be actively used when the GI is difficult to determine or as an approach for analysing periodontal diseases using imaging‐based methods.

These results suggest that HIC‐processed IOP analysis is a meaningful approach for gingivitis screening in orthodontic patients, and that the R/G ratio can be effectively diagnosed periodontal disease. Similar to Mayer et al., who reported the potential benefits of photometric analysis for assessing gingival changes after treatment [39], we showed that gingivitis can be assessed and diagnosed using gingival colour changes in IOPs. Our study confirms that diagnosis efficiency is increased by applying photocorrection technology. However, some limitations must be considered. First, the results may vary depending on the number of bins used in the histogram analysis. Bins are used to allocate the number of bars required to express the entire range of variables represented in a histogram, and the image intensity expression varies according to the bin width. This variation can be quantified using histogram distance methods, including the chi‐squared [40], cross‐bin [41], and Earth mover's distance [42, 43]. Future studies should examine the impact of these various approaches on the results obtained using the proposed HIC method. Second, periodontal region extraction was not standardized, as these parameters were individually selected for each patient. We selected six periodontal areas, including three papillary gingival areas between the four maxillary and mandibular incisors. This selection was based on the empirical judgement of researchers. However, further studies are required to standardize area selection. Third, the results of the high‐intensity outer area remaining after halo artefact removal were incomplete. Although the Otsu method has proven effective for removing distorted areas [44, 45], it operates based on thresholds, and residues may remain in the outer area, which can distort the HIC results. In the present analysis, calculations were performed by excluding the top 1% of the intensity values. Recently, several algorithms that remove halo artefacts more accurately have been introduced [46, 47]. Such advancements are expected to further improve the results when applied to the proposed HIC method. Finally, this study had a relatively small sample size. Although power analysis using G*Power indicated that 45 participants were required to achieve sufficient statistical power, only 34 participants were included in the final analysis. This discrepancy may reduce the robustness of the statistical tests and increase the risk of type II errors, potentially limiting the generalizability of the findings. The small sample size was due to the time constraints and logistical challenges inherent in conducting clinical imaging studies during the review period. Future studies should include larger cohorts to validate these findings and further strengthen the clinical applicability of the proposed image‐based gingivitis screening methods.

5. Conclusion

This confirms the viability of the HIC method as a novel diagnostic tool for gingivitis screening during orthodontic treatment. We expect that this approach will provide valuable data for digital healthcare diagnosis and management platforms for oral health promotion in the future.

6. Clinical Relevance

6.1. Scientific Rationale for Study

HIC‐ and GIC‐processed IOPs may serve as diagnostic tools for gingivitis screening in orthodontic patients.

6.2. Principal Findings

We revealed a strong correlation between the R/G ratio and GI during the late stages and completion of orthodontic treatment, underscoring the efficacy of HIC‐processed IOP analysis for gingivitis screening and periodontal disease diagnosis.

6.3. Practical Implications

The HIC method offers a novel diagnostic tool for screening gingivitis during orthodontic treatment and can provide valuable data for digital healthcare platforms for oral health management.

Author Contributions

K.K. and H.‐N.K. conceived the ideas; H.‐N.K. and J.‐Y.K. collected the data; K.K. and H.‐N.K. analysed the data; and J.‐Y.K. and H.‐N.K. led writing – review and editing.

Funding

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

Data sharing is not applicable to this article as no new data were created or analyzed in this study.