The prevalence of proximal caries remains high, regardless of the presence of primary or permanent dentition. It has been reported that proximal initial caries lesions account for 86 % of the total number of proximal caries lesions at 15 years of age [1,2]. With the emergence of new preventive and minimally invasive treatment strategies, it is becoming increasingly important for clinicians to detect and accurately identify proximal caries in the early stages. This enables an effective reduction of the disease burden associated with dental caries. Bitewing radiography (BWR), combined with visual and tactile examinations, is an important method for the early detection of proximal caries [3,4]. However, research has shown that its diagnostic efficacy is often unsatisfactory for initial enamel lesions [5], [6], [7].

Compared with the BWR examination, cone-beam computed tomography (CBCT) has demonstrated non-inferiority or even superiority in detecting proximal caries [8,9]. As previously reported, photostimulated phosphor radiographs and CBCT showed similar performances in detecting proximal caries, and the diagnostic efficacy was found to be independent of CBCT resolution [10,11]. Compared with “ground truth” methods such as micro-CT and histology, CBCT has also demonstrated good sensitivity and specificity [12,13]. For clinical orthodontic applications, employing CBCT for root-bone relationship assessments offers considerable advantages in achieving precise orthodontic treatment to avoid potential periodontal damage, as the pretreatment status of alveolar boundary conditions and their potential adaptation may set the anatomic limitations of tooth movement [14], particularly in patients who are at risk of bone dehiscence and fenestration [15,16]. If orthodontists can simultaneously use CBCT data for caries screening, the risk-to-benefit ratio of CBCT imaging can be further enhanced. However, it is important to emphasize that CBCT is not recommended to be routinely used for diagnosing caries in clinical practice because the required X-ray radiation may pose harm to patients, making it ethically unacceptable for the long-term follow-up of patients and nonoptimal for pregnant women [17]. Hence, there is a growing demand for new imaging technologies that facilitate the early detection and regular follow-up of proximal caries.

As a nonionizing alternative, near-infrared imaging (NIRI) has proven to be a valuable diagnostic method for the detection of proximal caries [18,19]. NIRI has shown excellent diagnostic performance for proximal caries in primary and permanent molars compared to traditional methods such as visual and tactile examinations and BWR [18,19]. Mechanistically, NIRI can be divided into reflectance and transillumination imaging [20]. In reflectance imaging, near-infrared light passes through the entire thickness of the enamel owing to its lower scattering coefficient, resulting in a dark area, whereas dentin appears bright because of the scattering effect caused by the orientation of the dentinal tubules. Carious lesions, or areas of demineralization, manifest as bright areas in NIRI due to increased scattering within the region. In transillumination imaging, the opposite is observed: healthy enamel appears bright, whereas dentin and suspected carious lesions appear dark. The miniaturization of NIRI devices has further enhanced their clinical applicability. However, it is important to note that there is significant variability in the diagnostic performance of in vivo studies using NIRI for caries diagnosis [21], [22], [23].

This study aimed to compare the diagnostic performance of NIRI in detecting proximal caries with that of CBCT, which was used as the reference test. This study also investigated potential improvements in diagnostic performance when combining NIRI with unaided visual examination (UVE). The null hypothesis of the current study was that UVE, NIRI, and their combination do not differ from CBCT regarding the diagnostic performance in detecting proximal caries.