Implant‑position reproducibility of wired and wireless intraoral scanners and the effects of Wi‑Fi distance and upload speed: an in vitro study

Abstract

Background

The impact of wireless communication on intraoral scanning reproducibility remains unclear. This in‑vitro study aimed to compare the implant‑position reproducibility of wireless (Wi‑Fi) and wired intraoral scanners using a mandibular edentulous model with six implants and to evaluate how Wi‑Fi distance and upload speed affect reproducibility in wireless intraoral scanners.

Methods

An edentulous mandibular model with six implants was scanned as the reference dataset using a high-accuracy laboratory scanner. Optical impressions were obtained using three wireless intraoral scanners (designated as Primescan 2, SIRIOS, and TRIOS 5) and one wired intraoral scanner (Primescan). For wireless devices, the scanner–router distance was set at 0.5, 2.0, and 5.0 m. The wired Primescan served as the control. Five scans per condition were acquired and analyzed. Upload speed was measured immediately before scanning at each distance tested. The reference and intraoral scanner datasets were superimposed for three-dimensional analysis. Reproducibility was expressed as the concordance rate, i.e., the percentage of surface area within ± 50 μm of the reference dataset. The association between upload speed and reproducibility was examined. The median and interquartile range were calculated for concordance rates; group differences were tested using the Kruskal–Wallis test with Steel–Dwass post‑hoc comparisons, and within‑device associations between upload speed and concordance were examined by linear regression (two‑sided α = 0.05).

Results

Primescan 2 showed stable concordance across distances (> 78.8%; p > 0.05) and exceeded Primescan at all distances (63.5%; p < 0.05). SIRIOS (64.6→50.4%) and TRIOS 5 (61.6→29.5%) showed decline with distance; at 0.5 m, both were comparable to Primescan, whereas at 5.0 m, TRIOS 5 < SIRIOS (p < 0.05). Across devices at identical distances, Primescan 2 > SIRIOS/TRIOS 5 (p < 0.05). Upload speed correlated with concordance for SIRIOS (β = 0.0369, R² = 0.725, p < 0.001) and TRIOS 5 (β = 0.0707, R² = 0.567, p = 0.001), but not for Primescan 2 (β = 0.0177, R² = 0.077, p = 0.318).

Conclusions

Primescan 2 showed higher implant‑position reproducibility than wired Primescan at all tested distances and was insensitive to Wi‑Fi distance and upload speed. SIRIOS and TRIOS 5 matched the wired comparator at 0.5 m but showed decline at 2.0–5.0 m; upload speed correlated positively with reproducibility for SIRIOS and TRIOS 5, but not for Primescan 2, indicating dependence on device architecture and Wi‑Fi link quality.

Background

Intraoral scanners have become indispensable tools in modern dentistry, with applications ranging from diagnosis and treatment planning to the fabrication of prostheses and enhancement of patient engagement [1–4]. The accuracy of these devices is a critical factor for successful clinical outcomes. While previous studies have shown that intraoral scanners provide clinically acceptable accuracy for short-span restorations [5, 6], their performance in full-arch implant cases remains debatable. Some reports have suggested limitations in full-arch scanning due to accumulated errors over long distances [7], whereas others demonstrate an accuracy of intraoral scanners equivalent or superior to that of conventional methods even in such scenarios [8].

To date, most of these studies have focused on wired intraoral scanners. However, wireless intraoral scanners, which transfer data via wireless fidelity (Wi-Fi), have recently gained popularity due to their improved maneuverability and cleaner operatory environment. Although a few studies have compared wired and wireless scanners [9], the effects of specific wireless communication parameters, such as Wi-Fi distance or upload speed, on scanning reproducibility have not been thoroughly evaluated. Upload speed refers to the rate at which data are transmitted from the scanner to an external device or cloud storage. Unstable communication conditions—such as increased router-to-scanner distance, signal attenuation, or limited bandwidth—may cause delayed or incomplete data transmission by wireless intraoral scanners. These disruptions can impair the real-time acquisition and processing of scan images, thereby compromising the accuracy of three-dimensional (3D) reconstruction.

Accurate implant-position transfer is essential for the fabrication of well-fitting prostheses. Poor reproducibility may result in misfit of superstructures, which can increase the risk of mechanical complications, such as screw loosening, framework fracture, and even biological complications such as peri-implantitis and marginal bone loss. Therefore, ensuring optimal scanner performance—even under variable clinical Wi-Fi conditions—is critical to achieving long-term success in implant therapy.

Based on these considerations, we hypothesized that greater Wi-Fi communication distances and slower upload speeds would negatively affect the implant-position reproducibility of wireless intraoral scanners. Therefore, this in‑vitro study aimed to compare the implant‑position reproducibility of wireless (Wi‑Fi) and wired intraoral scanners using a mandibular edentulous model with six implants, and to evaluate how Wi‑Fi distance and upload speed affect reproducibility in wireless intraoral scanners.

Methods

Fabrication of the reference model and acquisition of the reference dataset

Six implants (Roxolid Tissue Level Standard Implant Ø4.1 mm RN-SLActive Loxim-10.0 mm, Straumann, Basel, Switzerland) were embedded in a mandibular edentulous gypsum model at positions corresponding to 47, 44, 42, 32, 34, and 37 (FDI tooth numbering system). Three rigid cylindrical reference bodies (diameter 3 mm, height 2 mm) with a matte surface finish were rigidly affixed to the alveolar ridge—buccal to #47 and #37 and at the mid-lingual area between #42 and #32—and were used solely for dataset alignment, completing the reference model (Fig. 1). Scan bodies (CARES Mono Scan body RN, Straumann, Basel, Switzerland) were attached to the implants, and the reference dataset was acquired using a high-precision dental laboratory scanner (F8, 3Shape, Copenhagen, Denmark, ISO 12836 accuracy: 4 μm). These scan bodies were cylindrical and made of polyetheretherketone (PEEK). This reference dataset served as the near‑ground‑truth reference standard against which 3D surface deviations of all intraoral‑scanner (IOS) datasets were evaluated.

Fig. 1.

Reference model used in this study

Optical impression method using intraoral scanners

Three wireless intraoral scanners—Primescan 2 (Dentsply Sirona, York, PA, USA), SIRIOS (Straumann, Basel, Switzerland), and TRIOS 5 (3Shape, Copenhagen, Denmark)—were evaluated, and one wired intraoral scanner, Primescan (Dentsply Sirona, York, PA, USA), was used as the control.

The alveolar ridge of the reference model was scanned first (Fig. 2a). Subsequently, scan bodies were attached to each implant and scanned (Fig. 2b). Alveolar-ridge and scan-body datasets were automatically matched using dedicated software (Primescan 2: Connect Software v5.2.10; SIRIOS: Virtuo Vivo Software v3.10; TRIOS 5: 3Shape Unite v25.1; and Primescan: CEREC Software v5.2.13). All the scans were performed by a single operator, a board-certified prosthodontist with 10 years of clinical experience (T.Mu., certified by the Japan Prosthodontic Society), to minimize operator-dependent variability. Scan time was standardized to 60 s per acquisition, encompassing the alveolar ridge and all six scan bodies (two-pass sequence as in Fig. 2), and was monitored with a digital stopwatch (resolution 0.01 s). For all wireless scanners, the battery charge was set to 100% at the start of every scan.

Fig. 2.

Scanning sequence for the reference model. a Scanning sequence for the alveolar ridge. b Scanning sequence for scan bodies

For the wireless intraoral scanners, the distance between the scanner and its Wi-Fi communication device was set to 0.5, 2.0, and 5.0 m. Primescan 2 used a Wi-Fi router (WXR18000BE10P, Buffalo Inc., Nagoya, Japan), whereas SIRIOS and TRIOS 5 were connected to the host computer via a USB Wi‑Fi client adapter (TP-Link Archer T9UH, TP-Link Technologies Co., Ltd., Shenzhen, China) attached to the host computer (Fig. 3). Wireless communication was performed on the 5‑GHz band. Channel width was not explicitly logged during the experiments; however, the WXR18000BE10P supports up to 160 MHz channels on 5 GHz, and the Archer T9UH supports up to 80 MHz channels on 5 GHz. Router/adapter configurations (e.g., SSID, security mode, and channel selection) were kept constant across distances, and only the scanner–Wi‑Fi distance varied. These Wi-Fi communication devices were used as recommended by the manufacturer. For brevity, we refer to this separation as the scanner–Wi-Fi distance throughout the manuscript. All obstacles within a 5.0 m radius of the scanner and Wi-Fi communication device were removed to minimize interference.

Fig. 3.

Intraoral scanners and Wi-Fi communication devices used in this study

The host computers were identical Dell Precision 7680 Mobile Workstations (Dell Inc., Round Rock, TX, USA) with a 13th Gen Intel^® Core™ i7-13850HX vPro^® processor (Intel Corporation, Santa Clara, CA, USA). The processing computers were connected via wired Gigabit Ethernet from the same outlet/switch port, with on-board Wi-Fi/Bluetooth disabled; thus, the only intentionally varied network factor was the scanner–Wi-Fi distance. Because wireless architectures differed across devices, absolute upload speeds were not compared between devices; instead, effective upload throughput was used as a within-device, device-agnostic index of link quality.

A ceiling-mounted LED dental light was positioned directly above the operator and centered over the model at a fixed height of 80 cm above the occlusal plane and an angle of 45°. Intensity was kept constant, and ambient daylight was excluded; where available, illuminance was approximately 1,000 lx, verified with a digital meter. Room temperature was maintained at 25 °C, and potential sources of electromagnetic interference were eliminated.

Sample size calculation

We acquired five repeated scans per scanner–distance condition (0.5, 2.0, 5.0 m) for each of three wireless devices and five scans for the wired control (total N = 50). An a priori G*Power analysis v3.1.9.7 (one‑way fixed‑effects ANOVA; α = 0.05; 1–β = 0.80; k = 3; n = 5 per level) indicated that within‑device distance comparisons are powered to detect only very large effects (Cohen’s f ≈ 0.91; partial η² ≈ 0.46); for f = 0.62, power is ~ 0.46. Power for Kruskal–Wallis tests was approximated using this ANOVA setup. We report effect sizes with two‑sided 95% confidence intervals [CIs], and for regressions, unstandardized slopes (β) with 95% CIs and R².

Measurement of upload speeds of the Wi-Fi communication device

Due to software limitations and manufacturer-specific restrictions, different methods were used to measure the upload speeds. For the WXR18000BE10P (used with Primescan 2), we used the Speedtest application by Ookla [10]. For the TP-Link Archer T9UH (used with SIRIOS and TRIOS 5), because the host computer and scanner were connected via a direct local link (i.e., not routed through the Internet), Speedtest was inapplicable; instead, we obtained the OS-reported Wi-Fi link rate (transmit rate, Tx rate). Additionally, the host computer also employed a wired Ethernet connection; when Speedtest was launched, the operating system’s default route prioritized the wired path, preventing evaluation of the wireless link. Furthermore, because the scanner is a medical device, installation of third-party applications or traffic-generation tools was not permitted, making it infeasible in this configuration to standardize measurements using the same application-layer method.

Network performance was assessed using Speedtest by the Ookla app to measure application‑layer throughput, and the OS‑reported Wi‑Fi link speed (Tx rate) was used as a simple proxy for wireless link quality. These metrics are widely used in practice; however, the Tx rate represents a negotiated physical-layer upper bound and may overestimate effective throughput. Accordingly, we explicitly noted that the two metrics pertain to different layers and limited our analysis to distance dependence within each device.

At each scanner–Wi-Fi distance (0.5, 2.0, and 5.0 m), effective upload throughput was measured five times under the same condition immediately before scanning; the mean per condition was used in the analyses.

Analysis of scanning data

Scanning data from the reference model and each intraoral scanner were imported into a 3D analysis software (GOM Inspect 2020 Hotfix 6, GOM, Braunschweig, Germany). Prior to registration, each dataset was trimmed to remove nonrelevant areas and segmented to define two regions of interest (ROIs): (i) the three reference-body meshes used exclusively for alignment and (ii) the scan-body meshes used exclusively for deviation analysis. The three reference bodies attached to the reference model served as reference points to align the reference dataset with the intraoral scanner datasets. For alignment, the three cylindrical reference bodies on the alveolar ridge were segmented and used for coarse three-point registration, followed by an iterative closest point best‑fit with constraints to the reference‑body meshes only (scan‑body surfaces were masked during registration). Once the alignment was secured, deviation analysis was computed only on the scan-body ROIs using a point‑to‑surface distance algorithm. Subsequently, 3D positional deviations of the scan bodies in each dataset relative to the reference dataset were visualized using color mapping. Deviations of + 100 μm were indicated in red, ± 20 μm in green, and − 100 μm in blue, with intermediate values shown as gradients. The concordance rate was calculated as the percentage of scan-body surface area with an absolute deviation within ± 50 μm (|deviation| ≤ 50 μm) relative to the reference dataset. Implant-position reproducibility was evaluated by summarizing the distribution of this concordance rate using the median and interquartile range (IQR). The dataset with a concordance rate equal to the median value in each group was selected as the representative example for color mapping analysis. This approach was intended to accurately reflect typical deviation patterns within each group while minimizing visual complexity.

Statistical analysis

Prior to the statistical analysis, the normality of the data distribution was assessed using the Shapiro–Wilk test. Based on the results, nonparametric tests were selected. The median and interquartile range (IQR) of the matching areas of the scan bodies were calculated. Differences among groups were analyzed using the Kruskal–Wallis test. When significant differences were found, pairwise comparisons were performed using the Steel–Dwass test. All statistical analyses were conducted using IBM SPSS Statistics (version 22.0, IBM Corporation, Armonk, NY, USA), with the level of significance set at p < 0.05.

Additionally, to illustrate the relationship between upload speed and concordance rates for the six scan bodies, scatter plots of individual values for each condition were generated. A linear regression analysis was then performed to calculate the coefficient of determination (R²). Residual diagnostics were conducted to validate the regression models: residual normality was assessed with the Shapiro–Wilk test, homoscedasticity was confirmed through visual inspection of the residual plots, and linearity between the upload speed and concordance rate was verified using scatterplot analysis.

This study did not involve human or animal subjects; therefore, compliance with ARRIVE, STROBE, or CONSORT guidelines was not applicable.

Concordance rates for the six scan bodies

Figure 4 shows the concordance rates (defined as the percentage of scan‑body surface area within ± 50 μm [|deviation| ≤ 50 μm] of the reference dataset) for the six scan bodies for each intraoral scanner.

Fig. 4.

Concordance rates for the six scan bodies. Asterisks (*) indicate significant differences between groups (p < 0.05). Within identical distances, significant differences were noted between wireless intraoral scanners marked as follows: a and b; c and d; and among e, f, and g (p < 0.05)

The median concordance rates (IQR) for Primescan 2 were 82.3% (1.5), 82.3% (2.6), and 78.8% (5.2) at 0.5, 2.0, and 5.0 m, respectively. For SIRIOS, the rates were 64.6% (3.5), 54.4% (4.3), and 50.4% (4.6) at 0.5, 2.0, and 5.0 m, respectively. TRIOS 5 showed median concordance rates of 61.6% (19.2), 52.2% (24.0), and 29.5% (3.7) at 0.5, 2.0, and 5.0 m, respectively. The wired control, Primescan, showed a median concordance rate of 63.5% (4.7).

In the statistical analysis evaluating implant-position reproducibility across the six scan bodies, Primescan 2 showed significantly higher reproducibility than Primescan at all distances (p < 0.05), with no significant differences among distances for Primescan 2 (p > 0.05). SIRIOS showed significantly higher concordance rates at 0.5 m than those at 2.0 and 5.0 m (p < 0.05), but no significant differences were observed compared with Primescan (p > 0.05). For TRIOS 5, reproducibility at 5.0 m was significantly lower than that at 0.5 m and for Primescan (p < 0.05).

When the three wireless scanners were compared at identical distances, Primescan 2 was significantly higher than SIRIOS and TRIOS 5 at all distances (0.5, 2.0, and 5.0 m) (p < 0.05). A significant difference in the concordance rates between SIRIOS and TRIOS 5 at 5.0 m was also noted (p < 0.05).

Concordance rates for each scan body

Figure 5 shows the concordance rates for each scan body across the four intraoral scanners.

Fig. 5.

Concordance rates for each scan body. Subfigures are labeled consecutively as follows: (a) Primescan 2 (0.5 m), (b) Primescan 2 (2.0 m), (c) Primescan 2 (5.0 m), (d) SIRIOS (0.5 m), (e) SIRIOS (2.0 m), (f) SIRIOS (5.0 m), (g) TRIOS 5 (0.5 m), (h) TRIOS 5 (2.0 m), (i) TRIOS 5 (5.0 m), and (j) Primescan (wired). Within each subfigure, groups sharing the same uppercase letter are not significantly different, whereas different uppercase letters indicate significant differences (p < 0.05)

For Primescan 2, the median concordance rates (IQR) at 0.5 m were 59.6% (14.3), 88.9% (18.8), 84.5% (16.7), 88.7% (1.4), 91.8% (5.7), and 92.6% (3.1) at regions 47, 44, 42, 32, 34, and 37, respectively. At 2.0 m, the respective rates were 78.6% (24.1), 79.5% (5.8), 83.6% (6.3), 86.7% (4.2), 88.1% (1.7), and 89.9% (0.8). At 5.0 m, the corresponding rates were 74.0% (30.0), 87.9% (32.9), 87.7% (5.2), 88.4% (8.5), 90.8% (3.6), and 85.4% (13.3).

For SIRIOS, the median concordance rates (IQR) at 0.5 m were 51.0% (1.2), 53.5% (0.7), 58.8% (2.5), 60.7% (9.8), 80.0% (1.6), and 86.2% (4.2) at regions 47, 44, 42, 32, 34, and 37, respectively. At 2.0 m, the corresponding rates were 34.5% (5.2), 45.9% (6.7), 47.0% (9.9), 54.9% (21.6), 57.7% (18.8), and 76.4% (7.3). At 5.0 m, the respective rates were 32.6% (14.3), 51.6% (9.2), 48.6% (4.9), 49.3% (7.4), 61.9% (23.4), and 65.5% (35.4).

For TRIOS 5, the median concordance rates (IQR) at 0.5 m were 56.4% (18.1), 55.1% (16.5), 52.9% (18.4), 55.3% (10.9), 65.1% (13.6), and 73.5% (35.6) at regions 47, 44, 42, 32, 34, and 37, respectively. At 2.0 m, the respective rates were 37.6% (25.8), 48.8% (18.6), 49.6% (27.5), 45.8% (27.2), 50.6% (12.2), and 51.1% (16.6). At 5.0 m, the corresponding rates were 18.3% (4.9), 24.5% (5.9), 26.4% (20.0), 32.0% (10.8), 36.5% (4.9), and 46.4% (28.4).

For Primescan, the median concordance rates (IQR) were 37.2% (17.5), 67.5% (20.2), 66.9% (17.8), 66.1% (12.4), 80.1% (4.5), and 79.4% (1.9) at regions 47, 44, 42, 32, 34, and 37, respectively.

In the statistical analysis of implant-position reproducibility by scan body, Primescan 2 showed a significant difference between regions 47 and 37 at 0.5 m (p < 0.05), whereas no significant differences among regions were observed at 2.0–5.0 m (p > 0.05). For SIRIOS at 0.5 m, significant differences were found between concordance rates at region 47 and regions 32, 34, and 37; region 44 and regions 32, 34, and 37; region 42 and regions 34 and 37; and region 32 and regions 34 and 37 (p < 0.05). However, no significant differences in concordance rates were observed for SIRIOS at 2.0 and 5.0 m (p > 0.05). TRIOS 5 did not show any significant regional differences in concordance rates at any distance (p > 0.05). Primescan exhibited significant differences between concordance rates at region 47 and regions 32, 34, and 37 (p < 0.05).

Evaluation of color mapping

Median scanning data for Primescan 2, SIRIOS, TRIOS 5, and Primescan via color mapping are presented in Fig. 6.

Fig. 6.

Color-mapped 3D positional deviations of the scan bodies

Primescan 2 at 0.5 m showed minimal deviation within ± 20 μm across scan bodies. Slightly higher deviations (+ 100 μm) were noted at certain regions at greater distances. SIRIOS and TRIOS 5 showed significant deviations (± 100 μm) distributed broadly across scan bodies at all distances, particularly pronounced at 2.0 and 5.0 m. Primescan exhibited distinct directional deviations, notably − 100 μm on the buccal side and + 100 μm on the lingual side for regions 47, 44, and 42.

Relationship between the concordance rates for the six scan bodies and upload speeds

Figure 7; Table 1 show the relationships between the concordance rates for the six scan bodies and upload speeds. The average upload speeds were 186.0 Mbps (0.5 m), 150.8 Mbps (2.0 m), and 49.7 Mbps (5.0 m) for WXR18000BE10P, and 738.2 Mbps (0.5 m), 590.6 Mbps (2.0 m), and 369.1 Mbps (5.0 m) for TP-Link Archer T9UH. As the scanner–Wi-Fi distance increased from 0.5 m to 5.0 m, the effective upload speed declined for both wireless systems. Specifically, the average upload throughput dropped from 186.0 Mbps to 49.7 Mbps for the WXR18000BE10P router, and from 738.2 Mbps to 369.1 Mbps for the TP-Link Archer T9UH adapter, over this distance range.

Fig. 7.

Relationship between the concordance rates for the six scan bodies and upload speeds. Each point represents an individual scan result. Regression lines and coefficients of determination (R²) are shown. A weak correlation was observed for Primescan 2 (R² = 0.077), a strong positive correlation for SIRIOS (R² = 0.725), and a moderate positive correlation for TRIOS 5 (R² = 0.567)

Table 1.

Linear regression analysis between the upload speed and concordance rate for each intraoral scanner

Scanner	β	SE	p-value	95% CI	R²	Interpretation
Primescan 2	0.0177	0.017	0.318	−0.019 to 0.055	0.077	Not significant
SIRIOS	0.0369	0.006	< 0.001	0.023 to 0.051	0.725	Strong, significant
TRIOS 5	0.0707	0.017	0.001	0.034 to 0.108	0.567	Moderate, significant

β slope, CI Confidence interval, R² Coefficient of determination, SE Standard error

Linear regression analysis yielded the following results (Table 1):

• Primescan 2: β = 0.0177, standard error (SE) = 0.017, p = 0.318, 95% confidence interval (CI) = − 0.019 to 0.055, R² = 0.077 → not significant

• SIRIOS: β = 0.0369, SE = 0.006, p < 0.001, 95% CI = 0.023 to 0.051, R² = 0.725 → strong, significant positive correlation

• TRIOS 5: β = 0.0707, SE = 0.017, p = 0.001, 95% CI = 0.034 to 0.108, R² = 0.567 → moderate, significant positive correlation

These results indicate a significant dependence on the upload speed for SIRIOS and TRIOS 5, whereas Primescan 2 showed no significant correlation

Linear regression analysis demonstrated a strong positive correlation between upload speed and concordance rate for SIRIOS (R² = 0.725, p < 0.001) and a moderate positive correlation for TRIOS 5 (R² = 0.567, p = 0.001). In contrast, Primescan 2 showed no significant correlation between upload speed and implant-position reproducibility (concordance rate) (R² = 0.077, p = 0.318; see Table 1).

Discussion

This study aimed to evaluate the 3D reproducibility of implant positions using tissue-level implants embedded in a mandibular edentulous model, a crucial factor for the accurate fabrication of implant-supported prostheses. Tissue-level implants were specifically selected because they allow direct visual confirmation of the connection between the implant and scan body, facilitating highly accurate data acquisition. Additionally, manufacturer-certified PEEK scan bodies were used in this study, as a systematic review by Pachiou et al. [11] indicated superior scanning accuracy with PEEK scan bodies compared with that with metallic ones. Iwamoto et al. [8] investigated the performance of intraoral scanners using a permissible error threshold of 100 μm, which is acceptable performance for general clinical applications. In contrast, our study set a stricter permissible error threshold of 50 μm, which is more suitable for the precise fabrication of screw-retained superstructures. This decision was based on the findings of Wittneben et al. [12], who reported fewer mechanical and biological complications associated with screw-retained superstructures than those associated with cement-retained ones, as well as the recommendation by Katsoulis et al. [13] that the permissible error threshold for screw-retained prostheses should be within 50 μm. Although many studies employ best-fit algorithms to analyze scanning data [14–16], these algorithms optimize the overall deviation by minimizing global errors, potentially masking localized discrepancies. Consequently, this method may hinder the precise evaluation of impression accuracy, particularly regarding implant positioning, which is crucial for clinical success. To overcome this limitation, our study employed three strategically placed reference bodies attached directly to the alveolar ridge for data alignment. This approach facilitated a clinically relevant and precise evaluation of the spatial relationship between the alveolar ridge and implant positions, thereby closely reflecting actual clinical scenarios encountered during prosthetic fabrication. Intraoral scanners are generally designed to operate in either wired or wireless mode, and direct switching between the two within the same device is not technically available. For this reason, we employed Primescan 2 (wireless) and Primescan (wired) as control devices in this study. We selected the wired Primescan as the control because, in current clinical practice, a network‑independent wired intraoral scanner constitutes the standard benchmark, making it the appropriate comparator for isolating the effects of wireless communication.

Based on the analysis of concordance rates across the six scan bodies, Primescan 2 achieved the highest implant-position reproducibility at all tested distances, whereas the reproducibility of SIRIOS and TRIOS 5 declined as the scanner–Wi-Fi distance increased. Primescan 2 and Primescan both employ Dentsply Sirona’s proprietary Smart Pixel Sensor for high-speed image processing and Dynamic Depth Scan technology, enabling the acquisition of over 1 million 3D points per second while maintaining sharp, accurate scans at measurement depths up to approximately 20 mm [17, 18]. Despite this shared scanning approach, Primescan 2 consistently outperformed Primescan. This improvement is plausibly attributable to upgraded image sensors, optical systems, and image-processing pipelines, together with an optimized housing. In addition, because Primescan 2 uploads data directly to DS Core, Dentsply Sirona’s Google Cloud-based platform, its workflow is less dependent on local PC performance, which likely supported consistent reconstruction and data management. This cloud-based architecture may buffer transient network fluctuations and thereby deliver more stable, superior implant-position reproducibility compared with that delivered by the other wireless intraoral scanners (SIRIOS and TRIOS 5). This tendency is also corroborated by the clinical study by Schlenz et al., which found significant between‑system differences in trueness and precision among wireless intraoral scanners (Dexis IS 3800 W, Medit i700w, Primescan 2, and TRIOS 5), with Primescan 2 showing favorable—often superior—accuracy [19]. At 0.5 m, SIRIOS and TRIOS 5 exhibited implant-position reproducibility equivalent to that of Primescan. SIRIOS is characterized by structured-light, active triangulation, in which distortions of projected light patterns are analyzed from multiple viewpoints to reconstruct intraoral 3D geometry [20]. TRIOS 5 employs a confocal optical scanning approach that acquires multiple images at different focal planes to build high-definition 3D data [21]. These scanning principles, together with the short 0.5 m scanner–Wi-Fi distance, yielded reproducibility equivalent to that of the wired scanner. However, at 2.0 and 5.0 m, both SIRIOS and TRIOS 5 showed lower concordance than did Primescan. A likely explanation is their reliance on real-time, locally executed data processing that depends on the connected PC’s hardware specifications (e.g., central processing unit/graphics processing unit), introducing variability in the processing speed and stability. Mechanistically, increasing the scanner–Wi-Fi distance degrades link quality (rate adaptation and retransmissions), reduces effective throughput, and increases latency; in streaming, real-time reconstruction, such slowdowns lower the effective frame rate and can allow alignment errors to accumulate.

In the per-scan-body analysis, Primescan 2 maintained generally high concordance across all distances, and the color maps at 0.5 m showed only small deviations (approximately ± 20 μm). However, the IQR tended to widen at #47 as distance increased, and a consistent pattern emerged in which #47 exhibited lower concordance and #37 exhibited higher concordance. This may be attributed to #47 serving as the start/end point of the scan, where overlap is limited and cumulative registration drift is more likely to manifest, whereas #37 benefits from more continuous overlap during the scanning sequence. For SIRIOS, the median concordance decreased with increasing scanner–Wi-Fi distance, the IQR widened at 0.5 and 2.0 m and remained high at 5.0 m, and the color maps showed widespread deviations of approximately ± 100 μm—findings consistent with distance dependence. A plausible mechanism is that link-quality degradation with distance lowers the usable frame rate, promoting error accumulation; in addition, the device’s higher reliance on real-time local PC processing may make it more susceptible to instability from fluctuations in the processing load. In an in‑vitro study using wireless intraoral scanners (TRIOS 4 and TRIOS 5), Guo et al. showed that battery level and transmission distance significantly affect trueness and scan time (p < 0.05); moreover, precision was highest at high battery (76–100%) and short range (0.8 ± 0.5 m) [TRIOS 4: 54.92 ± 5.24 μm; TRIOS 5: 63.88 ± 12.86 μm]. The small degradation at 0.5 m and the distance‑dependent decline observed at 2.0–5.0 m in our study—together with the positive correlation between upload throughput and concordance—quantitatively support this dependence on link quality and operating conditions [22]. Notably, the publicly available specification lists 60 fps for SIRIOS [23], suggesting relatively greater sensitivity to frame rate drops as distance increases. For TRIOS 5, the median concordance declined with distance, but changes in the IQR were non-uniform and in some regions even decreased. Given the publicly reported frame rate of approximately 25 fps [24], the relative reduction in frame rate with increasing distance may have been smaller than that for SIRIOS, potentially limiting variance expansion. Nevertheless, the color maps at longer distances showed widely distributed deviations of ± 100 μm, aligning with the observed decline in medians. For Primescan, the #47–#37 contrast was the largest, and the color maps revealed bipolar buccolingual deviation at #47 (− 100 μm buccal / +100 μm lingual). Although Primescan and Primescan 2 share the same scanning engine/algorithms [17, 18], their execution environments differ: Primescan completes real-time processing on a local PC, whereas Primescan 2 uses a cloud-assisted workflow that includes asynchronous uploads to DS Core. These architectural differences may confer varying resilience to network fluctuations and processing load, which could in turn appear as between-device differences in concordance across scan bodies.

Regression analysis revealed that upload speed significantly correlated with implant-position reproducibility for SIRIOS and TRIOS 5, indicating substantial dependency on stable, high-speed Wi-Fi environments. In contrast, Primescan 2 revealed no significant correlation, demonstrating its ability to maintain high implant-position reproducibility even at lower upload speeds. This suggests that cloud-based architecture and asynchronous data transmission used by Primescan 2 may provide superior clinical flexibility regardless of the network conditions.

Our findings enable practical recommendations for clinicians using wireless intraoral scanners. First, clinicians should ensure a strong and stable Wi‑Fi connection by minimizing the distance between the scanner and Wi‑Fi access point. Our results show that SIRIOS and TRIOS 5 had better implant-position reproducibility at short ranges (0.5–2 m) than at longer distances (5 m). In practice, keeping the scanner’s receiver within a couple of meters of the router (line-of-sight if possible) can improve implant-position reproducibility. Second, clinicians should maintain high upload speeds when scanning. In this study, effective upload throughput above 500–600 Mbps (observed at 0.5–2 m distances) was associated with acceptable implant-position reproducibility for SIRIOS and TRIOS 5, whereas a decrease to 300–400 Mbps at 5 m led to notably reduced implant-position reproducibility. Therefore, clinicians should aim for a high-speed network environment. For example, they can use an up-to-date Wi‑Fi 6 router or adapter and a less congested 5 GHz band to achieve upload speeds in the several-hundred Mbps range. If network speed or stability is suboptimal (e.g., in clinics with congested Wi‑Fi or interference), it may be prudent to reposition the router closer to the operatory, use a wired connection if supported, or anticipate compromised implant-position reproducibility. Lastly, clinicians should be aware of device-specific capabilities. Our data revealed that the wireless scanner Primescan 2 was less affected by distance or speed variations than other scanners, likely due to its advanced hardware and cloud workflow. Clinicians should follow manufacturer guidelines and, when possible, perform test scans under their office network conditions to detect any reproducibility issues. Practitioners can optimize the implant-position reproducibility of wireless intraoral scans in daily clinical practice by implementing these measures, i.e., short scanner–Wi‑Fi distance, high upload bandwidth, and awareness of scanner performance.

This study has certain limitations. Because this was an in vitro study, it could not fully replicate the intra‑oral conditions. Specifically, our in vitro model lacked key intraoral factors, such as saliva, patient movement, and soft-tissue interference, which can impact scanning accuracy and the reproducibility of implant positioning; therefore, the direct clinical validity of our findings is limited, and caution is warranted when applying our findings to real-world, clinical scenarios. In this study, each wireless intraoral scanner was paired with the Wi‑Fi communication hardware recommended by its manufacturer. Parameters such as Wi‑Fi standard, channel width, transmit power, antenna design, and firmware can exert systematic effects beyond distance or nominal “speed.” Because these factors are collinear with the device, the device‑specific effects cannot be statistically separated from the router‑related contributions. Accordingly, the unit of inference in this study is the operational bundle of “device + router,” and generalization of the results is limited to these conditions. Although we controlled the environment as far as feasible—fixing the frequency band, channel, channel width, and transmit power; using a single client; and eliminating background traffic—this does not guarantee complete factor isolation. Future work will evaluate an orthogonal design using a common access point (AP) and a uniform measurement method. In addition, the network topology differed across devices (Primescan 2 via a router with WAN connectivity, whereas SIRIOS/TRIOS 5 used a direct local connection without WAN), making it impossible to apply a single speed‑measurement method (Speedtest) across all devices. We therefore used application‑layer effective throughput or PHY‑layer link rate, as appropriate to each configuration. Future studies will validate these findings using a common AP with local measurements. Finally, all acquisitions were performed by a single operator without randomization or blinding. Future research should consider using a standardized network setup (e.g., the same router or adapter across devices) and log comprehensive network quality data during scanning. When possible, comparing wired and wireless modes within the same scanner and using richer accuracy metrics (such as root mean square error) in a randomized setup would help isolate the impact of network factors on performance.

Conclusions

This in‑vitro study compared the implant‑position reproducibility of wireless and wired intraoral scanners and evaluated the impact of Wi‑Fi distance and upload speed on wireless devices. Primescan 2 exhibited higher implant‑position reproducibility than the wired Primescan across all distances and showed no distance‑ or upload‑speed sensitivity. SIRIOS and TRIOS 5 performed comparably to the wired comparator at 0.5 m but declined at 2.0–5.0 m. Upload speed correlated positively with reproducibility for SIRIOS and TRIOS 5 but not for Primescan 2, suggesting that wireless‑scanner reproducibility depends on device architecture and Wi‑Fi communication conditions.

Abbreviations

3D

Three-dimensional

CI

Confidence interval

IQR

Interquartile range

PC

Personal computer

PEEK

Polyetheretherketone

SE

Standard error

Wi-Fi

Wireless fidelity

Authors’ contributions

All authors contributed to the study conception and design. TMurakami acquired and interpreted the experimental data. KA, RI, TMiyashita, YM, YT, KS, and AO assisted in the acquisition and interpretation of the experimental data. All authors read and approved the final manuscript.

Funding

This work was supported by the Oral Implant Research Grant of the Japanese Society of Oral Implantology (No. JSOI-J2025-01).

Data availability

All data generated or analyzed during this study are included in this published article. Raw datasets underlying the summary tables and figures (beyond the presented charts) are available upon reasonable request to the corresponding author.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.