Image-enhanced endoscopy in upper gastrointestinal disease: focusing on texture and color enhancement imaging and red dichromatic imaging

Overview and development objectives of TXI

As discussed earlier, identifying lesions with subtle morphological or color changes using WLI alone is often challenging. Additionally, non-uniform illumination in the endoscopic image, caused by the curvature of the lumen and the direction of the illumination light, can make observation difficult because distant areas may appear darker. A novel image-enhancing technology, TXI, was developed to address these limitations with the aim of enhancing endoscopic images to improve the detection of lesions in the gastrointestinal tract. To achieve this goal, TXI utilizes retinex theory-based image processing technology to enhance three key image factors (texture, brightness, and color) to depict minute tissue differences more distinctly, while preserving the overall color and brightness consistency of the original white-light image (Fig. 1).17-19 Characteristically, TXI differs from other IEEs such as NBI, BLI, and LCI in that it does not emphasize specific wavelengths of light. In particular, while NBI is an optical filter technology that enhances vascular patterns by narrowing the light bandwidth, TXI is a computational post-processing technique that enhances the texture, brightness, and color tone of white-light images to improve lesion visibility.20 The main differences in the mechanisms and characteristics of TXI and NBI are summarized in Table 1.

Fig. 1.

Optimizing the three key elements of images in texture and color enhancement imaging (TXI). TXI emphasizes texture, brightness, and color tone to enhance the visibility of lesions. It seeks to augment the dimensional portrayal of subtle surface irregularities, enhance brightness in dark areas, and accentuate color alterations. WLI, white-light imaging.

Table 1.

Differences in mechanisms between NBI and TXI

The algorithm and fundamental principles of TXI

Understanding the basic principles and mechanisms of TXI would be beneficial before discussing their clinical usefulness. The process starts by dividing the input image into two layers: a base layer representing the illumination component and a detail layer highlighting the local contrast of brightness and color (Fig. 2).18,19 This division is accomplished using a single-scale retinex algorithm.21 Next, brightness correction is applied to the dark regions of the base layer. Achieving selective brightness correction across images is challenging owing to the limitations in controlling light sources. This can affect the entire image, causing a blurring effect around the edges of the brighter areas owing to the reflection and scattering of light. However, this correction can be achieved using processing techniques because the base layer corresponds to the illumination component. In this way, illumination of dark areas can be achieved while preventing overaugmented bright spots from obscuring the image. The subsequent steps focused on enhancing the texture and color. In Step 3, the dynamic range of the base layer is compressed to maintain the local contrast, which is necessary for subtle morphological or color changes in the detail layer. Step 4 involves enhancing the texture in the detail layer to improve subtle contrast. In the final step, the compressed base and enhanced detail layers are merged to generate a TXI image, which can be promptly displayed in TXI mode 2 or utilized for additional processing. Step 6 introduces a color enhancement algorithm designed to expand the color differences in the image, particularly between red and white hues.18 This further enhanced image is displayed as TXI mode 1.

Fig. 2.

A simplified flowchart of image processing algorithm of the texture and color enhancement imaging (TXI). Input image is initially divided into base and detail layers. Brightness correction is applied to dark regions of the base layer, and the dynamic range of the base layer is compressed to maintain local contrast. Texture in the detail layer is enhanced to improve subtle contrast. Compressed base and enhanced detail layers are then merged to generate a TXI image (mode 2). A color enhancement algorithm expands color differences, producing the final TXI mode 1 image. The goal of texture and color tone enhancement is to facilitate the recognition of structural and color tone disparities between lesions and the surrounding mucosa. WLI, white-light imaging.

In summary, the primary distinction between TXI modes 2 and 1 lies in their respective enhancements. TXI mode 2 combines brightness adjustments in darker regions with texture enhancements to emphasize subtle contrast. This generates images with color tones closer to those produced by WLI, which are more familiar to endoscopists. In contrast, TXI mode 1 includes additional color enhancement to delineate slight color contrast more clearly. The increased color contrast between red and white in TXI mode 1 renders the mucosa with a more pronounced reddish hue.

Gastric cancer

In clinical guidelines on gastric cancer, there has been no mention of TXI thus far, and limited publications have addressed its utility in gastric cancer diagnosis. Current studies on this subject have included a relatively small number of patients, emphasizing the color differences between tumor and non-tumor tissues (Table 2) and the enhancement of lesion visibility compared to WLI (Table 3).20,22-26

Table 2.

Color differences between gastric neoplasm and surrounding non-neoplastic mucosa, assessed with TXI and WLI in the evaluation of gastric cancer

Table 3.

Visibility scores assessed with TXI and WLI in the evaluation of gastric cancer

In a study by Ishikawa et al.,22 12 gastric cancer/adenoma lesions were evaluated. TXI mode 1 showed significantly higher color differences between neoplasm and surrounding mucosa (18.728±16.046) compared to WLI (8.000±4.263, p<0.01) or TXI mode 2 (10.246±8. 379, p=0.042), whereas TXI mode 2 was not superior to WLI. These trends persisted even when indigo carmine spray was applied, with the color difference increasing further in TXI mode. In visibility assessments, both TXI mode 1 and mode 2 showed significantly improved visibility compared to WLI (mode 1: 2.8±1.0, mode 2: 2.2±0.9 vs. WLI: 2.0±0.9; p<0.01 for both), with mode 1 exhibiting significantly higher values than mode 2. This trend persisted across all expertise levels, including both experts and trainees. While overall visibility trends were similar when indigo carmine spray was applied, trainees showed no difference between WLI and TXI, and the difference between TXI modes 1 and 2 disappeared.

In a study focusing on 20 cases of EGC, Abe et al.23 reported similar findings. Specifically, TXI mode 1 exhibited a significantly higher color difference compared to WLI (15.5±7.8 vs. 10.3±4.7, p=0.04), while there was no significant difference observed for TXI mode 2 compared to mode 1 or WLI. When evaluating lesion visibility, TXI modes 1 and 2 showed improvement in 7 out of 20 cases (35%) and 4 out of 20 cases (20%), respectively, compared to WLI. Particularly in TXI mode 1, there was a pronounced trend of improved visibility in macroscopic type 0-IIc or 0-IIb lesions compared to type 0-IIa or IIa+IIc lesions.

In another investigation involving 51 lesions of EGC undergoing endoscopic submucosal dissection (ESD), third-generation NBI, TXI, and WLI were compared to assess their ability to detect EGC lesions.20 In this study, Kawasaki et al.20 found that TXI mode 1 exhibited a significantly higher color difference compared to WLI (15.3 vs. 9.2, p<0.001), although there was no significant difference compared to NBI (15.3 vs. 13.5, p=0.330).

Koyama et al.,24 in their investigation comprising 31 EGC lesions using a high-definition ultrathin transnasal endoscope, observed that TXI mode 2 displayed a greater color difference than WLI (16.0±10.1 vs. 10.2±5.5, p<0.001). When 10 endoscopists evaluated visibility, TXI mode 2 showed significantly higher median visibility scores compared to WLI (4 [interquartile range, 4–4] vs. 4 [interquartile range, 3–4], p<0.001).

In a study by Futakuchi et al.,25 visibility of lesions was evaluated for 50 cases of gastric cancer and two cases of gastric adenoma. Both TXI modes 1 and 2 exhibited significantly higher visibility scores compared to WLI (TXI mode 1: 3.23±0.96, mode 2: 3.14±0.92 vs. WLI: 2.79±1.07; p<0.001). Interestingly, when analyzing expert endoscopists separately, there was no significant difference between TXI mode 2 and WLI. This trend was consistent regardless of whether a conventional or newly developed endoscope was used (GIF-H290Z; Olympus Corporation and GIF-XZ1200; Olympus Corporation, respectively).

There was also a study evaluating the utility of TXI in the diagnosis of gastric cancer after Helicobacter pylori eradication.26 After eradication therapy, EGCs often manifests as small, reddish, flat, or depressed lesions with indistinct borders, resembling a “gastritis-like” appearance, posing difficulty in detection during endoscopy.27 Therefore, efforts are required to effectively detect and evaluate EGC under this specific condition. In a study by Shijimaya et al.,26 which included a total of 58 lesions, both TXI modes 1 and 2 exhibited significantly higher color differences compared to WLI (mode 1: 22.90±0.96, mode 2: 15.32±0.71 vs. WLI: 11.88±0.26, both p<0.001). The mean color difference was significantly higher in TXI mode 1 compared to TXI mode 2 (p<0.001). Evaluation by six expert endoscopists revealed that the visibility score improved by 79.3% (46/58) and 54.4% (31/57) with TXI modes 1 and 2, respectively, compared with WLI. Interestingly, the mean visibility score was significantly higher for TXI mode 1 than for TXI mode 2 (p=0.0003). The authors also analyzed the diagnostic accuracy of the three trainee endoscopists. Despite the variability among examiners, in subjects with increased accuracy, the accuracies were 42/58 (72.4%) for WLI, 54/58 (93.1%) for TXI mode 1 (p<0.001), and 49/57 (86.0%) for TXI mode 2 (p<0.001).

Differing from the studies introduced above, Kemmoto et al.28 examined the efficacy of TXI in detecting gastric cancer during screening endoscopy in the general population. This study, the largest cohort to date, included 13,440 participants and compared gastric cancer detection rates between 10,745 in the WLI group and 2,695 in the TXI mode 2 group. The findings revealed a notably higher gastric cancer detection rate in the TXI mode 2 group than in the WLI group (0.71% vs. 0.29%, p=0.004). Similarly, when focusing solely on cases of gastric cancer developing after H. pylori eradication, the detection rate was significantly higher in the TXI mode 2 group than in the WLI group (1.36% vs. 0.43%, p=0.002). However, this significance was not observed when the subgroup analysis was restricted to subjects examined by expert endoscopists in the WLI group (gastric cancer: 0.71% vs. 0.68%, p=1.00; following H. pylori eradication: 1.36% vs. 0.78%, p=0.21).

Current research demonstrates that TXI offers advantages over WLI alone in the detection and visualization of gastric cancer or adenoma (Fig. 3A). Accumulating evidence also suggests that TXI mode 1 often exhibits even higher color differences and visibility improvements than TXI mode 2, although not consistently, implying that mode 1 may be the preferred option for evaluating gastric neoplasms or conducting screening endoscopies for gastric cancer. Additionally, TXI has been demonstrated to be effective in observing and detecting EGC after H. pylori eradication, which can sometimes be challenging owing to potential differences in morphology compared to typical cases of EGC. In some studies, the visibility effects varied depending on the macroscopic type of EGC or the expertise of the endoscopist, highlighting the need for further research to explore various situations in which TXI may be more effectively applied. This would enhance the practical implementation of TXI technology in clinical settings involving EGC.

Fig. 3.

Cases of upper gastrointestinal tract lesions observed using various modes of endoscopic imaging systems. (A) A case of early gastric cancer located on the posterior wall of the mid body of the stomach. The lesion is evaluated with white-light imaging (WLI), narrow-band imaging (NBI), and texture and color enhancement imaging (TXI) mode 1. (B) A case of atrophic gastritis with intestinal metaplasia evaluated with WLI, NBI, TXI mode 1, and TXI mode 2. The contrast between the lesion and the surrounding mucosa is enhanced in TXI images without darkening due to low illumination. TXI mode 2 shows an image more similar to WLI in color compared to TXI mode 1. (C) A case of esophageal low-grade dysplasia evaluated with WLI, TXI mode 1, and TXI mode 1 after Lugol solution application. TXI mode 1 highlights the redness of the lesion, clarifying its boundary, which aligns with the demarcation seen after Lugol solution application.

Atrophic gastritis and intestinal metaplasia

Atrophy and intestinal metaplasia (IM) are preneoplastic conditions that have gained significant attention due to their close association with gastric cancer development. Since the risk of cancer development varies significantly depending on the extent and severity of atrophy and IM, many international guidelines recommend risk stratification based on these conditions and surveillance for high-risk groups.6-8,29 However, due to various limitations associated with the pathological evaluation of atrophy and IM, such as difficulties in determining biopsy sites, increased risk of complications, costs, and inter-pathologist variability in results, interest in real-time endoscopic assessment for these conditions is increasing.6 Although there are growing expectations for endoscopic image-based assessment for this purpose, relying solely on WLI has limitations in terms of sensitivity, specificity, and interobserver agreement.30 This issue is now prompting the recommendation for the use of IEE to enhance the diagnostic capability of gastric atophy and IM.6,30

In recent years, endoscopic grading of gastric IM using NBI has been proposed in Europe, showing improved diagnostic accuracy compared to conventional WLI.31,32 Similarly, according to a prospective study in Korea, endoscopic evaluation using NBI was closely correlated with pathological evaluation for assessing gastric atrophy and IM, especially in groups with severe changes.13 Meanwhile, data on the effectiveness of TXI in diagnosing or assessing the risk of atrophy and IM are limited. To date, only a very small number of studies have been published on this topic (Table 4).22,33

Table 4.

Color differences and visibility scores assessed with TXI and WLI in the evaluation of atrophic gastritis

Ishikawa et al.22 conducted a study with 19 participants comparing the color difference between atrophic and nonatrophic mucosa. In this study, TXI mode 1 exhibited a significantly higher color difference compared to WLI (14.2±8.0 vs. 8.7±4.2, p=0.009). Interestingly, TXI mode 1 demonstrated notably higher color difference compared to mode 2 (14.2±8.0 vs. 10.0±4.2, p=0.017), paralleling the findings previously demonstrated in the detection and evaluation of gastric cancer.

TXI also demonstrated its advantage in another study by Sugimoto et al.,33 which investigated 60 individuals who underwent screening endoscopy as part of a health checkup using high-vision transnasal endoscopy. They divided the participants into NBI and TXI groups, with 30 individuals each. In particular, they evaluated the findings closely related to increased cancer risk and H. pylori infection status, including atrophic gastritis, IM, and map-like redness. In the TXI group, there was a significantly higher color difference compared to WLI for both atrophy and IM (20.8±9.7 vs. 14.0±7.3, p=0.003; 10.9±3.8 vs. 6.5±3.1, p<0.001, respectively). However, there was no significant difference in map-like redness between TXI and WLI ((13.5±5.6 vs. 13.1±3.1, p=0.885). When comparing TXI and NBI, no significant differences were observed in color differences surrounding atrophy, IM, and map-like redness (20.8±9.7 vs. 19.3±8.0, p=0.553; 10.9±3.8 vs. 13.5±4.7, p=0.057, and 13.5±5.6 vs. 11.7±9.3, p=0.703, respectively). This study concluded that third-generation high-resolution transnasal ultrathin endoscopy with TXI is effective in identifying atrophic borders and IM. These findings suggest that TXI could enhance the diagnostic efficiency of gastric atrophy and IM, thereby improving risk stratification for gastric cancer (Fig. 3B).

H. pylori infection status

The Kyoto Classification of Gastritis is a distinctive endoscopic classification that categorizes gastritis based on visual observations during endoscopic examination.34,35 It aims to diagnose H. pylori infection status and proposes a scoring system to predict the risk of gastric cancer based on endoscopic findings. However, the accurate identification of various endoscopic features can be challenging with WLI alone.

Kitagawa et al.14 evaluated the utility of TXI in diagnosing H. pylori gastritis. They retrospectively analyzed 22 image sets captured using WLI and TXI mode 1 from 60 patients. The diagnostic accuracy of TXI for active gastritis was significantly higher than that of WLI (85.3% vs. 78.7%, p=0.034), with TXI also demonstrating higher sensitivity (69.2% vs. 52.5%, p=0.012). This study also analyzed endoscopic findings related to the prediction of H. pylori infection status. Atrophy, IM, and diffuse redness were identified as features associated with the current infection state with high odds ratios (ORs) for both WLI and TXI. Specifically, diffuse redness was the only characteristic significantly associated with current infection for both WLI and TXI (OR, 22.0 and 56.1, respectively), whereas map-like redness was associated with past infection (OR, 6.3 and 11.0, respectively), and the regular arrangement of collecting venules was associated with non-infection (OR, 25.2 and 42.3, respectively). All specific endoscopic features associated with H. pylori infection status had a higher OR with TXI than with WLI. In terms of interobserver agreement, the kappa values among the reviewers were moderate to substantial for TXI (0.52–0.77), indicating higher values than those observed with WLI.

These results suggest that TXI could assist in predicting H. pylori infection status by enhancing the quality of the investigation of endoscopic features associated with H. pylori infection. However, further studies are required to validate these findings.

Esophageal diseases: Barrett’s esophagus, esophageal dysplasia and cancer

Esophageal cancer is the seventh most common cause of cancer-related deaths globally and presents as a multifaceted illness with varying etiologies depending on the histologic subtype and geographic location.36 Esophageal squamous cell carcinoma (ESCC) and adenocarcinoma exhibit distinct geographic distributions, with notably high rates of ESCC in Asia and Southeast Africa. The prognosis of esophageal cancer, particularly in advanced stages, remains poor, underscoring the importance of early detection. However, the early detection of esophageal cancer or its premalignant conditions is often challenging and requires meticulous inspection of the esophageal mucosa. For this purpose, it is beneficial to evaluate the surface pattern and microvascular structure.37 Observing the normal esophageal epithelium under ME reveals a fine vascular network distributed in the mucosa and submucosa, known as intrapapillary capillary loops (IPCLs).38,39 Changes in the morphology of IPCLs can be helpful for the differentiation of inflammatory lesions and neoplasms, and for the prediction of the severity of dysplastic change or invasion by esophageal cancer.

However, superficial ESCC is often detected in various forms such as erythema, subtle color changes, nodules, or combination of these findings, and concurrent lesions of different sizes are often present.40 These characteristics of esophageal neoplasms make it challenging to distinguish them from non-neoplastic lesions, and accurately delineating the boundaries, or horizontal margins, of lesions is sometimes difficult using WLI alone. IEE significantly improves the diagnostic accuracy for high-grade esophageal dysplasia and superficial ESCC, achieving over 90% accuracy, which is notably higher than that of WLI, and shows comparable results with Lugol chromoendoscopy.41,42 Therefore, the active use of Lugol chromoendoscopy or IEE is currently recommended to determine the extent of lesions before endoscopic resection of superficial ESCC.12,43 Some researchers have evaluated the efficacy of TXI in malignant and premalignant conditions of the esophagus (Fig. 3C).

Although rare, there are reports on the utility of TXI in the evaluation of squamous cell carcinoma (SCC) suspected lesions in the pharynx and esophagus. Dobashi et al.44 analyzed the color differences of 15 pharyngeal and 44 esophageal SCC suspected lesions using WLI, TXI mode 1, TXI mode 2, and NBI. After calculating the color difference between the lesions and surrounding mucosa, the mean values were 18.6 for TXI mode 1, 14.3 for TXI mode 2, and 17.2 for NBI, which were all significantly higher than the 11.6 observed for WLI (p<0.001). This trend persisted regardless of location, histology, gross type, and lesion size. Visibility assessed by six endoscopists showed improvement in 62.5% of lesions with TXI mode 1 and 88.1% with NBI. Moreover, significant differences were observed between all pairs of IEE modalities (TXI mode 1, mode 2, and NBI) for both experts (0.68±0.34 vs. 0.49±0.37 vs. 0.92±0.17, all significant) and non-experts (0.58±0.33 vs. 0.31±0.35 vs. 0.75±0.31, all significant) in terms of visibility scores, showing the highest scores in NBI and the lowest scores in TXI mode 2. In this study, TXI mode 1 enhanced color changes and improved the visibility of suspicious SCC lesions in the pharynx and esophagus compared to WLI, suggesting the utility of TXI in detecting and evaluating superficial SCCs in the esophagus. However, TXI exhibited slightly lower visibility outcomes than NBI, indicating the need for further research, including additional studies on the strength of TXI.

Barrett’s esophagus (BE) is an acquired condition characterized by the transformation of the stratified squamous epithelium of the distal esophagus into columnar epithelium, accompanied by the presence of specialized IM featuring morphologically typical goblet cells.45 BE is well known as a precursor lesion that increases the risk of esophageal adenocarcinoma, garnering considerable attention for its evaluation and management, especially in at-risk population.46 Accurate diagnosis of BE requires meticulous examination of key anatomical landmarks, such as the esophagogastric junction and the squamocolumnar junction.47 Integrated endoscopic imaging techniques, such as ME with NBI (ME-NBI), can enhance the visualization of microsurface patterns and microvascular patterns, potentially offering advantages in evaluating BE which is characterized by epithelial changes in the esophageal mucosa.

Ikeda et al.48 conducted a prospective study comparing TXI and WLI for the diagnosis of short-segment BE in 52 participants. When evaluating the color difference between esophageal mucosa and BE, TXI mode 1 demonstrated significantly higher values (27.2±7.7) compared to WLI (18.8±6.2, p<0.01) or TXI mode 2 (21.3±7.7, p<0.01). Similarly, when assessing the color difference between gastric mucosa and BE, TXI mode 1 showed significantly higher values compared to WLI (12.6±5.6 vs. 10.6±5.9, p<0.05), but there was no difference compared to TXI mode 2. In both cases, TXI mode 2 did not show a significant difference from WLI. Ten endoscopists evaluated the visibility; the percentage of improved lesion visibility compared to WLI was 78.8% for TXI-1 and 32.7% for TXI-2. The total visibility scores for all endoscopists were significantly higher with TXI-1 than with TXI-2 and significantly higher with both TXI-1 and TXI-2 than with NBI. When comparing visibility scores between the five trainees and five experts, both groups consistently showed higher values for TXI-1 than for TXI-2.

Another study compared the utility of TXI and NBI with WLI in evaluating BE in 40 patients using a third-generation ultrathin endoscope.49 The prevalence of BE showed no significant difference when assessed with WLI (82.5%), TXI (90.0%), and NBI (82.5%), respectively. Regarding the color difference between the esophageal mucosa and BE, NBI outperformed WLI and TXI. With regard to the color difference between the gastric mucosa and BE, both TXI and NBI were superior to WLI, with no significant differences between TXI and NBI. Particularly, the visibility score for detecting BE was higher with TXI (3.4±0.4) compared to WLI (2.9±0.8, p=0.016) or NBI (3.0±0.4, p=0.022).

These studies demonstrate that TXI holds promise for the sensitive and prompt detection of BE by enhancing subtle mucosal changes and highlighting color variations. Improvement in lesion visibility compared with WLI was notably more frequent for TXI mode 1 than for mode 2, implying that TXI mode 1 might be more effective for evaluating BE. Additionally, with TXI mode 1, there was a noticeable improvement in color separation, resulting in a clearer visualization of the BE. These results imply that TXI mode 1 facilitates clear visualization of BE and provides better contrast images not only at the squamocolumnar junction but also at the gastroesophageal junction. TXI mode 2 also improved the visibility of the BE, although it did not significantly enhance color differentiation. It is speculated that while the enhanced structure provided by texture enhancement may have contributed to this improvement with TXI mode 2, TXI mode 1 achieved a higher visibility score than TXI mode 2 because color tone enhancement leads to better color separation in BE.

Studies have shown that IEE in BE surveillance exhibits superior accuracy in detecting dysplasia or early carcinoma compared to WLI.50-52 According to the latest guidelines, routine application of acetic acid or the use of VCE is recommended during endoscopic surveillance of BE.46 Especially, careful inspection of Barrett's mucosa with the use of high-definition WLI and VCE is recommended during surveillance after endoscopic eradication therapy to monitor dysplasia recurrence.53 Hence, it can be inferred that TXI may also be beneficial for BE surveillance, although literature on this topic is limited, indicating the need for further research. Future studies investigating the utility of TXI in the surveillance of BE would be valuable.

Superficial non-ampullary duodenal epithelial tumor

Adenocarcinoma arising from the small bowel is exceedingly rare and primarily found in the duodenum.54 Some reports suggest that the risk of small bowel adenocarcinoma is higher among African Americans and males. Notably, individuals with specific hereditary mutations, such as familial adenomatous polyposis, Lynch syndrome, or Peutz–Jeghers syndrome, have a significantly increased risk of small bowel adenocarcinoma.54,55

Duodenal cancer is often associated with superficial non-ampullary duodenal epithelial tumor (SNADETs), presumed to be its precursors.56 Despite their low prevalence, ranging from 0.03% to 0.4% among patients undergoing screening or diagnostic endoscopy, SNADETs have clinical significance as they can progress to duodenal adenocarcinoma, which shows a poor prognosis.57-60

Therefore, an early diagnosis is crucial for patients in at-risk groups, ideally at the premalignant stage. European guidelines recommend the use of ME-NBI for the diagnosis and staging of SNADETs.61 Accurate localization and characterization of SNADET lesions are essential for guiding biopsy sampling and determining the appropriate treatment strategy, highlighting the importance of IEE. However, research on the utility of TXI in the detection and evaluation of SNADETs is extremely limited and is not mentioned in the guidelines.

Okimoto et al.62 conducted a prospective pilot study involving 12 SNADET lesions, evaluating the visibility of the surface structure and blood vessels. In this study, they employed ME with indigo carmine (ICME-TXI) application for the evaluation. The surface structure visibility score of ICME-TXI was significantly higher than that of ME-NBI, ME-TXI, and ICME-WLI (p<0.001, respectively). They also found very high intra- and interobserver agreements for visibility improvement. Moreover, all endoscopists reported that they preferred ICME-TXI for visualizing the surface structures in all lesions. This study demonstrated that TXI, in combination with classic IEE methods, could enhance the visibility of surface structures in SNADETs, potentially aiding their accurate preprocedural evaluation and diagnosis.

Experiences of TXI in real clinical practice

Few studies have compared TXI with NBI, which is a widely used IEE technique. Based on the research introduced earlier, the findings are as follows. There is no significant difference between TXI and NBI in evaluating gastric cancer or gastric mucosal conditions related to H. pylori status. In the case of BE, NBI exhibited a higher color difference, whereas TXI showed higher visibility, resulting in mixed outcomes. TXI demonstrated lower visibility than NBI for neoplastic lesions in the pharynx and esophagus. The effectiveness of TXI compared with that of NBI in diagnosis is generally similar or sometimes lower, which can be tentatively attributed to several factors.

First, NBI emphasizes changes in surface vascular patterns, but has the disadvantage of insufficient light intensity compared to WLI. This makes NBI particularly advantageous in narrower lumens such as the esophagus or colon, or during close mucosal observations. Consequently, the lower contrast of TXI compared with NBI for some lesions might be a contributing factor. Indeed, NBI has outperformed TXI in visualizing the surface patterns of serrated colorectal polyps in previous studies.63,64 Second, the current generation of NBI (second and third) has improved light intensity, overcoming previous disadvantages and providing bright, high-resolution images.65 This enhancement likely contributes to NBI’s superior performance. Third, many examiners are already well acquainted with the characteristics of lesions, such as cancer or dysplasia, under NBI, thanks to extensive experience and literature. Conversely, TXI is a relatively new technology with relatively low familiarity and the systematic classification of lesion findings is not yet well established, potentially leading to underperformance. Finally, most studies to date have evaluated color differences and visibility scores. Future well-designed studies and data accumulation on the detection rates and diagnostic capabilities in screening settings are warranted to fully assess the potential of TXI.

In conclusion, although TXI shows promise in the field of UGI disease, its overall effectiveness compared to NBI remains to be fully determined. Continued research and familiarity with TXI are crucial for their broad adoption in clinical practice.

Studies have also compared the TXI modes 1 and 2. For the evaluation of gastric neoplasms and mucosal atrophy, TXI mode 1 generally demonstrates a better performance in terms of color difference and visibility. The improvement in lesion visibility was also notably higher for TXI mode 1 than for mode 2 in the evaluation of BE and esophageal SCCs, suggesting that TXI mode 1 might be more effective for the evaluation of these conditions.

The primary distinction between TXI modes 2 and 1 is their respective enhancements. TXI mode 2 combines brightness adjustments in darker regions with texture enhancements to emphasize subtle contrast. This generates images with color tones similar to those of WLI, which are more familiar to endoscopists. Despite its generally lower efficacy than mode 1, TXI mode 2 is particularly beneficial in situations requiring improved visibility in dark regions and a more familiar range of colors. It could be useful in screening and surveillance as well as for training and familiarization purposes. In contrast, TXI mode 1 includes additional color enhancement. This leads to a clearer delineation of slight color contrast, making it particularly effective for highlighting fine mucosal structures and subtle lesions, which is highly useful for detailed inspection. Therefore, this mode can aid in the identification of minute mucosal irregularities and early neoplastic changes.

In addition to the type of lesion or the purpose of the examination, user factors such as the experience level, preference, and familiarity of the endoscopist also influence the efficacy of the mode and are significant considerations in choosing the mode. Understanding the specific advantages and appropriate uses of each mode would enable optimal utilization of TXI technology in clinical practice.

Enhanced visualization of deep blood vessels

As previously stated, the amber and red wavelengths penetrate deeply into the mucosa, allowing the visualization of deep blood vessels. Recognition of deep vessels could help prevent bleeding by avoiding major vessels during submucosal injections or incisions during procedures such as ESD or POEM. These characteristics can also be utilized for evaluating submucosal vessels such as esophageal varices.

In a pilot study involving 60 patients who underwent esophageal ESD, the analysis focused on whether the use of RDI reduced bleeding and hematoma occurrence during submucosal injection.71 Compared to the WLI group, the RDI group exhibited significantly lower incidence of hematomas per unit area (0/cm² vs. 0.18/cm², p=0.024) and total bleeding episodes (16.7% vs. 42.9%, p<0.001). Moreover, significantly more blood vessels were observed in the RDI group than in the WLI group (5 vs. 2, p=0.0020). These findings are likely attributed to the increased visibility of blood vessels in the submucosal layer observed using RDI.

Enhanced visualization of active bleeding point

2) Bleeding during ESD

Meanwhile, there has been a relatively greater volume of research conducted on bleeding during advanced endoscopic procedures. In a study including 20 patients undergoing gastric ESD, the efficacy of RDI in enhancing the visibility of bleeding was examined.75 Results revealed that RDI significantly outperformed WLI in terms of mean visibility score (3.69±0.60 vs. 3.20±0.84, p<0.01), and the color difference achieved with RDI was also notably higher than that with WLI (19.51±15.18 vs. 14.80±7.41, p<0.01). Furthermore, the submergence of bleeding points was identified as independently associated to the superior performance of RDI in the multivariate analysis (OR, 10.35; 95% confidence interval, 2.76–38.81; p<0.01).

Recently, another study by Mori et al.76 evaluated the recognition capabilities of RDI for 58 cases of active bleeding points in patients undergoing esophageal or gastric ESD. The results showed that RDI significantly outperformed WLI, with a higher median visibility score and recognition rate (2.17 vs. 1.42, p<0.001 and 62.1% vs. 27.6%, p<0.001, respectively). Additionally, the median color difference of RDI was notably higher than that of WLI (8.97 vs. 3.69, p<0.001). Moreover, a strong correlation was observed between the visibility score and color difference. Interestingly, the recognition rate of trainees using the RDI surpassed that of experts using the WLI (60.3% vs. 43.1%, p=0.067).

These findings collectively underscore the potential of RDI to improve bleeding visibility by enhancing the color contrast during UGI procedures that require hemostasis (Fig. 5). This suggests that RDI can be used to quickly identify the bleeding focus during ESD and achieve effective hemostasis, demonstrating its potential utility even for less experienced operators.

Fig. 5.

Identification of bleeding point using red dichromatic imaging (RDI) during gastric endoscopic submucosal dissection. The bleeding point is clearly visualized in RDI mode by enhancing the contrast between the highly concentrated blood and the surrounding diluted blood pool. WLI, white-light imaging.

Application in ESD or POEM

Other studies have explored the application of RDI in gastric ESD. In challenging scenarios where a hematoma was present in the submucosal layer, RDI aided in precisely locating the dissection line during the procedure, leveraging its see-through effect.77 Additionally, the effectiveness of RDI for ensuring clear visibility throughout ESD was also noted, even in situations with considerable adipose tissue in the submucosal layer.78

Kita et al.79 reported that employing RDI throughout the entire cutting process during ESD significantly enhances dissection speed compared to performing ESD with WLI alone. The study included 40 esophageal, 17 gastric, and 25 colorectal cancers, and the median dissection speed in the full-time RDI group was significantly faster than that in the WLI-ESD group (27.23 mm²/min vs. 20.94 mm²/min, p=0.025). The results showed that full-time RDI during ESD contributed to a faster dissection speed without compromising en bloc resection, negative resection margins, or increasing adverse events compared to ESD performed with WLI alone.

Another study by Nabi et al.80 reported the successful implementation of the TXI and RDI modes during ESD or POEM procedures in 25 patients. RDI proved to be beneficial in identifying the bleeding site during endoscopic dissection. Through this experience, the authors propose that the new image-enhanced technique facilitates the convenient execution of submucosal tunneling and endoscopic dissection procedures.

To summarize, studies have consistently reported partial or complete utilization of RDI in scenarios involving bleeding peptic ulcers or procedures such as ESD for esophageal cancer and EGC, as well as POEM in esophageal motility disorders.81-86 The advantages of RDI include better identification of blood vessels, leading to reduced bleeding during the injection and dissection process. Moreover, bleeding points were easily detected using RDI, enabling prompt and effective hemostasis.