Introduction

There have been many reports on the usefulness of magnifying endoscopy with image-enhanced endoscopic techniques. In recent years, the magnifying endoscopy simple diagnostic algorithm for early gastric cancer (MESDA-G)^１ has been proposed as a unified diagnostic system. However, it is a “static” diagnostic algorithm that uses still images. We focused on real-time red blood cell flow in subepithelial microvascular visualized by magnifying endoscopy, and clarified that microvascular blood flow rate was significantly lower in early gastric cancer (EGC) than in benign patchy redness in the pilot study².
Moreover, we uniquely developed an automated blood flow rate analysis system (hereinafter “analysis system”) and calculated the cut-off value (1090.9um/sec) for endoscopic diagnosis of EGC using the test dataset. However, comparative studies on diagnostic accuracy between endoscopists and analysis system have not yet been analyzed.

Aims & Methods

The aim of this study was to compare the qualitative diagnostic accuracy of EGC between endoscopists and analysis system. We retrospectively reviewed magnifying endoscopy with blue laser imaging (M-BLI) videos of differentiated-type EGC and patchy redness at our hospital between December 2017 and September 2021. M-BLI videos of 31 differentiated-type EGCs and 40 patchy redness in which red blood cell flow in subepithelial microvascular was visible were included in this study. In the analysis system group, the diagnosis was made according to the above-mentioned cut-off value (1.09 mm/sec) for cancer/non-cancer. In the endoscopist group, 9 endoscopists (5 experts and 4 trainees) reviewed these M-BLI videos and made the diagnosis as cancer/non-cancer according to MESDA-G. We evaluated the diagnostic accuracy in each group, such as sensitivity/specificity/positive predictive value (PPV)/negative predictive value (NPV).

Results

In 31 EGCs, there were more men than women (22/9), the mean age was 72.2±9.5 years, the tumor location was U/M/L=2/8/21, the macroscopic type was elevated/depressed=8/23, the histological type was tub1/tub2/pap=24/6/1, and depth of invasion was M/SM=27/4. In 40 patchy redness, there were more men than women (28/12), the mean age was 69.0±9.7 years, and the location was U/M/L=3/12/25. With regard to diagnostic accuracy, sensitivity/specificity/PPV/NPV was 90.3/89.7/87.2/92.3% in the analysis system group, 85.9/87.2/80.0/85.5% in expert endoscopists, and 70.0/82.7/80.6/71.0% in trainees endoscopists, respectively.

	Analysis system	Endoscopists
		Experts (n=5)	Trainees (n=4)	All (n=9)
Sensitivity % (95%CI)	90.3 (0.68-0.95)	85.9 (57.0-87.6)	70.0 (54.8-85.8)	77.9 (55.7-86.6)
Specificity % (95%CI)	89.7 (0.79-0.98)	87.2 (65.4-90.8)	82.7 (64.3-92.4)	85.0 (64.8-91.7)
PPV % (95%CI)	87.2 (0.74-0.98)	80.0 (55.5-84.0)	80.6 (59.8-90.8)	80.3 (58.0-87.9)
NPV % (95%CI)	92.3 (0.73-0.96)	85.5 (61.3-85.5)	71.0 (60.0-87.2)	78.3 (60.8-86.4)

Conclusion

The automated blood flow rate analysis system showed higher diagnostic accuracy in the differential diagnosis between differentiated-type EGC and patchy redness compared with endoscopists. It is suggested that the analysis system may be useful as a new diagnostic modality for EGC and could potentially diagnose lesions that are difficult for endoscopists to diagnose.

References

1. Muto M et al. Dig Endosc. 2016. doi: 10.1111/den.12638.
2. Ueyama H et al. J Gastroenterol Hepatol. 2021. doi: 10.1111/jgh.15425.

Introduction

Recently, artificial intelligence (AI)-based diagnostic support tools for early gastric cancer have been developed and reported to be superior to endoscopists (accuracy: 70% and sensitivity: 85%).¹ However, These tools could not guarantee reliable diagnostic accuracy in clinical practice. In previous reports, magnifying endoscopy has been recognized as superior to white-light imaging endoscopy for endoscopic diagnosis of early gastric cancers. Therefore, we developed an AI-based diagnostic support system using magnifying endoscopy with narrow band imaging (M-NBI) for early gastric cancers² and improved its application in clinical practice using nondegraded images obtained from selective depth inspection (SDI) output for real-time diagnosis. This study is the first to create an AI-based diagnostic support system using non-degraded M-NBI images obtained from SDI output, as several previous papers have analyzed damaged M-NBI images that were converted from various recording devices.^3,4

Aims & Methods

We built a new AI system (M-NBI AI) which was assembled using four different image recognition models with non-degraded M-NBI images (normal mucosa:1776 images, gastric cancers: 1947 images, and non-neoplastic lesions: 2451 images). We evaluated the performance of this M-NBI AI for diagnosis of early gastric cancer. The diagnostic performances of M-NBI AI and endoscopists (five specialists and five non-specialists) were compared using non-degraded M-NBI images (three images per lesion) of 50 early gastric cancer and 50 non-cancerous lesions. Early gastric cancers included histopathological types of well, moderately, and poorly differentiated adenocarcinomas and signet-ring cell adenocarcinomas. The primary outcome was the superiority of the M-NBI AI (based on accuracy) over the endoscopists. The secondary outcome was the superiority of the M-NBI AI (based on sensitivity) over the endoscopists.

Results

	M-NBI AI	Endoscopists
		Specialists	Non-specialists
Accuracy, % (95% CI)	87.0 (0.80–0.94)	80.0 (0.72–0.88)	71.0 (0.62–0.79)
Sensitivity, % (95% CI)	86.3 (0.77–0.96)	74.2 (0.63–0.85)	70.6 (0.58–0.83)
Specificity, % (95% CI)	88.0 (0.79–0.97)	89.0 (0.80–0.99)	71.0 (0.59–0.84)
PPV, % (95% CI)	88.0 (0.79–0.97)	92.0 (0.84–0.99)	72.0 (0.60–0.84)
NPV, % (95% CI)	86.0 (0.76–0.96)	68.0 (0.55–0.81)	70.0 (0.57–0.83)
PPV, positive predictive value; NPV, negative predictive value

The M-NBI AI was superior to the non-specialists based on accuracy (87.0%, 95%CI: 0.80–0.94 versus 71.0 %, 95%CI: 0.62–0.79; p<0.01), without being superior to the specialists (87.0%, 95%CI: 0.80–0.94 versus 80.0%, 95%CI: 0.72–0.88; p=0.18). The M-NBI AI was superior to the specialists and non-specialists based on sensitivity (86.3%, 95%CI: 0.77–0.96 versus 74.2%, 95%CI: 0.63–0.85 and 70.6%, 95%CI: 0.58–0.83; p<0.05, respectively).

Conclusion

The M-NBI AI was created as a new AI system that could be implemented for current endoscopy systems as a real-time diagnostic support tool for early gastric cancers. It demonstrated a robust performance and may indicate equalization of diagnostic accuracy and superadded effect over the white-light endoscopy for diagnosis of early gastric cancers in clinical practice.

References

1. Ishioka M et al. Dig Endosc. 2022. doi: 10.1111/den.14455.
2. Ueyama H et al. J Gastroenterol Hepatol. 2021;36(2):482-489.
3. Horiuchi Y et al. Dig. Dis. Sci. 2020;65:1355–1363.
4. He X et al. Gastrointest Endosc. 2022;95(4):671-678.

Introduction

The prevalence of Helicobacter pylori (H.pylori) positive patients has gradually decreased in Japan due to eradication therapy, and H.pylori-uninfected gastric cancer (HPUGC) is expected to increase relatively. Recently, HPUGC, such as gastric adenocarcinoma of fundic-gland type (GAFG), gastric adenocarcinoma of foveolar-type (GAFV), adenocarcinoma of differentiated type (DGA) and signet ring cell carcinoma (SRCC) has been increasingly reported. Since the detection and endoscopic diagnosis of HPUGC is considered to be difficult [1]. The effective observation methods in screening endoscopy to detect HPUGC. In this context, a new image enhancement technology called Texture and Color Enhancement Imaging (TXI; Olympus Medical Systems Corporation) which optimizes the three elements: structure (texture) enhancement, color tone, and brightness correction was developed in 2020. There have been several reports on the usefulness of TXI for the detection and endoscopic diagnosis of mucosal atrophy or gastric neoplasm [2-6]. However, the usefulness of TXI for the detection of HPUGC has been investigated.

Aims & Methods

The aim of the study is to investigate whether the visibility of HPUGC improved using TXI and NBI compared with WLI as a single-center retrospective clinical study. We enrolled 40 HPUGC lesions endoscopically imaged with WLI, TXI-1, TXI-2, and NBI in our facility between October 2020 and March 2023. In addition, the visibility of HPUGCs was analyzed for each factor, such as histological type, coloration, and macroscopic type. Ten endoscopists (5 experts and 5 non-experts) evaluated the images, and visibility changes were scored by endoscopists as follows: 5, improved; 4, somewhat improved; 3, equivalent; 2, somewhat decreased; and 1, decreased. The total visibility score of ≥40 in all endoscopists and /≥20 in each of the 5 endoscopists was defined as improved visibility, 21-39/11-19 as unchanged visibility, and 20≤/10≤ as decreased visibility. The intraclass correlation coefficients (ICC) were calculated and evaluated for inter-rater agreement.

Results

The histological types were classified as GAFG (n=24), GAFV (n=9), DGA (n=4), and SRCC (n=3). There were 17 reddish lesions and 23 whitish lesions. In the macroscopic type, there were type 0-Ⅰ (n=11), type 0-Ⅱa (n=13), type 0-Ⅱb (n=11), and type 0-Ⅱc (n=5) lesions. The visibility improvement rate for all endoscopists in all HPUGCs was TXI-1: 55%, TXI-2: 29.7%, and NBI: 7.5%. The overall ICCs were TXI-1: 0.86, TXI-2: 0.84, and NBI: 0.90, which were almost perfect. GAFG (all/experts/trainees: 79/54/79%) and SRCC (all/experts/trainees: 67/67/67%) showed improved visibility with TXI-1. In contrast, GAFV and DGA did not improve the visibility. The whitish lesions (all/experts/trainees: 74/52/74%) showed improved the visibility with TXI-1, but the reddish lesions did not improve visibility. Among the macroscopic type , type 0-Ⅱa (overall/experts/trainees: 85/54/85%) and type 0-Ⅱb (overall/experts/trainees: 73/64/73%) showed improved visibility with TXI-1. The all ICCs of each factor were above 0.7, which indicated generally good correlations.

Conclusion

The usefulness of TXI-1 compared with WLI led to the improved visibility of the whitish lesions such as GAFG and SRCC for both experts and trainees when evaluated subjectively and objectively.

References

1. Ishibashi, F., et al., Exploring quality indicators for the detection of Helicobacter pylori-naïve gastric cancer: a cross-sectional nationwide survey. Clinical Endoscopy, 2023. Apr 4. doi: 10.5946/ce.2022.167.
2. Abe, S., et al., Emerging texture and color enhancement imaging in early gastric cancer. Dig Endosc, 2022. 34(4): p. 714-720.
3. Sugimoto, M., et al., Efficacy of high-vision transnasal endoscopy using texture and colour enhancement imaging and narrow-band imaging to evaluate gastritis: a randomized controlled trial. Ann Med, 2022. 54(1): p. 1004-1013.
4. Abe, S., et al., Visibility of early gastric cancer in texture and color enhancement imaging. DEN Open, 2022. 2(1): p. e46.
5. Ishikawa, T., et al., Efficacy of Texture and Color Enhancement Imaging in visualizing gastric mucosal atrophy and gastric neoplasms. Sci Rep, 2021. 11(1): p. 6910.
6. Kawasaki, A., et al., Usefulness of third-generation narrow band imaging and texture and color enhancement imaging in improving visibility of superficial early gastric cancer: A study using color difference. DEN Open, 2022. 3(1): p. e186.

Introduction

Non-ampullary duodenal epithelial tumors (NADETs) have been reported to exhibit specific endoscopic and clinicopathological features based on mucin phenotypes: the gastric phenotype (GP) and the intestinal phenotype (IP) ¹. However, NADETs with the gastrointestinal mixed phenotype (MP) are often observed in clinical practice. And, the endoscopic and clinicopathological features including subclassification of NADETs with MP have not been well investigated.

Aims & Methods

The aim of this study is to clarify the endoscopic and clinicopathological features of each phenotype group; GP, IP, and MP. NADETs were collected from a database of endoscopic and surgical resections performed between February 2010 and December 2021 at our hospital. All NADETs were classified into three groups immunohistochemically: GP, IP, and MP. In addition, we subclassified MP into the gastric predominant phenotype (GPP) and the intestinal predominant phenotype (IPP) based on the predominance of gastric or intestinal phenotypic expression. The endoscopic and clinicopathological findings were compared between each phenotype group; GP, GPP, IPP, and IP.

Results

A total of 148 NADETs in 142 patients were classified into GP (n=16), GPP (n=16), IPP (n=43), and IP (n=73). As for the tumor location (first/second/third portion), GP (10/6/0) and GPP (9/5/2) were significantly more often located in the first portion compared to IPP (10/30/3) and IP (12/55/6) (p ‹ 0.01). The mean tumor size of GP (14.4 mm) and GPP (19.4 mm) were significantly larger than IPP (11.0 mm) and IP (10.7 mm) (p ‹ 0.01). In Vienna classification (Category 3/4 or 5), the proportion of category 4 or 5 was significantly higher in GP (7/9) and GPP (2/14) than in IPP (35/8) and IP (71/2) (p ‹ 0.01). Regarding the endoscopic findings (Table), the frequency of reddish coloration, 0-I macroscopic type, and lobular/granular pattern in white-light imaging was significantly higher in GP and GPP than in IPP and IP (p ‹ 0.01). The frequency of irregular microsurface pattern, negativity of white opaque substance, positivity of oval-shaped marginal epithelium and dilatation of the intervening part in magnifying endoscopy with narrow-band imaging was significantly higher in GP and GPP than in IPP and IP (p ‹ 0.01). In summary, the endoscopic and clinicopathological features of IPP were similar to those of IP, and GPP were similar to GP. The malignant potential of NADETs was presented in the descending order as follows; GPP › GP › IPP ≒ IP, and GPP demonstrated the highest malignant potential among 4 types.

	GP	GPP	IPP	IP
White-light imaging	n=16	n=16	n=43	n=73
Coloration (red/white)	13/3	11/5	17/26	21/52
Macroscopic type (0-I/0-IIa/0-IIc)	11/1/3	10/4/1	5/21/17	11/38/23
Lobular/granular pattern (%)	62.5%	62.5%	7.0%	11.0%
Magnifying endoscopy with narrow-band imaging	n=12	n=14	n=41	n=70
Microsurface pattern (regular/irregular/absent)	11/1/0	5/9/0	34/7/0	59/11/0
White opaque substance (%)	25.0%	64.3%	90.2%	90.0%
Oval-shaped marginal epithelium (%)	58.3%	64.3%	9.8%	4.3%
Dilatation of the intervening part (%)	75.0%	64.3%	22.0%	12.9%

Conclusion

NADETs exhibited specific features based on mucin phenotypes as previous reports¹, and MP (GPP and IPP) demonstrates the endoscopic and clinicopathological features of its predominant mucin phenotype (GP or IP). The accurate endoscopic diagnosis based on mucin phenotypes is necessary for appropriate treatment approach of NADETs, since the malignant potential of NADETs may differ between each mucin phenotype.

References

1. Akazawa Y. et al. Digestion 2021;102:663–670 DOI: 10.1159/000508040

Disclosure

Nothing.

Introduction

Fundic gland polyp associated with proton pump inhibitor (PPI-FGP) caused by long-term administration of proton pump inhibitors is known to be larger than normal FGP and have an edematous appearance, however, the endoscopic differentiation from normal FGP is sometimes difficult. Recently, Texture and Color Enhancement Imaging (TXI) has emerged as a novel imaged enhancement technology from Olympus Corporation and is expected to improve the visibility of the lesions by enhancing color tone and structure. The enhancement of color contrast of the mode1 is superior to that of mode2 in accordance with the color enhancement of the mode1 algorithm; however, the naturalness in mode2 is better than that of the mode1¹. Recently, we previously detected the gray color sign as a novel endoscopic finding that was clearly visible by TXI in the edematous appearance of PPI-FGP, and defined that it was gland duct dilatation below the surface containing mucus and other substances in histopathology.

Aims & Methods

The aim of this study was to evaluate the usefulness of TXI compared with white light imaging (WLI) in the visibility of the gray color sign in PPI-FGP. We enrolled 23 PPI-FGP lesions, endoscopically imaged with WLI, TXI-1, TXI-2, and Narrow Band Imaging (NBI) between April 2021 and October 2022. All PPI-FGP lesions were diagnosed by biopsy specimen histopathologically. Ten endoscopists (5 trainees, 5 experts) evaluated the images and scored the visibility of the gray color sign in TXI-1, TXI-2, and NBI compared to WLI. Visibility changes were scored by endoscopists as follows: 5, improved; 4, somewhat improved; 3, equivalent; 2, somewhat decreased; and 1, decreased. For total scores, 40 points or more was considered improved visibility, 21–39 points was comparable to WLI, and a score of less than 20 was considered decreased visibility. Inter-rater reliability was also evaluated using intraclass correlation (ICC). In addition, the color difference (ΔE*) of the gray color sign was calculated using CIELAB (color space system: L* a* b*) defined by Commission internationale de l'éclairage (CIE), and compared between WLI, TXI-1 and TXI-2.

Results

The visibility and ICC scores of TXI-1 and TXI-2 in trainees, experts, and all endoscopists are shown in Table1. TXI-1 and TXI-2 showed improved visibility of the gray color sign in PPI-FGP for all endoscopists. There was no significant difference in the visibility scores between trainees and experts for both TXI-1 and TXI-2. Compared with WLI, TXI showed improved visibility; TXI-1 82.6% (19/23), TXI-2 86.9% (20/23). In contrast, NBI showed decreased visibility; 78.2% (18/23). The intra-rater reliability for TXI-1 and TXI-2 were “almost perfect” for trainees and experts. TheΔE* values were 12.0 for WLI, 22.3 for TXI-1, and 25.9 for TXI-2 (WLI vs. TXI-1: p< 0.01, WLI vs. TXI-2: p< 0.01). In summary, improved visibility of the gray color sign in PPI-FGP was achieved for TXI-1 and TXI-2 compared to WLI for both trainees and experts and inter-rater reliability between trainees and experts was good. The ΔE* values were significantly larger in TXI-1 and TXI-2 compared to WLI.

Gray color sign		All endoscopists　(N:10)	Trainees　(N:5)	Experts　(N:5)	Trainees vs.Experts (P value)
TXI	Mode 1	44.9 ± 4.3	22.7 ± 2.1	22.2 ± 2.3	0.50
	Mode 2	42.9 ± 4.6	21.8 ± 2.4	21.1 ± 2.3	0.32
ICC（2.1）	Mode 1	0.931	0.858	0.864
	Mode 2	0.927	0.889	0.817

Conclusion

The usefulness of TXI compared with WLI led to the improved visibility of the gray color sign in PPI-FGP for both trainees and experts when evaluated subjectively and objectively.

References

1. Sato T et al. Journal of Healthcare Engineering Volume 2021.

Introduction

The depth of early gastric cancer is an important factor in determining the treatment plan such as endoscopic resection or surgery. We developed an artificial intelligence (AI) program using deep neural networks to diagnose mucosal and submucosal cancers using endoscopic images. Performance evaluation tests were conducted for clinical applications.

Aims & Methods

A validation dataset was established comprising 204 patients with early gastric cancer who underwent endoscopic resection or surgery at our hospital during 2018–2021. The lesion sites were the upper part in 16 (8%), middle part in 106 (52%), and lower part in 82 (40%) patients. The median lesion size (range) was 14 (2–90) mm; the lesion depth was mucosal cancer in 136 (67%) and submucosal cancer in 68 (33%) patients. Five white-light images per case at a minimum were used (median: 8, range: 5–22). The primary endpoint was sensitivity of mucosal cancer, defined as the lower limit of the 95% confidence interval (CI) > 75%, based on previous reports on the depth diagnosis of early gastric cancer. The secondary endpoints were specificity and accuracy.

Results

Sensitivity of this program for the diagnosis of mucosal cancer was 91.2% (95% CI: 85.2%–94.9%), indicating the usefulness of this program. Specificity was 66.2% (95% CI: 54.3%–76.3%), and accuracy was 82.8% (95% CI: 77.1%–87.4%).

Conclusion

The AI program developed for diagnosing the depth of early gastric cancer is useful and expected to be applied clinically in the near future.

Introduction

Early detection of gastric neoplasms remains challenging, even with enhanced endoscopic imaging. Photodynamic endoscopic imaging (PEI) with 5-aminolevulinic acid (5-ALA) is a novel diagnostic approach that exploits tumor-selective fluorescence. After oral administration, 5-ALA is converted via the heme biosynthesis pathway to protoporphyrin IX (PpIX), which accumulates in neoplastic cells and emits red fluorescence under violet-blue excitation. We developed specialized endoscope prototypes (first-generation prototype using a 410 nm laser, second-generation prototype with an enhanced 410 nm laser, and third-generation prototype employing a high-intensity LED light source) to visualize PpIX fluorescence in the stomach. These successive devices improved image brightness and enabled more objective tumor detection.

Aims & Methods

In this exploratory study at Tottori University Hospital, 54 patients with 60 early-stage gastric lesions (gastric adenomas or early gastric carcinomas) were examined. Patients received oral 5-ALA (20 mg/kg) three hours before endoscopy. PEI was performed using the prototype systems (excitation at 410 nm and detection of ~635 nm red fluorescence). Lesions were identified by their red fluorescent contrast against surrounding mucosa. After the procedure, patients were protected from strong light for 24 hours to prevent photosensitivity. Key outcomes were the fluorescence detection rate of known tumors, discovery of any additional lesions, and safety (adverse events).

Results

The overall fluorescence detection rate was 85.0% (51/60 lesions). PEI newly identified 11 synchronous lesions in 8 patients (14.8%). Among lesions with fluorescence, histological types included tubular adenocarcinoma and adenoma (tub/sig/adenoma: 49/0/13). All signet ring cell carcinomas (5/5) were fluorescence-negative. Among 12 post-H. pylori eradication gastric cancers, 11 (91.6%) exhibited detectable fluorescence. Adverse events included transient elevations in liver enzymes (9 cases), mild hypotension (1 case), and low-grade fever (1 case), all resolving spontaneously. No phototoxic reactions were observed under light-shielding protocols.

Conclusion

Oral 5-ALA-mediated photodynamic endoscopic imaging is a feasible and effective technique for objective visualization of early-stage gastric tumors. This method provides bright tumor-specific fluorescence in gastric adenomas and differentiated early cancers, enhancing lesion contrast and aiding detection. Although diffuse-type (signet ring cell) carcinoma may evade fluorescence due to limited PpIX accumulation, the overall ability of 5-ALA-PEI to highlight tumors could significantly augment current endoscopic screening practices. This emerging molecular imaging approach offers endoscopists a valuable tool for identifying gastric neoplasms, and ongoing device refinements may further improve sensitivity for subtle lesions.

Disclosure

Hiroki Kurumi and Yuichiro Ikebuchi declare no conflict of interest. Hajime Isomoto received research funding from FUJIFILM Corporation.