Artificial intelligence-assisted diagnosis of rectal neuroendocrine tumors during white-light endoscopy

doi:10.3748/wjg.v31.i48.112683

Advanced Search

BPG is committed to discovery and dissemination of knowledge

Home / Archive / Volume 31, Issue 48

This Article

Academic Content and Language Evaluation of This Article

CrossCheck and Google Search of This Article

Academic Rules and Norms of This Article

Supplementary Materials of This Article

Citation of this article

Corresponding Author of This Article

Research Domain of This Article

Article-Type of This Article

Open-Access Policy of This Article

Times Cited Counts in Google of This Article

Number of Hits and Downloads for This Article

Total Article Views (0)

All Articles published online

The chart showing PDF series, HTML series, Figures (1-4) series, Tables (1-7) series.

Item

Count

PDF

HTML

Figures (1-4)

Tables (1-7)

Sum=49

Featured Article

The chart showing Browse series, Download series.

Item

Count

Browse

Download

Sum=31

Publishing Process of This Article

Item

Count

Browse

Download

Sum=7

Dec 28, 2025 (publication date) through Dec 27, 2025

Times Cited of This Article

Times Cited (0)

Journal Information of This Article

Publication Name

World Journal of Gastroenterology

ISSN

1007-9327

Publisher of This Article

Baishideng Publishing Group Inc, 7041 Koll Center Parkway, Suite 160, Pleasanton, CA 94566, USA

Retrospective Cohort Study

World J Gastroenterol. Dec 28, 2025; 31(48): 112683
Published online Dec 28, 2025. doi: 10.3748/wjg.v31.i48.112683

Table 1 Hyperparameters used in the study

Hyperparameters	Value
Optimizer	Adam
Batch_size	32
Epochs	50
Image_size	640 × 640
Learning_rate	5e-4
Weight_decay	5e-4

Full Size Table

Table 2 Performance of the model in detecting rectal neuroendocrine tumors in training and validation sets

Dataset	Precision	Recall	F1
Training set	0.991	0.988	0.990
Validation set	0.910	0.831	0.868

Full Size Table

Table 3 Comparison of artificial intelligence and endoscopist performance in lesions detection

Group	PPV	Sensitivity	NPV	Specificity	Accuracy
Expert¹	0.981 (0.008)²	0.966 (0.004)	0.980 (0.002)	0.990 (0.005)	0.981 (0.001)
Senior¹	0.901 (0.021)^a	0.993 (0.000)^a	0.996 (0.000)^a	0.938 (0.014)^a	0.958 (0.010)
Novice¹	0.901 (0.009)^a	0.993 (0.000)^a	0.996 (0.000)^a	0.938 (0.006)^a	0.958 (0.004)^a
Mean-doctor	0.928 (0.043)^a	0.984 (0.014)^a	0.991 (0.008)^a	0.955 (0.028)^a	0.966 (0.013)^a
AI	0.995	0.961	0.979	0.997	0.984

¹Value is the mean of the values of the two endoscopists.

²Variance.

^aP < 0.05.

PPV: Positive predictive value; NPV: Negative predictive value; AI: Artificial intelligence.

Full Size Table

Table 4 Comparison of artificial intelligence and endoscopist performance in the detection of submucosal lesions

Group	PPV	Sensitivity	NPV	Specificity	Accuracy
Expert¹	0.909 (0.007)²	0.891 (0.002)^a	0.976 (0.001)	0.980 (0.001)^a	0.962 (0.001)^a
Senior¹	0.828 (0.030)	0.911 (0.015)	0.980 (0.004)	0.956 (0.009)	0.948 (0.011)
Novice¹	0.762 (0.014)^a	0.839 (0.002)^a	0.962 (0.001)^a	0.940 (0.002)^a	0.919 (0.001)^a
Mean-doctor	0.833 (0.068)^a	0.880 (0.034)	0.972 (0.008)	0.958 (0.018)^a	0.943 (0.020)^a
AI	0.939	0.909	0.979	0.987	0.971

¹Value is the mean of the values of the two endoscopists.

²Variance.

^aP < 0.05.

PPV: Positive predictive value; NPV: Negative predictive value; AI: Artificial intelligence.

Full Size Table

Table 5 Comparison of artificial intelligence and endoscopist performance in the diagnosis of rectal neuroendocrine tumors among the submucosal lesions

Group	PPV	NPV	Sensitivity	Specificity	Accuracy
Expert¹	0.855 (0.046)²	0.778 (0.047)	0.977 (0.006)	0.986 (0.008)	0.967 (0.001)^a
Senior¹	0.675 (0.001)^a	0.697 (0.071)	0.964 (0.008)	0.962 (0.001)^a	0.932 (0.009)
Novice¹	0.635 (0.071)	0.573 (0.016)^a	0.960 (0.001)^a	0.968 (0.012)	0.935 (0.012)
Mean-doctor	0.721 (0.111)¹	0.683 (0.100)	0.967 (0.009)¹	0.972 (0.013)¹	0.945 (0.018)¹
AI	0.940	0.797	0.981	0.994	0.977

¹Value is the mean of the values of the two endoscopists.

²Variance.

^aP < 0.05.

PPV: Positive predictive value; NPV: Negative predictive value; AI: Artificial intelligence.

Full Size Table

Table 6 Comparison of performance in the identification of rectal neuroendocrine tumors based on the image as the unit

Group	PPV	Sensitivity	NPV	Specificity	Accuracy
Expert¹	0.808 (0.006)²^,^a	0.755 (0.027)	0.982 (0.002)	0.987 (0.001)¹	0.972 (0.001)^a
Senior¹	0.473 (0.047)^a	0.778 (0.033)	0.983 (0.003)	0.938 (0.009)¹	0.927 (0.011)^a
Novice¹	0.487 (0.098)^a	0.392 (0.087)^a	0.956 (0.005)^a	0.968 (0.019)	0.929 (0.011)^a
Mean-doctor	0.589 (0.176)^a	0.642 (0.199)	0.974 (0.014)	0.964 (0.024)^a	0.942 (0.024)^a
AI	0.939	0.726	0.981	0.997	0.978

¹Value is the mean of the values of the two endoscopists.

²Variance.

^aP < 0.05.

PPV: Positive predictive value; NPV: Negative predictive value; AI: Artificial intelligence.

Full Size Table

Table 7 Comparison of different methods in identifying rectal neuroendocrine tumors from all images based on the lesion as the unit

Method	Precision	Recall	F1
YOLOv7	0.940	0.729	0.821
YOLOv5	0.927	0.720	0.810

Full Size Table

Citation: Liu K, Wang ZY, Yi LZ, Li F, He SH, Zhang XG, Lai CX, Li ZJ, Qiu L, Zhang RY, Wu W, Lin Y, Yang H, Liu GM, Guan QS, Zhao ZF, Cheng LM, Dai J, Bai Y, Xie F, Zhang MN, Chen SZ, Zhong XF. Artificial intelligence-assisted diagnosis of rectal neuroendocrine tumors during white-light endoscopy. World J Gastroenterol 2025; 31(48): 112683
URL: https://www.wjgnet.com/1007-9327/full/v31/i48/112683.htm
DOI: https://dx.doi.org/10.3748/wjg.v31.i48.112683

Liu K, Wang ZY, Yi LZ, Li F, He SH, Zhang XG, Lai CX, Li ZJ, Qiu L, Zhang RY, Wu W, Lin Y, Yang H, Liu GM, Guan QS, Zhao ZF, Cheng LM, Dai J, Bai Y, Xie F, Zhang MN, Chen SZ, Zhong XF. Artificial intelligence-assisted diagnosis of rectal neuroendocrine tumors during white-light endoscopy. World J Gastroenterol 2025; 31(48): 112683 [DOI: 10.3748/wjg.v31.i48.112683]

©2004-2025 Baishideng Publishing Group Inc. All rights reserved. 7041 Koll Center Parkway, Suite 160, Pleasanton, CA 94566, USA