Artificial intelligence-assisted detection and classification of colorectal polyps under colonoscopy: a systematic review and meta-analysis

Wang, Aling; Mo, Jiahao; Zhong, Cailing; Wu, Shaohua; Wei, Sufen; Tu, Binqi; Liu, Chang; Chen, Daman; Xu, Qianqian; Cai, Mengyi; Li, Zhuoyao; Xie, Wenting; Xie, Miao; Kato, Motohiko; Xi, Xujie; Zhang, Beiping

doi:10.21037/atm-21-5081

Cited by 14 publications

(5 citation statements)

References 46 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Recently, research and development of AI for the detection and diagnosis of colorectal tumors have been actively conducted and are commercially available in some countries [16][17][18][19][20][21][22]. Approximately a quarter of the respondents replied that it is used based on necessity in clinical practice.…”

Section: Discussionmentioning

confidence: 99%

Current Status of Diagnosis and Treatment of Colorectal Cancer in Asian Countries: A Questionnaire Survey

et al. 2023

View full text Add to dashboard Cite

Introduction: Diagnostic and therapeutic methods for colorectal cancer (CRC) have advanced; however, they may be inaccessible worldwide, and their widespread use is challenging. This questionnaire survey investigates the current status of diagnosis and treatment of early-stage CRC in Asian countries. Methods: Responses to the questionnaire were obtained from 213 doctors at different institutions in 8 countries and regions. The questionnaire consisted of 39 questions on the following four topics: noninvasive diagnosis other than endoscopy (6 questions), diagnosis by magnification and image-enhanced endoscopy (IEE) including artificial intelligence (AI) (10 questions), endoscopic submucosal dissection (ESD), proper use among other therapeutic methods (11 questions), and pathologic diagnosis and surveillance (12 questions). Results: Although 101 of 213 respondents were affiliated with academic hospitals, there were disparities among countries and regions in the dissemination of advanced technologies, such as IEE, AI, and ESD. The NICE classification is widely used for the diagnosis of colorectal tumors using IEE, while the JNET classification with magnification was used in countries such as Japan (65/70, 92.9%) and China (16/22, 72.7%). Of the 211 respondents, 208 (98.6%) assumed that en bloc resection should be achieved for carcinomas, and 180 of 212 (84.9%) believed that ESD was the most suitable in cases with a diameter larger than 2 cm. However, colorectal ESD is not widespread in countries such as Thailand, the Philippines, and Indonesia. Conclusion: The promotion of advanced technologies and education should be continual to enable more people to benefit from them.

show abstract

Section: Discussionmentioning

confidence: 99%

Current Status of Diagnosis and Treatment of Colorectal Cancer in Asian Countries: A Questionnaire Survey

et al. 2023

View full text Add to dashboard Cite

show abstract

“…analysis of these studies was recently published [10]. They reported that the ADR associated with CADe had 95% sensitivity and good discrimination ability.…”

Section: Accepted Manuscriptmentioning

confidence: 99%

“…In recent years, the usefulness of artificial intelligence (AI) has been reported in solving complex multinomial problems in various fields, including computer-aided detection (CADe) in medicine. Large prospective trials have been conducted on colonoscopy AI, and a meta-analysis of these studies was recently published 10 . The authors reported that the ADR associated with CADe had 95% sensitivity and good discrimination ability 10 .…”

Section: Introductionmentioning

confidence: 99%

“…Large prospective trials have been conducted on colonoscopy AI, and a meta-analysis of these studies was recently published 10 . The authors reported that the ADR associated with CADe had 95% sensitivity and good discrimination ability 10 . Other meta-analyses of randomized controlled trials (RCTs) also reported that the ADR was significantly higher with the use of CADe 11 12 13 .…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Linked-color imaging with or without artificial intelligence for adenoma detection: a randomized trial

Miyaguchi,

Tsuzuki,

Hirooka

et al. 2024

Endoscopy

View full text Add to dashboard Cite

Background and study aims: The adenoma detection rate (ADR) is an important indicator of colonoscopy quality and colorectal cancer incidence. Both linked color imaging (LCI) with artificial intelligence-assisted colonoscopy (LCA) and LCI alone (LC) increase adenoma detection, although it remains unclear whether one modality is superior. This study aimed to compare the ADR between LCA and LC. Furthermore, we evaluated differences in ADRs according to endoscopists’ experience (experts and trainees) and polyp size between the LCA and LC groups. Patients and methods: Patients undergoing colonoscopy for positive fecal immunochemical tests, follow-up of colon polyps, and abdominal symptoms at a single institution were randomly assigned to the LCA or LC group. The ADR, adenoma per colonoscopy (APC), cecal intubation time, withdrawal time, number of adenomas per location, and adenoma size were compared. Results: The LCA (n=400) and LC (n=400) groups showed comparable cecal intubation and withdrawal times. The LCA group showed a significantly higher ADR (58.8% vs 43.5%, p<0.001) and mean APC (1.31±1.65 vs. 0.94±1.42, p<0.001), particularly in the ascending colon (0.30±0.62 vs. 0.20±0.54, p=0.016). The total number of non-polypoid-type adenomas was also significantly higher in the LCA group (0.15±0.51 vs. 0.08±0.30, p=0.019). Small polyps (≤5 and 6–9 mm) were detected significantly more frequently in the LCA group (0.75±1.13 vs. 0.48±0.89, p<0.001 and 0.34±0.74 vs. 0.24±0.58, p=0.035). In both the groups, there was no significant difference in the ADR between experts and trainees. Conclusions: LCA is significantly superior to LC in terms of the ADR.

show abstract

“…8 In order to address this issue and reduce EDMR, computer-aided diagnostic methods employing artificial intelligence algorithms have increasingly been used for improving the detection and characterization of cancer polyps. [9][10][11][12][13][14][15][16][17][18] Algorithms such as Support Vector Machines (SVMs), k-nearest neighbors (kNNs), ensemble methods and random forests, [9][10][11][12] and convolutional neural networks (CNNs) with various architectures such as Residual networks (ResNet), Densely connected CNNs (DenseNet), and AlexNet, [13][14][15][16][17][18] have shown promising results in this field. However, most such machine learning (ML) algorithms have limitations, most notably that they tend to perform poorly on smaller and unbalanced datasets, leading to overfitting, biases, and under-coverage of less represented classes.…”

Section: Introductionmentioning

confidence: 99%

Enhancing colorectal cancer diagnosis through generative models and vision-based tactile sensing: a Sim2Real study

Kapuria,

Ikoma,

Chinchali

et al. 2024

Medical Imaging 2024: Imaging Informatics for Healthcare, Research, and Applications

View full text Add to dashboard Cite

Towards the goal of developing an informed, intuitive, and generalized artificial intelligence model for the earlystage diagnosis of colorectal cancer (CRC), in this work, we present a generative model-based technique to improve the training and generalization performance of machine learning classification algorithms. Through this approach, we address the challenge of acquiring sizable and well-balanced datasets within the clinical domain. Our methodology involves training generative models on already available medical data, learning the latent representations, and finally generating new synthetic samples to be used for downstream tasks. We train dedicated UNet2D-based Denoising Diffusion Probabilistic Models (DDPMs) using our custom dataset, which consists of textural images captured by our novel Vision-based Tactile Sensor (VS-TS), called Hysense. These UNet2D DDPMs are employed to generate synthetic images for each potential class. To thoroughly study the effectiveness of using synthetic images during training, we compared the performance of multiple classification models, ranging from simple to state-of-the-art approaches, with our evaluation focusing solely on real images. Specifically for our dataset, we also extend the use of dedicated UNet2D DDPMs to generate synthetic images of not just possible classes, but also other features that may be present in the image, such as whole or partial contact of sensor with polyp phantoms. Through our experimental analyses, we demonstrated that the utilization of generative models to enrich existing datasets with synthetic images leads to improved classification performance and a reduction in model biases.

show abstract

Artificial intelligence-assisted detection and classification of colorectal polyps under colonoscopy: a systematic review and meta-analysis

Cited by 14 publications

References 46 publications

Current Status of Diagnosis and Treatment of Colorectal Cancer in Asian Countries: A Questionnaire Survey

Current Status of Diagnosis and Treatment of Colorectal Cancer in Asian Countries: A Questionnaire Survey

Linked-color imaging with or without artificial intelligence for adenoma detection: a randomized trial

Enhancing colorectal cancer diagnosis through generative models and vision-based tactile sensing: a Sim2Real study

Contact Info

Product

Resources

About