The histologic features, molecular features, detection and management of serrated polyps: a review

Multiclassification of Colorectal Polyps from Colonoscopy Images Using AI for Early Diagnosis

Background/Objectives: Colorectal cancer (CRC) remains one of the leading causes of cancer-related mortality worldwide, emphasizing the critical need for the accurate classification of precancerous polyps. This research presents an extensive analysis of the multiclassification framework leveraging various deep learning (DL) architectures for the automated classification of colorectal polyps from colonoscopy images. Methods: The proposed methodology integrates real-time data for training and utilizes a publicly available dataset for testing, ensuring generalizability. The real-time images were cautiously annotated and verified by a panel of experts, including post-graduate medical doctors and gastroenterology specialists. The DL models were designed to categorize the preprocessed colonoscopy images into four clinically significant classes: hyperplastic, serrated, adenoma, and normal. A suite of state-of-the-art models, including VGG16, VGG19, ResNet50, DenseNet121, EfficientNetV2, InceptionNetV3, Vision Transformer (ViT), and the custom-developed CRP-ViT, were trained and rigorously evaluated for this task. Results: Notably, the CRP-ViT model exhibited superior capability in capturing intricate features, achieving an impressive accuracy of 97.28% during training and 96.02% during validation with real-time images. Furthermore, the model demonstrated remarkable performance during testing on the public dataset, attaining an accuracy of 95.69%. To facilitate real-time interaction and clinical applicability, a user-friendly interface was developed using Gradio, allowing healthcare professionals to upload colonoscopy images and receive instant classification results. Conclusions: The CRP-ViT model effectively predicts and categorizes colonoscopy images into clinically relevant classes, aiding gastroenterologists in decision-making. This study highlights the potential of integrating AI-driven models into routine clinical practice to improve colorectal cancer screening outcomes and reduce diagnostic variability.

Characteristics of Clinically Significant Hyperplastic Polyps: Distinctions Between Microvesicular and Goblet Cell-Rich Types

BACKGROUND

Clinically significant serrated polyps (CSSPs) are defined as sessile serrated lesions (SSLs), SSLs with dysplasia, traditional serrated adenomas (TSAs), hyperplastic polyps (HPs) ≥ 10 mm, and HPs ≥ 6 mm in the proximal colon. HPs are further classified as microvesicular HPs (MVHPs) and goblet cell-rich HPs (GCHPs). Among CSSPs, HPs were categorized into clinically significant MVHPs (CS-MVHPs) and clinically significant GCHPs (CS-GCHPs). This study compares the characteristics of CS-MVHPs, CS-GCHPs, and SSLs.

METHODS

This study included patients who underwent colonoscopy at the Toyoshima Endoscopy Clinic between March 2021 and April 2024. Lesions diagnosed as adenomas or CSSPs were removed. Age, sex, number of polyps, detection rate, and polyp size were compared among CS-MVHPs, CS-GCHPs, and SSLs.

RESULTS

In total, 14 065 patients were enrolled. The detection rates for CS-MVHPs, CS-GCHPs, and SSLs were 5.24%, 1.22%, and 6.36%, respectively. Patients with CS-MVHPs or SSLs were significantly younger and more often female than those with CS-GCHPs. The mean sizes of CS-MVHPs and SSLs were significantly larger than that of CS-GCHPs. The detection rate of CS-GCHPs increased with age, whereas the detection rates of CS-MVHPs and SSLs did not show a similar trend.

CONCLUSIONS

Compared with CS-GCHPs, CS-MVHPs were larger, more frequent, and more likely to be found in younger patients and females. The characteristics of CS-MVHPs are similar to those of SSLs, supporting the hypothesis that CS-MVHPs are precursors of SSLs.

Mucin 5AC as a Biomarker for Sessile Serrated Lesions: Results From a Systematic Review and Meta-Analysis

INTRODUCTION

Sessile serrated lesions (SSLs) are a class of colon polyps challenging to detect through current screening methods but highly associated with colon cancer. To improve detection, we sought a biomarker sensitive for SSLs. Recent endoscopic and histopathologic studies suggest that SSLs are associated with alterations in intestinal mucin expression, but the frequency with which this occurs is not known.

METHODS

We performed a meta-analysis of available pathologic studies comparing mucin expression on SSLs to normal colonic mucosa, tubular adenomas, villous adenomas, traditional serrated adenomas (TSAs), and hyperplastic polyps (HPs). We searched Medline, Pubmed, and Embase and found 440 publications in this topic, and 18 total studies met inclusion.

RESULTS

We found that MUC5AC expression was more common in SSLs compared to normal colonic mucosa (OR = 82.9, P < 0.01), tubular adenoma (OR = 11, P < 0.01), and TSAs (OR = 3.6, P = 0.04). We found no difference in MUC5AC expression between SSLs versus HPs (OR = 2.1, P = 0.09) and no difference in MUC5AC expression between left colon and right colon HPs, with an OR = 1.8, P = 0.23.

DISCUSSION

We found that MUC5AC expression was found commonly on villous adenoma, SSLs, and TSAs while the frequency on colon cancers declined. MUC5AC is also upregulated in inflammatory bowel disease and in response to intestinal infections. MUC5AC expression highlights the potential of mucins as useful biomarkers, though not specific to SSLs. Further research into the clinical utility of MUC5AC as a pathologic or fecal biomarker could enhance SSL detection.

Clinical and endoscopic characteristics of colorectal sessile serrated lesions with or without dysplasia/carcinoma: A systematic review and meta-analysis

OBJECTIVE

We aimed to compare the clinical and endoscopic characteristics of sessile serrated lesions (SSLs) with dysplasia/carcinoma (SSLD/Cs) and SSLs without dysplasia in this systematic review and meta-analysis.

METHODS

MEDLINE, EMBASE, and Cochrane Library databases and Clinicaltrials.gov were searched for relevant studies published up to August 28, 2023. The primary outcome was lesion size in SSLD/Cs and SSLs without dysplasia. The secondary outcomes included risk of dysplasia/carcinoma, morphology (classified based on the Paris classification), and lesion features such as mucus cap and nodules/protrusions in the two groups.

RESULTS

Thirteen studies with 14 381 patients were included. The proportion of SSLD/Cs ≥10 mm was significantly higher than that of SSLs without dysplasia (odds ratio [OR] 3.82, 95% confidence interval [CI] 1.21-12.02, p = 0.02). There was no significant difference in the risk of dysplasia/carcinoma between the proximal (OR 0.80, 95% CI 0.57-1.14) and distal colon (OR 1.25, 95% CI 0.88-1.77, p = 0.21). The 0-Ip (OR 2.47, 95% CI 1.50-4.09) and 0-IIa + Is (OR 10.38, 95% CI 3.08-34.98) morphologies were more prevalent among SSLD/Cs, whereas the 0-IIa morphology (OR 0.38, 95% CI 0.22-0.65) was more prevalent among SSLs without dysplasia (all p < 0.001). Furthermore, mucus cap (OR 0.61, 95% CI 0.42-0.89, p = 0.01) was more common among SSLs without dysplasia, whereas nodules/protrusions (OR 7.80, 95% CI 3.07-19.85, p < 0.001) were more common in SSLD/Cs.

CONCLUSION

SSLs >10 mm, 0-Ip or 0-IIa + Is morphologies, and those with nodules/protrusions are significantly associated with dysplasia/carcinoma.