Establishment and validation of an artificial intelligence-based model for real-time detection and classification of colorectal adenoma

Colorectal cancer (CRC) prevention requires early detection and removal of adenomas. We aimed to develop a computational model for real-time detection and classification of colorectal adenoma. Computationally constrained background based on real-time detection, we propose an improved adaptive lightweight ensemble model for real-time detection and classification of adenomas and other polyps. Firstly, we devised an adaptive lightweight network modification and effective training strategy to diminish the computational requirements for real-time detection. Secondly, by integrating the adaptive lightweight YOLOv4 with the single shot multibox detector network, we established the adaptive small object detection ensemble (ASODE) model, which enhances the precision of detecting target polyps without significantly increasing the model's memory footprint. We conducted simulated training using clinical colonoscopy images and videos to validate the method's performance, extracting features from 1148 polyps and employing a confidence threshold of 0.5 to filter out low-confidence sample predictions. Finally, compared to state-of-the-art models, our ASODE model demonstrated superior performance. In the test set, the sensitivity of images and videos reached 87.96% and 92.31%, respectively. Additionally, the ASODE model achieved an accuracy of 92.70% for adenoma detection with a false positive rate of 8.18%. Training results indicate the effectiveness of our method in classifying small polyps. Our model exhibits remarkable performance in real-time detection of colorectal adenomas, serving as a reliable tool for assisting endoscopists.

Evaluation of an Artificial Intelligence-Augmented Digital System for Histologic Classification of Colorectal Polyps

Importance

Colorectal polyps are common, and their histopathologic classification is used in the planning of follow-up surveillance. Substantial variation has been observed in pathologists' classification of colorectal polyps, and improved assessment by pathologists may be associated with reduced subsequent underuse and overuse of colonoscopy.

Objective

To compare standard microscopic assessment with an artificial intelligence (AI)-augmented digital system that annotates regions of interest within digitized polyp tissue and predicts polyp type using a deep learning model to assist pathologists in colorectal polyp classification.

Design, Setting, and Participants

In this diagnostic study conducted at a tertiary academic medical center and a community hospital in New Hampshire, 100 slides with colorectal polyp samples were read by 15 pathologists using a microscope and an AI-augmented digital system, with a washout period of at least 12 weeks between use of each modality. The study was conducted from February 10 to July 10, 2020.

Main Outcomes and Measures

Accuracy and time of evaluation were used to compare pathologists' performance when a microscope was used with their performance when the AI-augmented digital system was used. Outcomes were compared using paired t tests and mixed-effects models.

Results

In assessments of 100 slides with colorectal polyp specimens, use of the AI-augmented digital system significantly improved pathologists' classification accuracy compared with microscopic assessment from 73.9% (95% CI, 71.7%-76.2%) to 80.8% (95% CI, 78.8%-82.8%) (P < .001). The overall difference in the evaluation time per slide between the digital system (mean, 21.7 seconds; 95% CI, 20.8-22.7 seconds) and microscopic examination (mean, 13.0 seconds; 95% CI, 12.4-13.5 seconds) was -8.8 seconds (95% CI, -9.8 to -7.7 seconds), but this difference decreased as pathologists became more familiar and experienced with the digital system; the difference between the time of evaluation on the last set of 20 slides for all pathologists when using the microscope and the digital system was 4.8 seconds (95% CI, 3.0-6.5 seconds).

Conclusions and Relevance

In this diagnostic study, an AI-augmented digital system significantly improved the accuracy of pathologic interpretation of colorectal polyps compared with microscopic assessment. If applied broadly to clinical practice, this tool may be associated with decreases in subsequent overuse and underuse of colonoscopy and thus with improved patient outcomes and reduced health care costs.

Colonoscopy polyp detection and classification: Dataset creation and comparative evaluations

Colorectal cancer (CRC) is one of the most common types of cancer with a high mortality rate. Colonoscopy is the preferred procedure for CRC screening and has proven to be effective in reducing CRC mortality. Thus, a reliable computer-aided polyp detection and classification system can significantly increase the effectiveness of colonoscopy. In this paper, we create an endoscopic dataset collected from various sources and annotate the ground truth of polyp location and classification results with the help of experienced gastroenterologists. The dataset can serve as a benchmark platform to train and evaluate the machine learning models for polyp classification. We have also compared the performance of eight state-of-the-art deep learning-based object detection models. The results demonstrate that deep CNN models are promising in CRC screening. This work can serve as a baseline for future research in polyp detection and classification.

A comparative study on polyp classification using convolutional neural networks

Colorectal cancer is the third most common cancer diagnosed in both men and women in the United States. Most colorectal cancers start as a growth on the inner lining of the colon or rectum, called 'polyp'. Not all polyps are cancerous, but some can develop into cancer. Early detection and recognition of the type of polyps is critical to prevent cancer and change outcomes. However, visual classification of polyps is challenging due to varying illumination conditions of endoscopy, variant texture, appearance, and overlapping morphology between polyps. More importantly, evaluation of polyp patterns by gastroenterologists is subjective leading to a poor agreement among observers. Deep convolutional neural networks have proven very successful in object classification across various object categories. In this work, we compare the performance of the state-of-the-art general object classification models for polyp classification. We trained a total of six CNN models end-to-end using a dataset of 157 video sequences composed of two types of polyps: hyperplastic and adenomatous. Our results demonstrate that the state-of-the-art CNN models can successfully classify polyps with an accuracy comparable or better than reported among gastroenterologists. The results of this study can guide future research in polyp classification.

A Web-Based Education Program for Colorectal Lesion Diagnosis with Narrow Band Imaging Classification

BACKGROUND

An education system for narrow band imaging (NBI) interpretation requires sufficient exposure to key features. However, access to didactic lectures by experienced teachers is limited in the United States.

AIM

To develop and assess the effectiveness of a colorectal lesion identification tutorial.

METHODS

In the image analysis pretest, subjects including 9 experts and 8 trainees interpreted 50 white light (WL) and 50 NBI images of colorectal lesions. Results were not reviewed with subjects. Trainees then participated in an online tutorial emphasizing NBI interpretation in colorectal lesion analysis. A post-test was administered and diagnostic yields were compared to pre-education diagnostic yields.

RESULTS

Under the NBI mode, experts showed higher diagnostic yields (sensitivity 91.5% [87.3-94.4], specificity 90.6% [85.1-94.2], and accuracy 91.1% [88.5-93.7] with substantial interobserver agreement [κ value 0.71]) compared to trainees (sensitivity 89.6% [84.8-93.0], specificity 80.6% [73.5-86.3], and accuracy 86.0% [82.6-89.2], with substantial interobserver agreement [κ value 0.69]). The online tutorial improved the diagnostic yields of trainees to the equivalent level of experts (sensitivity 94.1% [90.0-96.6], specificity 89.0% [83.0-93.2], and accuracy 92.0% [89.3-94.7], p < 0.001 with substantial interobserver agreement [κ value 0.78]).

CONCLUSIONS

This short, online tutorial improved diagnostic performance and interobserver agreement.

[Management of unexpected cancer in locally resected colorectal polyps]

Excised colorectal polyps may contain unexpected malignancy. Once diagnosed histologically, further management is controversial, and little evidence exists to help in the choice between salvage bowel resection, local resection, systemic oncological treatment or watchful waiting. Choice of treatment must take patient-related factors as age and co-morbidity into consideration. Clinicians often face a dilemma concerning further treatment because of the scarce evidence. We want to establish a suggestion of a tailor-made treatment of malignant colorectal polyps based on the given evidence.

Automatic Detection and Classification of Colorectal Polyps by Transferring Low-Level CNN Features From Nonmedical Domain

Colorectal cancer (CRC) is a leading cause of cancer deaths worldwide. Although polypectomy at early stage reduces CRC incidence, 90% of the polyps are small and diminutive, where removal of them poses risks to patients that may outweigh the benefits. Correctly detecting and predicting polyp type during colonoscopy allows endoscopists to resect and discard the tissue without submitting it for histology, saving time, and costs. Nevertheless, human visual observation of early stage polyps varies. Therefore, this paper aims at developing a fully automatic algorithm to detect and classify hyperplastic and adenomatous colorectal polyps. Adenomatous polyps should be removed, whereas distal diminutive hyperplastic polyps are considered clinically insignificant and may be left in situ . A novel transfer learning application is proposed utilizing features learned from big nonmedical datasets with 1.4-2.5 million images using deep convolutional neural network. The endoscopic images we collected for experiment were taken under random lighting conditions, zooming and optical magnification, including 1104 endoscopic nonpolyp images taken under both white-light and narrowband imaging (NBI) endoscopy and 826 NBI endoscopic polyp images, of which 263 images were hyperplasia and 563 were adenoma as confirmed by histology. The proposed method identified polyp images from nonpolyp images in the beginning followed by predicting the polyp histology. When compared with visual inspection by endoscopists, the results of this study show that the proposed method has similar precision (87.3% versus 86.4%) but a higher recall rate (87.6% versus 77.0%) and a higher accuracy (85.9% versus 74.3%). In conclusion, automatic algorithms can assist endoscopists in identifying polyps that are adenomatous but have been incorrectly judged as hyperplasia and, therefore, enable timely resection of these polyps at an early stage before they develop into invasive cancer.

Directional wavelet based features for colonic polyp classification

In this work, various wavelet based methods like the discrete wavelet transform, the dual-tree complex wavelet transform, the Gabor wavelet transform, curvelets, contourlets and shearlets are applied for the automated classification of colonic polyps. The methods are tested on 8 HD-endoscopic image databases, where each database is acquired using different imaging modalities (Pentax's i-Scan technology combined with or without staining the mucosa), 2 NBI high-magnification databases and one database with chromoscopy high-magnification images. To evaluate the suitability of the wavelet based methods with respect to the classification of colonic polyps, the classification performances of 3 wavelet transforms and the more recent curvelets, contourlets and shearlets are compared using a common framework. Wavelet transforms were already often and successfully applied to the classification of colonic polyps, whereas curvelets, contourlets and shearlets have not been used for this purpose so far. We apply different feature extraction techniques to extract the information of the subbands of the wavelet based methods. Most of the in total 25 approaches were already published in different texture classification contexts. Thus, the aim is also to assess and compare their classification performance using a common framework. Three of the 25 approaches are novel. These three approaches extract Weibull features from the subbands of curvelets, contourlets and shearlets. Additionally, 5 state-of-the-art non wavelet based methods are applied to our databases so that we can compare their results with those of the wavelet based methods. It turned out that extracting Weibull distribution parameters from the subband coefficients generally leads to high classification results, especially for the dual-tree complex wavelet transform, the Gabor wavelet transform and the Shearlet transform. These three wavelet based transforms in combination with Weibull features even outperform the state-of-the-art methods on most of the databases. We will also show that the Weibull distribution is better suited to model the subband coefficient distribution than other commonly used probability distributions like the Gaussian distribution and the generalized Gaussian distribution. So this work gives a reasonable summary of wavelet based methods for colonic polyp classification and the huge amount of endoscopic polyp databases used for our experiments assures a high significance of the achieved results.

The frequency of early colorectal cancer derived from sessile serrated adenoma/polyps among 1858 serrated polyps from a single institution

BACKGROUND AND AIM

Sessile serrated adenoma/polyps (SSAPs) are suspected to have a high malignant potential, although few reports have evaluated the incidence of carcinomas derived from SSAPs using the new classification for serrated polyps (SPs). The aim of study was to compare the frequency of cancer coexisting with the various SP subtypes including mixed polyps (MIXs) and conventional adenomas (CADs).

METHODS

A total of 18,667 CADs were identified between April 2005 and December 2011, and 1858 SPs (re-classified as SSAP, hyperplastic polyp (HP), traditional serrated adenoma (TSA), or MIX) were removed via snare polypectomy, endoscopic mucosal resection, or endoscopic sub-mucosal dissection.

RESULTS

Among 1160 HP lesions, 1 (0.1%) coexisting sub-mucosal invasive carcinoma (T1) was detected. Among 430 SSAP lesions, 3 (0.7%) high-grade dysplasia (HGD/Tis) and 1 (0.2%) T1 were detected. All of the lesions were detected in the proximal colon, with a mean tumor diameter of 18 mm (SD 9 mm). Among 212 TSA lesions, 3 (1%) HGD/Tis were detected but no T1 cancer. Among 56 MIX lesions, 9 (16%) HGD/Tis and 1 (2%) T1 cancers were detected, and among 18,677 CAD lesions, 964 (5%) HGD/Tis and 166 (1%) T1 cancers were identified.

CONCLUSIONS

Among the resected lesions that were detected during endoscopic examination, a smaller proportion (1%) of SSAPs harbored HGD or coexisting cancer, compared to CAD or MIX lesions. Therefore, more attention should be paid to accurately identifying lesions endoscopically for intentional resection and the surveillance of each SP subtype.

The malignant colonic polyp Review of biological, clinical parameters and treatment

We know the significance of adenomas about the risk of neoplastic transformation defined as adenoma-carcinoma sequence. Although the majority of adenomas removed are small, it is well recognized that the risk of malignant transformation increases with an increased adenoma size. The term "malignant polyp" refers to an adenoma that macroscopically appears benign, but in which there is an invasion of malignant neoplastic cells within the submucosa through the muscularis mucosae. Malignant Polyps are substantially adenocarcinomas at an early stage; it is estimated that they represent the 0.75-5.6% of all adenomas removed during endoscopic exams. The management of a malignant polyp, diagnosed after an endoscopic removal, is complicated because the presence of residual malignant cells is a possibility. Also the presence of regional lymph nodes metastasis is different in literature and related to different prognostic factors. In this review we will analyze the incidence, the most appropriate methods of diagnosis, the biological parameters that characterize the various classes of risk of malignant polyps, in order to choice a correct treatment. The goal should be the improvement of the survival rate, decreasing the likelihood of residual disease evaluating the risk of overtreatment.

KEY WORDS

Adenoma, Adenoma-carcinoma, "Malignant polyp".

Management of malignant colon polyps: Current status and controversies

Colon cancer remains a significant clinical problem worldwide and in the United States it is the third most common cancer diagnosed in men and women. It is generally accepted that most malignant neoplasms of the colon arise from precursor adenomatous polyps. This stepwise progression of normal epithelium to carcinoma, often with intervening dysplasia, occurs as a result of multiple sequential, genetic mutations-some are inherited while others are acquired. Malignant polyps are defined by the presence of cancer cells invading through the muscularis mucosa into the underlying submucosa (T1). They can appear benign endoscopically but the presence of malignant invasion histologically poses a difficult and often controversial clinical scenario. Emphasis should be initially focused on the endoscopic assessment of these lesions. Suitable polyps should be resected en-bloc, if possible, to facilitate thorough evaluation by pathology. In these cases, proper attention must be given to the risks of residual cancer in the bowel wall or in the surrounding lymph nodes. If resection is not feasible endoscopically, then these patients should be referred for surgical resection. This review will discuss the important prognostic features of malignant polyps that will most profoundly affect this risk profile. Additionally, we will discuss effective strategies for their overall management.

Kudo’s pit pattern classification for colorectal neoplasms: A meta-analysis

AIM: To analyze the current available evidence of Kudo’s pit pattern classification for diagnosing colorectal neoplasms.

METHODS: A search was performed on Pubmed/Embase to identify studies reporting the outcomes of the pit pattern classification in colorectal polyps. Retrieved records were evaluated and selected by two independent investigators. The number of patients, polyps and diagnostic performance of Kudo’s pit pattern classification were retrieved from suitable studies. Pooled sensitivities and specificities were calculated using fixed or random effect models according to their heterogeneity. Publication bias was evaluated using funnel plot, Egger’s test, and Begg’s test. Sensitivity analysis was performed by omitting one study at a time and selecting a subgroup consisting of 11 magnifying chromoendoscopy studies.

RESULTS: 20 eligible studies were included in which a total of 5111 colorectal lesions in 3418 patients were identified for the differentiation of neoplastic and non-neoplastic polyps. Pit pattern classification in all the studies of mucosal patterns with magnification resulted in a pooled sensitivity of 89.0% (95%CI: 85.2-91.9) and pooled specificity of 85.7% (95%CI: 81.3-89.2) and the area under the SROC curve was 0.9354. There was significant publication bias (P = 0.038 and 0.006 for sensitivity and specificity using Egger’s test, P = 0.035 and 0.139 for sensitivity and specificity using Begg’s test, respectively). No single study significantly affected the pooled result, and the magnifying chromoendoscopy subgroup yielded a sensitivity of 92.7% (95%CI: 89.2-95.2) and specificity of 87.3% (95%CI: 81.6-91.4).

CONCLUSION: Kudo’s pit pattern classification is an accurate diagnostic method for the differentiation of neoplastic colorectal lesions. Publication bias is significant in the current available literature.

Variation in the association between colorectal cancer susceptibility loci and colorectal polyps by polyp type

We conducted a case-control study of the association between subsets of colorectal polyps, including adenomas and serrated polyps, and single-nucleotide polymorphisms (SNPs) related to colorectal cancer through prior genome-wide association studies (GWAS). Participants were enrollees in the Group Health Cooperative (Seattle, Washington) aged 24-79 years who received a colonoscopy from 1998 to 2007, donated a buccal or blood sample, and completed a structured questionnaire. We performed genotyping of 13 colorectal cancer susceptibility SNPs. Polytomous logistic regression models were used to estimate odds ratios and 95% confidence intervals for associations between polyps and the colorectal cancer risk allele for each SNP under a log-additive model. Analyses included 781 controls, 489 cases with adenoma, 401 cases with serrated polyps, and 188 cases with both polyp types. The following SNPs were associated with advanced adenomas: rs10936599, rs10795668, rs16892766, and rs9929218 (P < 0.05). For nonadvanced adenomas and for serrated polyps overall, only rs961253 was statistically significant (P < 0.05). These associations were in the same directions as those in prior colorectal cancer GWAS. No SNP was significantly associated with hyperplastic polyps, and only rs6983267 was significantly associated with sessile serrated polyps, but this association was opposite of that found in colorectal cancer GWAS. Our results suggest that the association between colorectal cancer susceptibility SNPs and colorectal polyps varies by polyp type.

Hyperplastic polyp? Look again... the impact of the new classification for serrated polyps

INTRODUCTION

The World Health Organization reviewed the classification for serrated colonic polyps in 2010. A new entity, sessile serrated adenoma, was included with two variants: with and without cytological dysplasia. This lesion's malignant potential has been recognized and according to the new classification, many polyps may be reclassified. The impact of this change is yet to be assessed.

OBJECTIVE

Analyze the proportion of lesions that were reclassified according to the new World Health Organization classification and the variables that influenced it.

MATERIAL AND METHODS

Every patient with at least one sessile serrated adenoma diagnosed in a 5 year period was included. All polyps (regardless of type) resected during the study period were reviewed. Data concerning polyp's characteristics and patient variables were collected. Forty consecutive patients were included [13 female, mean age at 1st sessile serrated adenoma -59 yrs (34-80)].

RESULTS

Were reviewed 247 polyps: hyperplastic--42%; conventional adenomas--29%; sessile serrated adenoma--24%; serrated adenomas--5%. Sixty-three polyps were reclassified: 43 hyperplastic, 12 serrated adenomas, 7 sessile serrated adenoma and 1 conventional adenoma with low grade dysplasia. Reclassification was significantly greater for hyperplastic polyps when compared with the other subtypes. Forty-three of one hundred and four (41%) hyperplastic polyps were reclassified all as sessile serrated adenoma. In these polyps the probability of reclassification was independent from polyp location but was greater if polyp size ≥ 5 mm.

DISCUSSION

This is a single center, rectrospective study. The fact that it was done in an Oncology Referral Institution with a Family Risk Clinic may have influenced the results. Nevertheless the impressive reclassification rate for Hyperplastic Polyps and the fact that they were reclassified mainly as Serrated Adenomas makes these results relevant to daily practice.

CONCLUSION

Our results suggest that, according to the new World Health Organization classification for serrated colonic polyps, a considerable proportion of hyperplastic polyps will be reclassified. The serrated pathway of colorectal carcinogenesis has probably been underestimated and at-risk patients may have been under inappropriate surveillance.

Serrated pathway: alternative route to colorectal cancer

Serrated polyps have been an area of intense focus for gastroenterologists over the past several years. Contrary to what was thought before, a growing body of literature indicates that these polyps can be precursors of colorectal cancer (CRC). Most of these lesions, particularly those in the proximal colon, have so far been under-recognized and missed during colonoscopy, qualifying these lesions to be the main cause of interval cancers. It is estimated that 10%-20% of CRCs evolve through this alternative, serrated pathway, with a distinct genetic and epigenetic profile. Aberrant DNA methylation plays a central role in the development of this CRC subtype. This characteristic molecular background is reflected in a unique pathological and clinical manifestation different from cancers arising via the traditional pathway. In this review we would like to highlight morphological, molecular and clinical features of this emerging pathway that are essential for gastroenterologists and may influence their everyday practice.

Software for automated classification of probe-based confocal laser endomicroscopy videos of colorectal polyps

AIM

To support probe-based confocal laser endomicroscopy (pCLE) diagnosis by designing software for the automated classification of colonic polyps.

METHODS

Intravenous fluorescein pCLE imaging of colorectal lesions was performed on patients undergoing screening and surveillance colonoscopies, followed by polypectomies. All resected specimens were reviewed by a reference gastrointestinal pathologist blinded to pCLE information. Histopathology was used as the criterion standard for the differentiation between neoplastic and non-neoplastic lesions. The pCLE video sequences, recorded for each polyp, were analyzed off-line by 2 expert endoscopists who were blinded to the endoscopic characteristics and histopathology. These pCLE videos, along with their histopathology diagnosis, were used to train the automated classification software which is a content-based image retrieval technique followed by k-nearest neighbor classification. The performance of the off-line diagnosis of pCLE videos established by the 2 expert endoscopists was compared with that of automated pCLE software classification. All evaluations were performed using leave-one-patient-out cross-validation to avoid bias.

RESULTS

Colorectal lesions (135) were imaged in 71 patients. Based on histopathology, 93 of these 135 lesions were neoplastic and 42 were non-neoplastic. The study found no statistical significance for the difference between the performance of automated pCLE software classification (accuracy 89.6%, sensitivity 92.5%, specificity 83.3%, using leave-one-patient-out cross-validation) and the performance of the off-line diagnosis of pCLE videos established by the 2 expert endoscopists (accuracy 89.6%, sensitivity 91.4%, specificity 85.7%). There was very low power (< 6%) to detect the observed differences. The 95% confidence intervals for equivalence testing were: -0.073 to 0.073 for accuracy, -0.068 to 0.089 for sensitivity and -0.18 to 0.13 for specificity. The classification software proposed in this study is not a "black box" but an informative tool based on the query by example model that produces, as intermediate results, visually similar annotated videos that are directly interpretable by the endoscopist.

CONCLUSION

The proposed software for automated classification of pCLE videos of colonic polyps achieves high performance, comparable to that of off-line diagnosis of pCLE videos established by expert endoscopists.

Evaluation of magnifying colonoscopy in the diagnosis of serrated polyps

AIM

To elucidate the colonoscopic features of serrated lesions of the colorectum using magnifying colonoscopy.

METHODS

Broad division of serrated lesions of the colorectum into hyperplastic polyps (HPs), traditional serrated adenomas (TSAs), and sessile serrated adenomas/polyps (SSA/Ps) has been proposed on the basis of recent molecular biological studies. However, few reports have examined the colonoscopic features of these divisions, including magnified colonoscopic findings. This study examined 118 lesions excised in our hospital as suspected serrated lesions after magnified observation between January 2008 and September 2011. Patient characteristics (sex, age), conventional colonoscopic findings (location, size, morphology, color, mucin) and magnified colonoscopic findings (pit pattern diagnosis) were interpreted by five colonoscopists with experience in over 1000 colonoscopies, and were compared with histopathological diagnoses. The pit patterns were categorized according to Kudo's classification, but a more detailed investigation was also performed using the subclassification [type II-Open (type II-O), type II-Long (type II-L), or type IV-Serrated (type IV-S)] proposed by Kimura T and Yamano H.

RESULTS

Lesions comprised 23 HPs (23/118: 19.5%), 39 TSAs (39/118: 33.1%: with cancer in one case), 50 SSA/Ps (50/118: 42.4%: complicated with cancer in three cases), and six others (6/118: 5.1%). We excluded six others, including three regular adenomas, one hamartoma, one inflammatory polyp, and one juvenile polyp for further analysis. Conventional colonoscopy showed that SSA/Ps were characterized as larger in diameter than TSAs and HPs (SSA/P vs HP, 13.62 ± 8.62 mm vs 7.74 ± 3.24 mm, P < 0.001; SSA/Ps vs TSA, 13.62 ± 8.62 mm vs 9.89 ± 5.73 mm, P < 0.01); common in the right side of the colon [HPs, 30.4% (7/23): TSAs, 20.5% (8/39): SSA/P, 84.0% (42/50), P < 0.001]; flat-elevated lesion [HPs, 30.4% (7/23): TSAs, 5.1% (2/39): SSA/Ps, 90.0% (45/50), P < 0.001]; normal-colored or pale imucosa [HPs, 34.8% (8/23): TSAs, 10.3% (4/39): SSA/Ps, 80% (40/50), P < 0.001]; and with large amounts of mucin [HPs, 21.7% (5/23): TSAs, 17.9% (7/39): SSA/Ps, 72.0% (36/50), P < 0.001]. In magnified colonoscopic findings, 17 lesions showed either type II pit pattern alone or partial type II pit pattern as the basic architecture, with 14 HPs (14/17, 70.0%) and 3 SSA/Ps. Magnified colonoscopy showed the type II-O pit pattern as characteristic of SSA/Ps [sensitivity 83.7% (41/49), specificity 85.7% (54/63)]. Cancer was also present in three lesions, in all of which a type VI pit pattern was also present within the same lesion. There were four HPs and four TSAs each. The type IV-S pit pattern was characteristic of TSAs [sensitivity 96.7% (30/31), specificity 89.9% (72/81)]. Cancer was present in one lesion, in which a type VI pit pattern was also present within the same lesion. In our study, serrated lesions of the colorectum also possessed the features described in previous reports of conventional colonoscopic findings. The pit pattern diagnosis using magnifying colonoscopy, particularly magnified colonoscopic findings using subclassifications of surface architecture, reflected the pathological characteristics of SSA/Ps and TSAs, and will be useful for colonoscopic diagnosis.

CONCLUSION

We suggest that this system could be a good diagnostic tool for SSA/Ps using magnifying colonoscopy.

In vivo diagnosis of colonic precancer and cancer using near-infrared autofluorescence spectroscopy and biochemical modeling

The aim of this study is to evaluate the biochemical foundation and clinical capability of an image-guided near-infrared (NIR) autofluorescence (AF) spectroscopy technique for in vivo diagnosis of colonic malignancies during clinical colonoscopy. A novel endoscopic fiber-optic AF system was utilized for in vivo NIR AF measurements at 785 nm excitation. A total of 263 in vivo NIR AF spectra of colonic tissues were measured from 100 patients, in which 164 spectra were from benign tissue (116 normal and 48 hyperplastic polyps), 34 spectra were from precancer (adenomatous polyps), and 65 spectra were from cancer. The non-negativity constrained least squares minimization biochemical modeling was explored to estimate the biochemical compositions of colonic tissue using nine basis reference spectra from the representative biochemicals (i.e., collagen I, elastin, β-nicotinamide adenine dinucleotide, flavin adenine dinucleotide, L-tryptophan, hematoporphyrin, 4-pyridoxic acid, pyridoxal 5'-phosphate, and water) associated with structural or cellular metabolic progression in colonic precancer and cancer. High-quality in vivo NIR AF spectra in the spectral range of 810 to 1000 nm were acquired from colonic benign, precancerous, and cancerous mucosa under white-light reflectance endoscopic imaging guidance. Partial least squares discriminant analysis, together with the leave-one tissue site-out, cross validation on in vivo NIR AF spectra yields diagnostic sensitivities of 85.4%, 76.5%, and 84.6%, and specificities of 89.9%, 93.4%, and 91.4%, respectively, for classification of benign, precancer, and cancer in the colon. This work demonstrates that image-guided NIR AF spectroscopy in conjunction with biochemical modeling has promising potential for improving in vivo detection and diagnosis of colonic precancer and cancer during clinical colonoscopic screening.

Analysis of mtDNA sequence variants in colorectal adenomatous polyps

Colorectal tumors mostly arise from sporadic adenomatous polyps. Polyps are defined as a mass of cells that protrudes into the lumen of the colon. Adenomatous polyps are benign neoplasms that, by definition display some characteristics of dysplasia. It has been shown that polyps were benign tumors which may undergo malignant transformation. Adenomatous polyps have been classified into three histologic types; tubular, tubulovillous, and villous with increasing malignant potential. The ability to differentially diagnose these colorectal adenomatous polyps is important for therapeutic intervention. To date, little efforts have been directed to identifying genetic changes involved in adenomatous polyps. This study was designed to examine the relevance of mitochondrial genome alterations in the three adenomatous polyps. Using high resolution restriction endonucleases and PCR-based sequencing, fifty-seven primary fresh frozen tissues of adenomatous polyps (37 tumors and 20 matched surrounding normal tissues) obtained from the southern regional Cooperative Human Tissue Network (CHTN) and Grady Memorial Hospital at Atlanta were screened with three mtDNA regional primer pairs that spanned 5.9 kbp. Results from our data analyses revealed the presence of forty-four variants in some of these mitochondrial genes that the primers spanned; COX I, II, III, ATP 6, 8, CYT b, ND 5, 6 and tRNAs. Based on the MITODAT database as a sequence reference, 25 of the 44 (57%) variants observed were unreported. Notably, a heteroplasmic variant C8515G/T in the MT-ATP 8 gene and a germline variant 8327delA in the tRNAlys was observed in all the tissue samples of the three adenomatous polyps in comparison to the referenced database sequence. A germline variant G9055A in the MT-ATP 6 gene had a frequency of 100% (17/17) in tubular and 57% (13/23) in villous adenomas; no corresponding variant was in tubulovillous adenomas. Furthermore, A9006G variant at MT-ATP 6 gene was observed at frequency of 57% (13/23) in villous adenomas only. Interestingly, variants A9006G and G9055A were absent in the villous tissue samples that were clinicopathological designated as "polyvillous adenomas". Our current data provide a basis for continued investigation of certain mtDNA variants as predictors of the three adenomatous polyps in a larger number of clinicopathological specimens.

Pathological characterization of colorectal polyps in Ilorin, Nigeria

Colorectal polyps especially the adenomas are recognized precursors of colorectal carcinoma. Identification and removal of such polyps before malignant transformation could reduce the burden of colorectal carcinoma. To document the demography and the histopathological types ofcolorectal polyps received by the Department of Pathology of University of Ilorin Teaching Hospital over a period of thirty years. This is a retrospective review of all cases ofcolorectal polyps that were received, processed and had histological diagnosis in our centre between 1979 and 2008 using the request cards and hematoxylin and eosin stained slides. Forty-four cases of colorectal polyps were reviewed constituting 6.7 percent of all colorectal biopsies/resections received in the same period. The male to female ratio was 2.1:1. Seventeen (38.6%) were adenomas, 9 (20.5%) were juvenile polyps, 8 (18.2%) were inflammatory polyps, 4 cases were lipomatous polyps, 3 were leiomatous polyps, and one each of lymphoid polyp, hamartomatous polyp and neurofibromatous polyp. Of the adenomas, 11 (58.8%) were tubular, 5 (29.4%) were villous, 1 (5.9%) was tubulovillous, and one was a villous adenoma with a focus of malignant transformation. Adenomatous polyp is the commonest pathological type ofcolorectal polyps in our centre. This study therefore sets out to review the age and sex distribution, location and morphological characteristics of all cases of colorectal polyps in our centre over the study period.

Serrated polyps of the colon and rectum (hyperplastic polyps, sessile serrated adenomas, traditional serrated adenomas, and mixed polyps)-proposal for diagnostic criteria

Until recently, two major types of colorectal epithelial polyps were distinguished: the adenoma and the hyperplastic polyp. While adenomas-because of their cytological atypia-were recognized as the precursor lesions for colorectal carcinoma, hyperplastic polyps were perceived as harmless lesions without any potential for malignant progression mainly because hyperplastic polyps are missing cytological atypia. Meanwhile, it is recognized that the lesions, formerly classified as hyperplastic, represent a heterogeneous group of polyps with characteristic serrated morphology some of which exhibit a significant risk of neoplastic progression. These serrated lesions show characteristic epigenetic alterations not commonly seen in colorectal adenomas and progress to colorectal carcinoma via the so-called serrated pathway (CpG-island-methylation-phenotype pathway). This group of polyps is comprised not only of hyperplastic polyps, but also of sessile serrated adenomas, traditional serrated adenomas and mixed polyps, showing serrated and "classical" adenomatous features. Diagnostic criteria and nomenclature for these lesions are not uniform and, therefore, somewhat confusing. In a consensus conference of the Working Group of Gastroenterological Pathology of the German Society of Pathology, standardization of nomenclature and diagnostic criteria as well as recommendations for clinical management of these serrated polyps were formulated and are presented herein.

SVM based colon polyps classifier in a wireless active stereo endoscope

This work focuses on the recognition of three-dimensional colon polyps captured by an active stereo vision sensor. The detection algorithm consists of SVM classifier trained on robust feature descriptors. The study is related to Cyclope, this prototype sensor allows real time 3D object reconstruction and continues to be optimized technically to improve its classification task by differentiation between hyperplastic and adenomatous polyps. Experimental results were encouraging and show correct classification rate of approximately 97%. The work contains detailed statistics about the detection rate and the computing complexity. Inspired by intensity histogram, the work shows a new approach that extracts a set of features based on depth histogram and combines stereo measurement with SVM classifiers to correctly classify benign and malignant polyps.

Colorectal polyps: a clinical, endoscopic and pathologic study in Iranian children

OBJECTIVE

To study the clinical presentation, histology and colonoscopic features of lower gastrointestinal polyps in Iranian children.

MATERIAL AND METHODS

Medical reports of children with colorectal polyps were retrospectively reviewed from 1996 to 2005 at the Children's Medical Center Hospital, Iran. A total of 563 cases were studied. Data related to age, sex, family history, signs and symptoms, the size, location, polyp types and associated lesions were collected and analyzed.

RESULTS

The mean age of children was 5.66 +/- 2.88 years (range 2 months to 17 years), with a male-to-female ratio of 1.61:1.0. The highest incidence was between ages 2 and 10 years (85.1%). Rectal bleeding was the presenting symptom in 78.5% cases. The polyps were solitary in 94% of cases. A majority of polyps (86.3%) were juvenile and 86.7% located in the rectosigmoid area. Three percent of cases had a positive family history. One case of Turcot syndrome was also identified.

CONCLUSION

Juvenile polyps remain the most common polyps in Iranian children. Although the presence of a solitary polyp in the rectosigmoid colon is more prevalent, in a significant number of cases they are multiple and located in proximal parts. Polyps must be removed even when asymptomatic because of their probable neoplastic potential.

Is serum protein electrophoresis useful in separating the "high risk group" in patients with colonic polyps?

BACKGROUND

Adenomas have the highest potential or clinical value from among colonic polyps of developing into adenocarcinoma. The aims of this paper are: to establish criteria to identify the high risk group of patients in a group of patients with colonic polyps, to work out a simple scheme for follow-up care after endoscopic polypectomy, and to establish indications for surgery. The usefulness of determination of electrophoresis of serum proteins has been specially analysed to detect early development of malignant growths in patients with colonic polyps regarding alfa-1/alfa-2 and alfa/beta. 67 cases - 21 women, 46 men were tested. Follow-up endoscopy with the electrophoresis was performed after 6 weeks, 6 and 12 months after polypectomy. 97 polyps were resected with endoscopy in 67 patients. 38 patients (39.17%), those constituting the high risk group, were selected. Included were all polyps with grade II and III of cellular differentiation.

CONCLUSIONS

1) alfa-1/alfa-2 and alfa/beta is a helpful test in identifying the high risk group among patients with colonic polyps and it can be used as a screening test, 2) the determination of beta-2-macroglobuline is not useful in the diagnosis of this group of patients, 3) the electrophoresis of proteins should be the first test to perform on patients with colonic polyps. The relation of electrophoresis to endoscopic polypectomy aids evaluations of patients specially predisposed to malignant.

Transparent rendering of intraluminal contrast for 3D polyp visualization at CT colonography

We developed a classifier that permits transparent rendering of both tagging material and air to facilitate interpretation of tagged computed tomographic (CT) colonography. With this technique, a reader can simultaneously appreciate polyps on endoluminal views both covered with tagging material and against air, along with unmodified 2-dimensional CT images. Evaluated with 49 polyps from 26 patients (data from public National Library of Medicine, Health Insurance Portability and Accountability Act compliant), 3 readers were able to determine the presence/absence of polyps in tagged locations with equivalent accuracy compared with polyps in air. This method offers an alternative way to visualize tagged CT colonography.

Role of chromoendoscopy in colon cancer surveillance in inflammatory bowel disease

Inflammation in the intestine is a well-known risk factor for neoplastic changes in the mucosa. In fact, it has been shown that long-standing ulcerative colitis and colonic Crohn's disease have a significantly increased risk for developing colorectal cancer, although the estimates vary widely between studies. Conventional colonoscopy is effective in detecting polypoid changes in the mucosa. However, it is now generally accepted that neoplastic changes in colitis are frequently flat and depressed, which are easily missed by use of routine colonoscopy. The introduction of chromoendoscopy, especially in combination with magnifying endoscopy, has greatly advanced our means to detect and differentiate neoplastic lesions in the colorectum. Accumulating evidence-based data indicate that implementation of chromoendoscopy into colon cancer surveillance protocols for patients with inflammatory bowel disease is effective. However, the introduction of chromoendoscopy into surveillance programs requires meticulous training and further studies to compare the value of chromoendoscopy to newer endoscopic devices and techniques, such as narrow band imaging.

Short- and long-term outcomes of standardized EMR of nonpolypoid (flat and depressed) colorectal lesions > or = 1 cm (with video)

BACKGROUND

Nonpolypoid (flat and depressed) colorectal lesions are increasingly recognized. Their endoscopic removal requires specialized EMR techniques, which are more complex to perform. Outcomes data on EMR of nonpolypoid neoplasms in the United States is needed.

OBJECTIVE

To determine the safety and efficacy of EMR in the resection of nonpolypoid colorectal neoplasms > or = 1 cm.

DESIGN

Retrospective analysis.

SETTING

Veterans Affairs Palo Alto Health Care System.

PATIENTS

Over a 5-year period, patients who underwent EMR for nonpolypoid colorectal lesions > or = 1 cm.

INTERVENTION

A standardized approach that included lesion assessment, classification, inject-and-cut EMR technique, reassessment, and treatment of residual tissue.

MAIN OUTCOME MEASUREMENTS

Complete resection, bleeding, perforation, development of advanced cancer, and death.

RESULTS

A total of 100 patients (125 lesions: 117 flat and 8 depressed) met inclusion criteria. Mean size was 16.7 +/- 7 mm (range, 10-50 mm). Histology included 5 submucosal invasive cancers, 5 carcinomas in situ, and 91 adenomas. Thirty-eight patients (48 lesions) did not receive surveillance colonoscopy: 8 had surgery, 16 had hyperplastic pathology, and 14 did not undergo repeat examination. Surveillance colonoscopy was performed on 62 patients (77 lesions). Complete resection was achieved in 100% of these patients after 1 to 3 surveillance colonoscopies. All patients received follow-up (mean [standard deviation] = 4.5 +/- 1.4 years); none developed colorectal cancer or metastasis.

LIMITATIONS

Single endoscopist, retrospective study.

CONCLUSIONS

A standardized EMR (inject-and-cut) technique is a safe and curative treatment option in nonpolypoid colorectal neoplasms (> or = 1 cm) in the United States.

Recent advances in chromoscopic colonoscopy and endomicroscopy

The basic rationale for secondary prevention of colorectal carcinoma is by endoscopic polypectomy. New technologies in the form of high-magnification or "zoom" colonoscopy complemented by chromoscopic agents permit early detection of neoplastic colorectal lesions, particularly flat and depressed types. Detailed morphologic characteristics of the surface crypt or "pit pattern" can be obtained with these techniques, enabling an in vivo "optical biopsy" and staging tool. Establishing suitability for endoscopic resection or surgical excision can be enhanced using these techniques. Furthermore, chromoscopic colonoscopy may have a role in routine endoscopic colorectal cancer surveillance programs in patients at high risk for colorectal neoplasia, such as those with long-standing ulcerative colitis and familial colorectal cancer syndromes. This review summarizes recent data regarding the prevalence and histopathologic characteristics of flat and depressed colorectal lesions in Western cohorts and describes how their detection and management can be improved by chromoscopy and magnification technology. We outline these techniques from a clinical perspective and describe the basic principles of endoscopic mucosal resection.

Segmentation and size measurement of polyps in CT colonography

Virtual colonoscopy is a relatively new method for the detection of colonic polyps. Their size, which is measured from reformatted CT images, mainly determines diagnosis. We present an automatic method for measuring the polyp size. The method is based on a robust segmentation method that grows a surface patch over the entire polyp surface starting from a seed. Projection of the patch points along the polyp axis yields a 2D point set to which we fit an ellipse. The long axis of the ellipse denotes the size of the polyp. We evaluate our method by comparing the automated size measurement with those of two radiologists using scans of a colon phantom. We give data for inter-observer and intra-observer variability of radiologists and our method as well as the accuracy and precision.

MIB-1 and MCM-2 immunohistochemical analysis does not aid in identification of serrated colorectal polyps with abnormal proliferation

We investigated the staining characteristics of serrated polyps with abnormal proliferation (SPAP) using MIB-1 and MCM-2 to determine if they could provide assistance in delineating SPAPs from traditional hyperplastic polyps (HPs). Using published morphologic criteria we reviewed H&E slides of 107 polyps from 80 patients. Thirty-nine (36.4%) polyps met the criteria for SPAP Within a given region, polyps in the transverse colon had the largest percentage of SPAPs (50.0%) followed by the right colon (40.9%). The majority of SPAPs (82.1%) and HPs (72.1%) showed MIB-1 staining confined to the basal third of the crypts. The majority of SPAPs (59.0%) and HPs (52.9%) showed MCM-2 staining extending into the apical third of the crypts. We do not recommend MIB-1 or MCM-2 staining to differentiate SPAPs from conventional HPs, since staining characteristics are not significantly different between the 2 groups, and frequent variable crypt staining within a given polyp is difficult to interpret.

Sessile Serrated Adenoma: A Brief History and Current Status

Sessile serrated adenoma (SSA) is a polyp of the large intestine, which was first described in 1996. It presents as a solitary lesion or in a setting of a polyposis previously coined as "serrated adenomatous polyposis." The importance of correct recognition of this lesion and distinction from other serrated polyps of the colon is in that the newly described "serrated pathway" of colorectal carcinogenesis seems to apply mostly to SSA and not to other serrated polyps of the colon. This review describes the history of morphologic discovery of SSA and reports on current status of this lesion.

Serrated pathway and APC (conventional)-type colorectal polyps: molecular-morphologic correlations, genetic pathways, and implications for classification

This review addresses the genetic mutations and cell signaling pathway alterations in colorectal premalignant polyps, focusing on the link between molecular changes and morphologic features. Biallelic APC (adenomatous polyposis coli) mutations are directly responsible for the specific and characteristic cytologic features of dysplastic cells in conventional tubular adenomas. Sessile serrated adenomas (SSAs) are the precursor lesions of the serrated neoplasia pathway. The BRAF activating mutation and hypermethylation of SLC5A8, which mediates short chain fatty acid transport, may be the important events in the genesis of SSAs. Intracellular butyrate inhibits histone deacetylase, allowing histone hyperacetylation and, eventually, transcriptional activation of specific genes. Decreased p21(WAF1/CIP1) and activation of the mitogen-activated protein kinase pathway may be the key intermediary alterations. Progressive loss of cell cycle control and decreased and altered cytoplasmic differentiation produce the characteristic constellation of morphologic changes of SSAs and traditional serrated adenomas.

Serrated adenomas have a pattern of genetic alterations that distinguishes them from other colorectal polyps

BACKGROUND

Serrated adenomas are characterized by serrated crypts with dysplasia, and are distinguished from other polyps by their histology, but the genetic basis of serrated adenomas is unknown. We investigated genetic alterations in colorectal polyps to determine if a specific pattern were associated with serrated adenomas.

METHODS

Sixty-six small (<10 mm) colorectal polyps were studied, including 11 hyperplastic polyps, 27 serrated adenomas, 9 tubular adenomas, 6 tubulovillous adenomas, and 3 villous adenomas. Allelic imbalance and microsatellite instability were detected by analysis of microsatellites on 5q, 18q, 17p, 2p, and 3p; K-ras mutations were detected by oligonucleotide hybridization.

RESULTS

Each polyp subset had its own characteristic mutational signature. Allelic imbalance of 18q was significantly more common (P < 0.05), whereas allelic imbalance of 5q and K-ras mutations were significantly less common (P < 0.05) in serrated adenomas compared with other polyps. Allelic imbalance of 17p was not found in any polyp.

CONCLUSIONS

Serrated adenomas are significantly more likely to have allelic imbalance at 18q than other types of adenomas, and significantly less likely to have allelic imbalance at 5q or K-ras mutations. Serrated adenomas seem to evolve through a different genetic pathway than other types of polyps in the colon.

The Association Between Cigarette Smoking and Colorectal Polyp Recurrence (United States)

OBJECTIVE

Although evidence exists linking smoking to precancerous colorectal adenomatous polyps, few studies have examined the association between cigarette smoking and recurrence of colorectal polyps. This association was investigated prospectively with data from the Polyp Prevention Trial.

METHODS

Cigarette smoking data were collected through baseline interviews. The study was completed by 1872 men and women with presence of adenomas at baseline colonoscopy. Multiple logistic regression analysis was used to examine the association between cigarette smoking and polyp recurrence (adenomatous and hyperplastic) up to four years from baseline.

RESULTS

Adenoma recurrence was not related to cigarette smoking. Current smokers had increased odds of hyperplastic polyps at follow-up compared to never smokers (OR 2.88, 95% CI 2.06-4.01). Current smoking was associated with subsequent distal (OR 3.44, 95% CI 2.38-4.95) and rectal (OR 3.53, 95% CI 2.15-5.78) hyperplastic polyps, but not subsequent proximal hyperplastic polyps. Cigarette smoking was associated with subsequent multiple and small size (4 mm) hyperplastic polyps. Significant linear trends were observed between development of subsequent hyperplastic polyps and all smoking variables.

CONCLUSIONS

Although no association with recurrent adenomas was observed, cigarette smoking was significantly associated with hyperplastic polyp development, except for those in the proximal colon. This prospective study confirms that cigarette smoking has a significant effect on the development of hyperplastic colorectal polyps.

A retrospective study of colonic polyps in King Chulalongkorn Memorial Hospital

To evaluate and classify polyps from colon in Thai patients, the authors retrospectively analyzed the 776 polyps from 696 subjects in King Chulalongkorn Memorial Hospital during the past five-year period from 1999 to 2003. All colonic polyps were included in the study. There were 461 (59%) male and 315 (410%) female with the mean age of 51 years. Non-neoplastic and neoplastic polyps were documented 50% each. Hyperplastic polyp was the most frequent diagnosis (39%), followed by tubular adenoma (36%). According to neoplastic polyp, 8%, 3%, and 14% cases were identified as high-grade dysplastic change, intramucosal carcinoma, and invasive carcinoma, respectively.

Prevalence of advanced colonic polyps in asymptomatic Chinese

AIM: To investigate the prevalence of advanced polyps in asymptomatic Chinese and to determine the risk of proximal advanced colonic polyps in subjects with and without polyps in the distal colon.

METHODS: Data were collected prospectively during colonoscopic examinations performed in 5 973 subjects as part of health evaluation at our unit from December 1997 to December 2003. Polyps were considered advanced, if they were larger than 10 mm or were tubovillous, villous or malignant. Proximal colon was defined as the splenic flexure and more proximal portions of the colon.

RESULTS: Colon polyps were detected in 971 (16.3%) subjects (613 males and 358 females) with their mean age being 56.6 ± 10.7 years. Advanced polyps were noted in 199 (3.3%) individuals. Subjects were sub-classified according to the location of polyps into three groups: distal (569, 58.6%), proximal (284, 29.2%), and combined proximal and distal (118, 12.2%) groups. Subjects with advanced polyps in these three groups were 95 (9.8%), 56 (5.8%), and 48 (4.9%) respectively. In the 48 subjects with advanced combined polyps, 13 advanced polyps were distributed at the distal colon, 17 at the proximal colon, and 18 at both. Eighteen colon cancers including 12 at sigmoid and 6 at ascending colon were confirmed by final pathology. The relative risk for advanced proximal polyp according to distal findings was 3.1 (95%CI: 1.3-7.4) for hyperplastic polyp, 2.7 (95%CI: 1.4-5.3) for tubular polyp and 13.5 (95%CI: 5.1-35.4) for advanced polyp as compared to that for no polyp. However, 56 (28.2%) of 199 subjects with advanced polyps had no index polyps at the distal colon and might go undetected under sigmoidoscopic screening.

CONCLUSION: Although distal lesions can predict the risk of advanced proximal polyps, a substantial portion of Chinese with advanced proximal polyps is not associated with any distal sentinel lesions. These data have implications for screening policy of colon cancers in Taiwanese Chinese.

Support vector machines committee classification method for computer-aided polyp detection in CT colonography

RATIONALE AND OBJECTIVES

A new classification scheme for the computer-aided detection of colonic polyps in computed tomographic colonography is proposed.

MATERIALS AND METHODS

The scheme involves an ensemble of support vector machines (SVMs) for classification, a smoothed leave-one-out (SLOO) cross-validation method for obtaining error estimates, and use of a bootstrap aggregation method for training and model selection. Our use of an ensemble of SVM classifiers with bagging (bootstrap aggregation), built on different feature subsets, is intended to improve classification performance compared with single SVMs and reduce the number of false-positive detections. The bootstrap-based model-selection technique is used for tuning SVM parameters. In our first experiment, two independent data sets were used: the first, for feature and model selection, and the second, for testing to evaluate the generalizability of our model. In the second experiment, the test set that contained higher resolution data was used for training and testing (using the SLOO method) to compare SVM committee and single SVM performance.

RESULTS

The overall sensitivity on independent test set was 75%, with 1.5 false-positive detections/study, compared with 76%-78% sensitivity and 4.5 false-positive detections/study estimated using the SLOO method on the training set. The sensitivity of the SVM ensemble retrained on the former test set estimated using the SLOO method was 81%, which is 7%-10% greater than the sensitivity of a single SVM. The number of false-positive detections per study was 2.6, a 1.5 times reduction compared with a single SVM.

CONCLUSION

Training an SVM ensemble on one data set and testing it on the independent data has shown that the SVM committee classification method has good generalizability and achieves high sensitivity and a low false-positive rate. The model selection and improved error estimation method are effective for computer-aided polyp detection.

Colonic polyp segmentation in CT colonography-based on fuzzy clustering and deformable models

An automatic method to segment colonic polyps in computed tomography (CT) colonography is presented in this paper. The method is based on a combination of knowledge-guided intensity adjustment, fuzzy c-mean clustering, and deformable models. The computer segmentations were compared with manual segmentations to validate the accuracy of our method. An average 76.3% volume overlap percentage among 105 polyp detections was reported in the validation, which was very good considering the small polyp size. Several experiments were performed to investigate the intraoperator and interoperator repeatability of manual colonic polyp segmentation. The investigation demonstrated that the computer-human repeatability was as good as the interoperator repeatability. The polyp segmentation was also applied in computer-aided detection (CAD) to reduce the number of false positive (FP) detections and provide volumetric features for polyp classification. Our segmentation method was able to eliminate 30% of FP detections. The volumetric features computed from the segmentation can further reduce FP detections by 50% at 80% sensitivity.

Hyperplastic (serrated) polyps of the colorectum: relationship of CpG island methylator phenotype and K-ras mutation to location and histologic subtype

We investigated the frequency of promoter region CpG island methylation (CIM) of hMLH1, MGMT, MINT1, MINT2, and p16 and K-ras mutations in a total of 79 hyperplastic (serrated) polyps (HPs) from 75 patients and correlated the molecular profiles to polyp location in the colorectum, histologic variation, and other factors. Methylation-specific PCR (MS-PCR) was used to assay CIM status. HPs that showed CIM of one or more or two or more of the genes assayed were classified as CpG island methylator phenotype (CIMP) and CIMP-high (CIMP-H), respectively. PCR restriction fragment length polymorphism was used to assay K-ras codon 12 and 13 mutations. Logistic regression indicated a statistically significant trend for increasing odds for CIMP (P = 0.002) and CIMP-H (P < 0.001) according to proximity to the cecum or distance from the rectum. Conversely, K-ras codon 12 mutation was present in 13 of 40 (32.5%) distally located HPs compared with 2 of 39 (5.1%) proximal HPs (P = 0.006). Histologic subtype distribution varied by proximal and distal locations. Frequency of CIMP in serrated polyps with abnormal proliferation (SPAPs), differed significantly from goblet cell serrated polyps (GCSPs) (24 of 26, 92.3% vs. 6 of 13, 46.2%) (P = 0.003) and microvesicular serrated polyps (MVSPs) (26 of 38, 68.4%) (P = 0.03). Frequency of K-ras mutation in GCSPs (7 of 13, 54%) differed from that of MVSPs (6 of 38, 16%) (P = 0.01) and SPAPs (2 of 26, 8%) (P = 0.003). Location in the colorectum and histologic subtype were major determinants of the molecular profile of HPs. The molecular findings of CIMP and K-ras mutations appear to encompass most if not all HPs; CIMP profiles suggest that SPAP is the most advanced morphologic variant. We postulate that MVSP and GCSP may be precursor lesions that, if proximally located or larger, can progress to SPAP. Frequent K-ras mutations and infrequent CIMP distinguish the distal GCSP variant.

Computer-aided tumor detection in endoscopic video using color wavelet features

We present an approach to the detection of tumors in colonoscopic video. It is based on a new color feature extraction scheme to represent the different regions in the frame sequence. This scheme is built on the wavelet decomposition. The features named as color wavelet covariance (CWC) are based on the covariances of second-order textural measures and an optimum subset of them is proposed after the application of a selection algorithm. The proposed approach is supported by a linear discriminant analysis (LDA) procedure for the characterization of the image regions along the video frames. The whole methodology has been applied on real data sets of color colonoscopic videos. The performance in the detection of abnormal colonic regions corresponding to adenomatous polyps has been estimated high, reaching 97% specificity and 90% sensitivity.

Pit pattern analysis by magnifying chromoendoscopy for the diagnosis of colorectal polyps

BACKGROUND AND PURPOSE

The development of magnifying chromoendoscopy has facilitated the observation of mucosal pit patterns. This study investigated the value of this technology in predicting the histologic findings of colorectal lesions.

METHODS

A total of 954 colorectal polyps were included. After identifying the lesions at colonoscopy, 0.2% indigocarmine solution was sprayed and then the zoom apparatus was switched to make a magnified view of the stained crypt orifice at a maximum 100 times magnification. The observed pit patterns were classified into 6 categories (I, II, IIIL, IIIS, IV, and V) according to Kudo's classification. Type I and II were designated as non-neoplastic patterns whereas other types were neoplastic. Correlation of the pit pattern with the findings of histologic examinations of resected or biopsied polyps was performed.

RESULTS

There were 678 diminutive (</= 5 mm) polyps (71.1%) and 705 neoplastic polyps (73.9%), including 695 adenomas and 10 carcinomas. When comparing histologically confirmed neoplastic lesions to non-neoplastic lesions, prediction of neoplastic lesions by endoscopists based on magnifying chromoendoscopy had a sensitivity of 90.8%, a specificity of 72.7%, a positive predictive value of 90.4%, a negative predictive value of 73.6%, and an overall accuracy of 86.1%. The diagnostic accuracy for neoplastic lesions was not associated with polyp size and location but was related to the operator's experience.

CONCLUSIONS

Characteristic pit patterns obtained by magnifying chromoendoscopy provide useful clues for differentiation of adenomatous from non-adenomatous polyps. Used appropriately in experienced hands, this technique offers a valuable adjunct to standard colonoscopy in predicting the histologic characteristics of colorectal polyps.

Feature-guided analysis for reduction of false positives in CAD of polyps for computed tomographic colonography

We evaluated the effect of our novel technique of feature-guided analysis of polyps on the reduction of false-positive (FP) findings generated by our computer-aided diagnosis (CAD) scheme for the detection of polyps from computed tomography colonographic data sets. The detection performance obtained by use of feature-guided analysis in the segmentation and feature analysis of polyp candidates was compared with that obtained by use of our previously employed fuzzy clustering technique. We also evaluated the effect of a feature called modified gradient concentration (MGC) on the detection performance. A total of 144 data sets, representing prone and supine views of 72 patients that included 14 patients with 21 colorectal polyps 5-25 mm in diameter, were used in the evaluation. At a 100% by-patient (95% by-polyp) detection sensitivity, the FP rate of our CAD scheme with feature-guided analysis based on round-robin evaluation was 1.3 (1.5) FP detections per patient. This corresponds to a 70-75% reduction in the number of FPs obtained by use of fuzzy clustering at the same sensitivity levels. Application of the MGC feature instead of our previously used gradient concentration feature did not improve the detection result. The results indicate that feature-guided analysis is useful for achieving high sensitivity and a low FP rate in our CAD scheme.