Automatic detection of different types of small-bowel lesions on capsule endoscopy images using a newly developed deep convolutional neural network

Abstract Background Previous computer-aided detection systems for diagnosing lesions in images from wireless capsule endoscopy (WCE) have been limited to a single type of small-bowel lesion. We developed a new artificial intelligence (AI) system able to diagnose multiple types of lesions, including erosions and ulcers, vascular lesions, and tumors. Methods We trained the deep neural network system RetinaNet on a data set of 167 patients, which consisted of images of 398 erosions and ulcers, 538 vascular lesions, 4590 tumors, and 34 437 normal tissues. We calculated the mean area under the receiver operating characteristic curve (AUC) for each lesion type using five-fold stratified cross-validation. Results The mean age of the patients was 63.6 years; 92 were men. The mean AUCs of the AI system were 0.996 (95 %CI 0.992 – 0.999) for erosions and ulcers, 0.950 (95 %CI 0.923 – 0.978) for vascular lesions, and 0.950 (95 %CI 0.913 – 0.988) for tumors. Conclusion We developed and validated a new computer-aided diagnosis system for multiclass diagnosis of small-bowel lesions in WCE images.

Download Full-text

Is It Possible to Predict the Presence of Intestinal Angioectasias?

Diagnostic and Therapeutic Endoscopy ◽

10.1155/2014/461602 ◽

2014 ◽

Vol 2014 ◽

pp. 1-6 ◽

Cited By ~ 7

Author(s):

Tiago Cúrdia Gonçalves ◽

Joana Magalhães ◽

Pedro Boal Carvalho ◽

Maria João Moreira ◽

Bruno Rosa ◽

...

Keyword(s):

Gastrointestinal Tract ◽

Small Bowel ◽

Gastrointestinal Bleeding ◽

Predictive Factors ◽

Capsule Endoscopy ◽

Diagnostic Tool ◽

Diagnostic Yield ◽

Vascular Lesions ◽

Important Diagnostic Tool ◽

The Mean

Background and Aim. Angioectasias are the most common vascular anomalies found in the gastrointestinal tract. In small bowel (SB), they can cause obscure gastrointestinal bleeding (OGIB) and in this setting, small bowel capsule endoscopy (SBCE) is an important diagnostic tool. This study aimed to identify predictive factors for the presence of SB angioectasias, detected by SBCE. Methods. We retrospectively analyzed the results of 284 consecutive SBCE procedures between April 2006 and December 2012, whose indication was OGIB, of which 47 cases with SB angioectasias and 53 controls without vascular lesions were selected to enter the study. Demographic and clinical data were collected. Results. The mean age of subjects with angioectasias (70.9±14.7) was significantly higher than in controls (53.1±18.6; P<0.001). The presence of SB angioectasias was significantly higher when the indication for the exam was overt OGIB versus occult OGIB (13/19 versus 34/81, P=0.044). Hypertension and hypercholesterolemia were significantly associated with the presence of SB angioectasias (38/62 versus 9/38, P<0.001 and 28/47 versus 19/53, P=0.027, resp.). Other studied factors were not associated with small bowel angioectasias. Conclusions. In patients with OGIB, overt bleeding, older age, hypercholesterolemia, and hypertension are predictive of the presence of SB angioectasias detected by SBCE, which may be used to increase the diagnostic yield of the SBCE procedure and to reduce the proportion of nondiagnostic examinations.

Download Full-text

A Neural Network-based Algorithm for Assessing the Cleanliness of Small Bowel during Capsule Endoscopy

Endoscopy ◽

10.1055/a-1301-3841 ◽

2020 ◽

Author(s):

Romain Leenhardt ◽

Marc Souchaud ◽

Guy Houist ◽

Jean-Philippe Le Mouel ◽

Jean-Christophe Saurin ◽

...

Keyword(s):

Neural Network ◽

Small Bowel ◽

Capsule Endoscopy ◽

Third Generation ◽

Automatic Assessment ◽

Point Scale ◽

Automated Assessment ◽

Calculation Time ◽

Highly Sensitive ◽

The Mean

Background and Aims. Cleanliness scores in small bowel (SB) capsule endoscopy (CE) have poor reproducibility. The aim of this study was to evaluate a neural network (NN)-based algorithm for automated assessment of the SB cleanliness during CE. Methods: First, 600 normal third-generation SBCE still frames were categorized as “adequate” or “inadequate” in terms of cleanliness by three expert readers, according to a 10-point scale and served as a training database. Then, 156 third-generation SBCE recordings were categorized in a consensual manner as “adequate” or “inadequate” in terms of cleanliness. This testing database was split into two independent 78-video subsets for the tuning and evaluation of the algorithm. Results: Using a threshold of 79% adequate still frames per video to achieve the best performance, the algorithm yielded a sensitivity of 90.3%, a specificity of 83.3%, and an accuracy of 89.7%. The reproducibility was perfect. The mean calculation time per video was 3 ± 1 minutes. Conclusion: This NN-based algorithm allowing automatic assessment of SB cleanliness during CE was highly sensitive and paves the way for automated, standardized SBCE reports.

Download Full-text

Automatic Detection of Small Intestinal Hookworms in Capsule Endoscopy Images Based on a Convolutional Neural Network

Gastroenterology Research and Practice ◽

10.1155/2021/5682288 ◽

2021 ◽

Vol 2021 ◽

pp. 1-8

Author(s):

Tao Gan ◽

Yulin Yang ◽

Shuaicheng Liu ◽

Bing Zeng ◽

Jinlin Yang ◽

...

Keyword(s):

Neural Network ◽

Artificial Intelligence ◽

Small Bowel ◽

Convolutional Neural Network ◽

Capsule Endoscopy ◽

High Speed ◽

Characteristic Curve ◽

Area Under The Curve ◽

Intelligence System ◽

Sensitivity Specificity

Ancylostomiasis is a fairly common small bowel parasite disease identified by capsule endoscopy (CE) for which a computer-aided clinical detection method has not been established. We sought to develop an artificial intelligence system with a convolutional neural network (CNN) to automatically detect hookworms in CE images. We trained a deep CNN system based on a YOLO-V4 (You Look Only Once-Version4) detector using 11236 CE images of hookworms. We assessed its performance by calculating the area under the receiver operating characteristic curve and its sensitivity, specificity, and accuracy using an independent test set of 10,529 small-bowel images including 531 images of hookworms. The trained CNN system required 403 seconds to evaluate 10,529 test images. The area under the curve for the detection of hookworms was 0.972 (95% confidence interval (CI), 0.967-0.978). The sensitivity, specificity, and accuracy of the CNN system were 92.2%, 91.1%, and 91.2%, respectively, at a probability score cut-off of 0.485. We developed and validated a CNN-based system for detecting hookworms in CE images. By combining this high-accuracy, high-speed, and oversight-preventing system with other CNN systems, we hope it will become an important supplement for detecting intestinal abnormalities in CE images. This trial is registered with ChiCTR2000034546 (a clinical research of artificial-intelligence-aided diagnosis for hookworms in small intestine by capsule endoscope images).

Download Full-text

Deep learning and capsule endoscopy: automatic identification and differentiation of small bowel lesions with distinct haemorrhagic potential using a convolutional neural network

BMJ Open Gastroenterology ◽

10.1136/bmjgast-2021-000753 ◽

2021 ◽

Vol 8 (1) ◽

pp. e000753

Author(s):

Miguel José Mascarenhas Saraiva ◽

João Afonso ◽

Tiago Ribeiro ◽

João Ferreira ◽

Helder Cardoso ◽

...

Keyword(s):

Small Bowel ◽

Capsule Endoscopy ◽

Predictive Value ◽

Diagnostic Yield ◽

Characteristic Curve ◽

Vascular Lesions ◽

Automatic Identification ◽

Wide Range ◽

Sensitivity Specificity

ObjectiveCapsule endoscopy (CE) is pivotal for evaluation of small bowel disease. Obscure gastrointestinal bleeding most often originates from the small bowel. CE frequently identifies a wide range of lesions with different bleeding potentials in these patients. However, reading CE examinations is a time-consuming task. Convolutional neural networks (CNNs) are highly efficient artificial intelligence tools for image analysis. This study aims to develop a CNN-based model for identification and differentiation of multiple small bowel lesions with distinct haemorrhagic potential using CE images.DesignWe developed, trained, and validated a denary CNN based on CE images. Each frame was labelled according to the type of lesion (lymphangiectasia, xanthomas, ulcers, erosions, vascular lesions, protruding lesions, and blood). The haemorrhagic potential was assessed by Saurin’s classification. The entire dataset was divided into training and validation sets. The performance of the CNN was measured by the area under the receiving operating characteristic curve, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV).ResultsA total of 53 555 CE images were included. The model had an overall accuracy of 99%, a sensitivity of 88%, a specificity of 99%, a PPV of 87%, and an NPV of 99% for detection of multiple small bowel abnormalities and respective classification of bleeding potential.ConclusionWe developed and tested a CNN-based model for automatic detection of multiple types of small bowel lesions and classification of the respective bleeding potential. This system may improve the diagnostic yield of CE for these lesions and overall CE efficiency.

Download Full-text

Video Capsule Endoscopy In Comparison To Radiological Examination In Detection Of Small Bowel Lesions

QJM ◽

10.1093/qjmed/hcab100.029 ◽

2021 ◽

Vol 114 (Supplement_1) ◽

Author(s):

Amir Helmy Samy ◽

Nevine Ibrahim Musa ◽

Shereen Abou Bakr Saleh ◽

Ahmed Sayed Elgammal

Keyword(s):

Small Bowel ◽

Capsule Endoscopy ◽

Positive Contrast ◽

Video Capsule Endoscopy ◽

Vascular Lesions ◽

Deficiency Anemia ◽

Mesenteric Angiography ◽

Upper Gi ◽

Negative Results ◽

Upper Gi Endoscopy

Abstract BACKGROUND Small bowel obscured its lesions as secrets which were difficult to diagnose before video capsule endoscopy as a new modality for investigation. Aim of the study Evaluation of video capsule endoscopy in comparison to radiological examination in detection of small bowel lesions. Patients and methods Fifty patients were recruited from Kafrawy Video Capsule Endoscopy Unit of Internal Medicine Department and endoscopy unit of Ain Shams University Hospital. The study included patients with occult or overt GIT bleeding, patients with unexplained microcytic iron deficiency anemia, patients with chronic diarrhea and abdominal pain, with normal upper GI endoscopy and colonoscopy. Exclusion of any patient younger than 18 years old, has intestinal stricture, achalasia, or dysphagia. All patients were studied biochemically with CBC and radiological by CT pelvis and abdomen with IV and oral positive contrast some of them were radiologically examined with CTE or CT mesenteric angiography. All patients were endoscopically examined by OGD, colonoscopy, VCE, and some of them were examined also with enteroscope. Results The study revealed that the detection rate of SB lesions with VCE was 84%. In the current study, (44%) of cases had AVMs, (72.73) % of them were above the age of forty five, and (27.27) % were below the age of forty five. All patients who were investigated with CT mesenteric angiography revealed negative results. In this study (20) % of patients had SB masses and polyps, (70) % of them were at age of forty five or more and only (30) % of them were below the age of forty five. All patients underwent CT pelvis and abdomen with IV and oral positive contrast, and we found that all patients had a negative results regarding the SB lesions. In comparison between CTE and VCE in detection of SB vascular lesions CTE did not detect SB vacular lesions. On the other hand, VCE detected the AVMs in the cases with negative CTE results. In this study one patient (2)% was diagnosed with hookworm infection. All patient underwent for OGD. We found that (20)% of patients had a significant gastric or duodenal lesions (proximal to the papilla) by VCE but missed by upper GI endoscopy. In our study the concomitant of VCE and enteroscope increase the detection of SB vascular lesions than isolated use of VCE only. Conclusion VCE has a high detection rate of SB lesions (84)%. CTE has a low significance in detection of SB vascular lesions and CT mesenteric angiography sensitivity relatively low. AVMs more common with increasing the age. PHE and SB ectopic varicies, were found to be common causes of GIT bleeding in CLD patient. There is a significant rate of missed gastric and duodenal (proximal to the papilla) lesions that the cause of GIT bleeding and unexplained iron deficiency anemia in OGD examination that were detected by VCE.

Download Full-text

A neural network algorithm for detection of GI angiectasia during small-bowel capsule endoscopy

Gastrointestinal Endoscopy ◽

10.1016/j.gie.2018.06.036 ◽

2019 ◽

Vol 89 (1) ◽

pp. 189-194 ◽

Cited By ~ 58

Author(s):

Romain Leenhardt ◽

Pauline Vasseur ◽

Cynthia Li ◽

Jean Christophe Saurin ◽

Gabriel Rahmi ◽

...

Keyword(s):

Neural Network ◽

Small Bowel ◽

Capsule Endoscopy ◽

Network Algorithm ◽

Neural Network Algorithm

Download Full-text

Centered Kernel Alignment Enhancing Neural Network Pretraining for MRI-Based Dementia Diagnosis

Computational and Mathematical Methods in Medicine ◽

10.1155/2016/9523849 ◽

2016 ◽

Vol 2016 ◽

pp. 1-10 ◽

Cited By ~ 10

Author(s):

David Cárdenas-Peña ◽

Diego Collazos-Huertas ◽

German Castellanos-Dominguez

Keyword(s):

Neural Network ◽

Characteristic Curve ◽

Principal Component ◽

Imaging Data ◽

Resonance Imaging ◽

Dementia Diagnosis ◽

Target Matrix ◽

Computer Aided ◽

Artificial Neural Network Ann ◽

Aided Diagnosis

Dementia is a growing problem that affects elderly people worldwide. More accurate evaluation of dementia diagnosis can help during the medical examination. Several methods for computer-aided dementia diagnosis have been proposed using resonance imaging scans to discriminate between patients with Alzheimer’s disease (AD) or mild cognitive impairment (MCI) and healthy controls (NC). Nonetheless, the computer-aided diagnosis is especially challenging because of the heterogeneous and intermediate nature of MCI. We address the automated dementia diagnosis by introducing a novel supervised pretraining approach that takes advantage of the artificial neural network (ANN) for complex classification tasks. The proposal initializes an ANN based on linear projections to achieve more discriminating spaces. Such projections are estimated by maximizing the centered kernel alignment criterion that assesses the affinity between the resonance imaging data kernel matrix and the label target matrix. As a result, the performed linear embedding allows accounting for features that contribute the most to the MCI class discrimination. We compare the supervised pretraining approach to two unsupervised initialization methods (autoencoders and Principal Component Analysis) and against the best four performing classification methods of the 2014CADDementiachallenge. As a result, our proposal outperforms all the baselines (7% of classification accuracy and area under the receiver-operating-characteristic curve) at the time it reduces the class biasing.

Download Full-text

Artificial intelligence using a convolutional neural network for automatic detection of small‐bowel angioectasia in capsule endoscopy images

Digestive Endoscopy ◽

10.1111/den.13507 ◽

2020 ◽

Vol 32 (3) ◽

pp. 382-390 ◽

Cited By ~ 19

Author(s):

Akiyoshi Tsuboi ◽

Shiro Oka ◽

Kazuharu Aoyama ◽

Hiroaki Saito ◽

Tomonori Aoki ◽

...

Keyword(s):

Neural Network ◽

Artificial Intelligence ◽

Small Bowel ◽

Convolutional Neural Network ◽

Capsule Endoscopy ◽

Automatic Detection

Download Full-text

Automatic Detection of Arrow Annotation Overlays in Biomedical Images

International Journal of Healthcare Information Systems and Informatics ◽

10.4018/jhisi.2011100102 ◽

2011 ◽

Vol 6 (4) ◽

pp. 23-41 ◽

Cited By ~ 9

Author(s):

Beibei Cheng ◽

R. Joe Stanley ◽

Soumya De ◽

Sameer Antani ◽

George R. Thoma

Keyword(s):

Neural Network ◽

Artificial Neural Network ◽

Computational Intelligence ◽

Medical Images ◽

Characteristic Curve ◽

Data Set ◽

Negative Correlation Learning ◽

Artificial Neural ◽

Winner Take All ◽

Take All

Images in biomedical articles are often referenced for clinical decision support, educational purposes, and medical research. Authors-marked annotations such as text labels and symbols overlaid on these images are used to highlight regions of interest which are then referenced in the caption text or figure citations in the articles. Detecting and recognizing such symbols is valuable for improving biomedical information retrieval. In this research, image processing and computational intelligence methods are integrated for object segmentation and discrimination and applied to the problem of detecting arrows on these images. Evolving Artificial Neural Networks (EANNs) and Evolving Artificial Neural Network Ensembles (EANNEs) computational intelligence-based algorithms are developed to recognize overlays, specifically arrows, in medical images. For these discrimination techniques, EANNs use particle swarm optimization and genetic algorithm for artificial neural network (ANN) training, and EANNEs utilize the number of ANNs generated in an ensemble and negative correlation learning for neural network training based on averaging and Linear Vector Quantization (LVQ) winner-take-all approaches. Experiments performed on medical images from the imageCLEFmed’08 data set, yielded area under the receiver operating characteristic curve and precision/recall results as high as 0.988 and 0.928/0.973, respectively, using the EANNEs method with the winner-take-all approach.

Download Full-text