scholarly journals Algorithm combining virtual chromoendoscopy features for colorectal polyp classification

2021 ◽  
Vol 09 (10) ◽  
pp. E1497-E1503
Author(s):  
Ramon-Michel Schreuder ◽  
Qurine E.W. van der Zander ◽  
Roger Fonollà ◽  
Lennard P.L. Gilissen ◽  
Arnold Stronkhorst ◽  
...  
Keyword(s):  
Predictive Value ◽  
Gold Standard ◽  
Network Architecture ◽  
Colorectal Polyp ◽  
Colorectal Polyps ◽  
Training Dataset ◽  
High Definition ◽  
Still Images ◽  
Novel Approach ◽  
Incidence And Mortality

Abstract Background and study aims Colonoscopy is considered the gold standard for decreasing colorectal cancer incidence and mortality. Optical diagnosis of colorectal polyps (CRPs) is an ongoing challenge in clinical colonoscopy and its accuracy among endoscopists varies widely. Computer-aided diagnosis (CAD) for CRP characterization may help to improve this accuracy. In this study, we investigated the diagnostic accuracy of a novel algorithm for polyp malignancy classification by exploiting the complementary information revealed by three specific modalities. Methods We developed a CAD algorithm for CRP characterization based on high-definition, non-magnified white light (HDWL), Blue light imaging (BLI) and linked color imaging (LCI) still images from routine exams. All CRPs were collected prospectively and classified into benign or premalignant using histopathology as gold standard. Images and data were used to train the CAD algorithm using triplet network architecture. Our training dataset was validated using a threefold cross validation. Results In total 609 colonoscopy images of 203 CRPs of 154 consecutive patients were collected. A total of 174 CRPs were found to be premalignant and 29 were benign. Combining the triplet network features with all three image enhancement modalities resulted in an accuracy of 90.6 %, 89.7 % sensitivity, 96.6 % specificity, a positive predictive value of 99.4 %, and a negative predictive value of 60.9 % for CRP malignancy classification. The classification time for our CAD algorithm was approximately 90 ms per image. Conclusions Our novel approach and algorithm for CRP classification differentiates accurately between benign and premalignant polyps in non-magnified endoscopic images. This is the first algorithm combining three optical modalities (HDWL/BLI/LCI) exploiting the triplet network approach.

BASIC (BLI Adenoma Serrated International Classification) classification for colorectal polyp characterization with blue light imaging

Endoscopy ◽  
2017 ◽  
Vol 50 (03) ◽  
pp. 211-220 ◽  
Author(s):  
Raf Bisschops ◽  
Cesare Hassan ◽  
Pradeep Bhandari ◽  
Emmanuel Coron ◽  
Helmut Neumann ◽  
...  
Keyword(s):  
Blue Light ◽  
Interobserver Agreement ◽  
Colorectal Polyp ◽  
High Definition ◽  
Still Images ◽  
Heterogeneous Distribution ◽  
Modified Delphi ◽  
Optical Magnification ◽  
High Concordance ◽  
Video Library

Abstract Background and study aim Advanced endoscopic imaging has revolutionized the characterization of lesions during colonoscopy. The aim of this study was to create a new classification for differentiating subcentimetric hyperplastic and adenomatous polyps, and deeply invasive malignant lesions using blue-light imaging (BLI) with high definition, with and without optical magnification, as well as to assess its interobserver concordance. Methods A video library consisting of 48 videos/still images (with/without optical magnification) from 24 histologically verified polyps/cancer with BLI was prospectively created. In the first step, seven endoscopists with experience in electronic chromoendoscopy reviewed 12 BLI videos/still images with/without magnification representative of the different histotypes, and individually identified possible descriptors. In the second step, these descriptors were categorized and summarized with a modified Delphi methodology. In the third step, the seven endoscopists independently reviewed the remaining 36 videos/still images with/without optical magnification, and the interobserver agreement for the new descriptors was assessed. The interobserver agreement between endoscopists was assessed using Gwet’s AC1. Results By reviewing the initial 12 videos/still images, 43 descriptors were proposed. By a modified Delphi process, the endoscopists eventually agreed on summarizing 12 descriptors into three main domains. The main domains identified were: polyp surface (mucus, yes/no; regular/irregular; [pseudo]depressed, yes/no), pit appearance (featureless, yes/no; round/nonround with/without dark spots; homogeneous/heterogeneous distribution with/without focal loss), and vessels (present/absent, lacy, pericryptal, irregular). Interobserver agreement for the polyp surface domain appeared to be almost perfect for mucus (AC1 0.92 with and 0.88 without optical magnification), substantial for the regular/irregular surface (AC1 0.67 with and 0.66 without optical magnification). For the pit appearance domain, interobserver agreement was good for featureless (AC1 0.9 with and 0.8 without optical magnification), and round/nonround (AC1 0.77 with and 0.69 without optical magnification) descriptors, but less consistent for the homogeneity of distribution (AC1 with/without optical magnification 0.58). Agreement was almost perfect for the vessel domain (AC1 0.81 – 0.85). Conclusions The new BASIC classification takes into account both morphological features of the polyp, as well as crypt and vessel characteristics. A high concordance among the observers was shown for most of the summarized descriptors. Optical magnification had a beneficial effect in terms of interobserver agreement for most of the descriptors.

scholarly journals Predictive rules for optical diagnosis of < 10-mm colorectal polyps based on a dedicated software

Endoscopy ◽  
2019 ◽  
Vol 52 (01) ◽  
pp. 52-60 ◽  
Author(s):  
Cesare Hassan ◽  
Raf Bisschops ◽  
Pradeep Bhandari ◽  
Emmanuel Coron ◽  
Helmut Neumann ◽  
...  
Keyword(s):  
Area Under The Curve ◽  
Colorectal Polyp ◽  
Colorectal Polyps ◽  
Hyperplastic Polyps ◽  
High Confidence ◽  
Still Images ◽  
The Individual ◽  
Video Library ◽  
Leave One Out

Abstract Background The BASIC classification for predicting in vivo colorectal polyp histology incorporates both surface and pit/vessel descriptor domains. This study aimed to define new BASIC classes for adenomatous and hyperplastic polyps. Methods A video library (102 still images/videos of < 10-mm polyps using white-light [WLI] and blue-light imaging [BLI]) was reviewed by seven expert endoscopists. Polyps were rated according to the individual descriptors of the three BASIC domains (surface/pit/vessel). A model to predict polyp histology (adenomatous or hyperplastic) was developed using multivariable logistic regression and subsequent “leave-one-out” cross-validation. New BASIC rules were then defined by Delphi agreement. The overall accuracy of these rules when used by experts was evaluated according to the level of confidence and light type. Results The strength of prediction for adenomatous histology from 2175 observations assessed by area under the curve (AUC; 95 % confidence interval) was poor-to-fair for the surface descriptors (0.50 [0.33 – 0.69] for mucus; 0.68 [0.57 – 0.79] for irregular surface), but stronger for pits (0.87 [0.80 – 0.96] for featureless/round/not round) and vessels (0.80 [0.65 – 0.87] for not present/lacy/pericryptal). By combining the domains, a good-to-excellent prediction was shown (AUC 0.89 [0.81 – 0.96]). After the definition of new BASIC rules for adenomatous and hyperplastic polyps, accuracy for high confidence BLI predictions was 90.3 % (86.3 % – 93.2 %), which was superior to high confidence WLI (83.7 % [77.3 % – 87.7 %]) and low confidence BLI predictions (77.7 % [61.1 % – 88.6 %]). Conclusions Based on the strength of prediction, the new BASIC classes for adenomatous and hyperplastic histology show favorable results for accuracy and confidence levels.

Digital otoscopy versus microscopy: How correct and confident are ear experts in their diagnoses?

2017 ◽  
Vol 24 (7) ◽  
pp. 453-459 ◽  
Author(s):  
Aaron C Moberly ◽  
Margaret Zhang ◽  
Lianbo Yu ◽  
Metin Gurcan ◽  
Caglar Senaras ◽  
...  
Keyword(s):  
Gold Standard ◽  
Correct Diagnosis ◽  
Sufficient Information ◽  
High Definition ◽  
Standard View ◽  
Diagnostic Ability ◽  
Still Images ◽  
Still Image ◽  
Degree Of Confidence ◽  
Ear Effusion

Introduction With the growing popularity of telemedicine and tele-diagnostics, clinical validation of new devices is essential. This study sought to investigate whether high-definition digital still images of the eardrum provide sufficient information to make a correct diagnosis, as compared with the gold standard view provided by clinical microscopy. Methods Twelve fellowship-trained ear physicians (neurotologists) reviewed the same set of 210 digital otoscope eardrum images. Participants diagnosed each image as normal or, if abnormal, they selected from seven types of ear pathology. Diagnostic percentage correct for each pathology was compared with a gold standard of diagnosis using clinical microscopy with adjunct audiometry and/or tympanometry. Participants also rated their degree of confidence for each diagnosis. Results Overall correctness of diagnosis for ear pathologies ranged from 48.6–100%, depending on the type of pathology. Neurotologists were 72% correct in identifying eardrums as normal. Reviewers’ confidence in diagnosis varied substantially among types of pathology, as well as among participants. Discussion High-definition digital still images of eardrums provided sufficient information for neurotologists to make correct diagnoses for some pathologies. However, some diagnoses, such as middle ear effusion, were more difficult to diagnose when based only on a still image. Levels of confidence of reviewers did not generally correlate with diagnostic ability.

scholarly journals SynPo-Net—Accurate and Fast CNN-Based 6DoF Object Pose Estimation Using Synthetic Training

Sensors ◽  
2021 ◽  
Vol 21 (1) ◽  
pp. 300
Author(s):  
Yongzhi Su ◽  
Jason Rambach ◽  
Alain Pagani ◽  
Didier Stricker
Keyword(s):  
Pose Estimation ◽  
Network Architecture ◽  
Single Channel ◽  
Domain Adaptation ◽  
Tracking System ◽  
Ground Truth ◽  
Training Dataset ◽  
Extensive Evaluation ◽  
Novel Approach ◽  
Synthetic Images

Estimation and tracking of 6DoF poses of objects in images is a challenging problem of great importance for robotic interaction and augmented reality. Recent approaches applying deep neural networks for pose estimation have shown encouraging results. However, most of them rely on training with real images of objects with severe limitations concerning ground truth pose acquisition, full coverage of possible poses, and training dataset scaling and generalization capability. This paper presents a novel approach using a Convolutional Neural Network (CNN) trained exclusively on single-channel Synthetic images of objects to regress 6DoF object Poses directly (SynPo-Net). The proposed SynPo-Net is a network architecture specifically designed for pose regression and a proposed domain adaptation scheme transforming real and synthetic images into an intermediate domain that is better fit for establishing correspondences. The extensive evaluation shows that our approach significantly outperforms the state-of-the-art using synthetic training in terms of both accuracy and speed. Our system can be used to estimate the 6DoF pose from a single frame, or be integrated into a tracking system to provide the initial pose.

scholarly journals Utility of the narrow-band imaging international colorectal endoscopic classification for optical diagnosis of colorectal polyp histology in clinical practice: a retrospective study

2021 ◽  
Vol 21 (1) ◽  
Author(s):  
Yasuhiko Hamada ◽  
Kyosuke Tanaka ◽  
Masaki Katsurahara ◽  
Noriyuki Horiki ◽  
Reiko Yamada ◽  
...  
Keyword(s):  
Narrow Band ◽  
Narrow Band Imaging ◽  
Colorectal Polyp ◽  
Colorectal Polyps ◽  
Optical Diagnosis ◽  
High Definition ◽  
Clinical Tool ◽  
Optical Magnification ◽  
High Level ◽  
Sensitivity Specificity

Abstract Background Narrow-band imaging (NBI) highlights the surface structures and vessels of colorectal polyps and is useful for determining the polyp histology. The narrow-band imaging international colorectal endoscopic (NICE) classification is a diagnostic tool for determining colorectal polyp histology based on NBI without optical magnification. In this study, we aimed to investigate the value of each type of the NICE classification for determining colorectal polyp histology using endoscopy data accumulated in a clinical setting. Methods Endoscopy data for 534 colorectal polyps (316 patients) treated at our facility were retrospectively analyzed. First, we investigated the diagnostic performance of each type of the NICE classification for the optical diagnosis of colorectal polyp histology. The procedures were performed by experienced endoscopists using high-definition colonoscopy without optical magnification. Second, inter-observer and intra-observer agreements were assessed after providing experts and non-experts with a short lecture on the NICE classification. Using 50 fine NBI images of colorectal polyps without optical magnification, the inter-observer and intra-observer agreements between five experts and five non-experts were assessed. Results The sensitivity, specificity, and accuracy values were 86.0%, 99.6%, and 98.5% for NICE type 1 lesions; 99.2%, 85.2%, and 97.8% for NICE type 2 lesions; and 81.8%, 99.6%, and 99.3% for NICE type 3 lesions, respectively. The inter-observer and intra-observer agreements ranged from substantial to excellent for both experts and non-experts. Conclusions The NICE classification had good diagnostic ability in terms of determining the polyp histology and demonstrated a high level of reproducibility among experts and non-experts. Thus, the NICE classification is a useful clinical tool that can be used without optical magnification.

Non-Blind Image Deconvolution Based on “Ringing” Removal Using Convolutional Neural Network

2020 ◽  
Vol 2020 (10) ◽  
pp. 181-1-181-7
Author(s):  
Takahiro Kudo ◽  
Takanori Fujisawa ◽  
Takuro Yamaguchi ◽  
Masaaki Ikehara
Keyword(s):  
Neural Network ◽  
Convolutional Neural Network ◽  
Network Architecture ◽  
Large Scale ◽  
Blind Deconvolution ◽  
Training Dataset ◽  
Image Deconvolution ◽  
Classic Problem ◽  
Key Points ◽  
Blind Image

Image deconvolution has been an important issue recently. It has two kinds of approaches: non-blind and blind. Non-blind deconvolution is a classic problem of image deblurring, which assumes that the PSF is known and does not change universally in space. Recently, Convolutional Neural Network (CNN) has been used for non-blind deconvolution. Though CNNs can deal with complex changes for unknown images, some CNN-based conventional methods can only handle small PSFs and does not consider the use of large PSFs in the real world. In this paper we propose a non-blind deconvolution framework based on a CNN that can remove large scale ringing in a deblurred image. Our method has three key points. The first is that our network architecture is able to preserve both large and small features in the image. The second is that the training dataset is created to preserve the details. The third is that we extend the images to minimize the effects of large ringing on the image borders. In our experiments, we used three kinds of large PSFs and were able to observe high-precision results from our method both quantitatively and qualitatively.

scholarly journals A new hybrid record linkage process to make epidemiological databases interoperable: application to the GEMO and GENEPSO studies involving BRCA1 and BRCA2 mutation carriers

2021 ◽  
Vol 21 (1) ◽  
Author(s):  
Yue Jiao ◽  
Fabienne Lesueur ◽  
Chloé-Agathe Azencott ◽  
Maïté Laurent ◽  
Noura Mebirouk ◽  
...  
Keyword(s):  
Record Linkage ◽  
Gold Standard ◽  
Brca2 Mutation ◽  
Epidemiological Studies ◽  
Supervised Machine Learning ◽  
Training Dataset ◽  
Support Vector ◽  
Genetic Modifiers ◽  
Brca1 And Brca2 ◽  
Mutation Carriers

Abstract Background Linking independent sources of data describing the same individuals enable innovative epidemiological and health studies but require a robust record linkage approach. We describe a hybrid record linkage process to link databases from two independent ongoing French national studies, GEMO (Genetic Modifiers of BRCA1 and BRCA2), which focuses on the identification of genetic factors modifying cancer risk of BRCA1 and BRCA2 mutation carriers, and GENEPSO (prospective cohort of BRCAx mutation carriers), which focuses on environmental and lifestyle risk factors. Methods To identify as many as possible of the individuals participating in the two studies but not registered by a shared identifier, we combined probabilistic record linkage (PRL) and supervised machine learning (ML). This approach (named “PRL + ML”) combined together the candidate matches identified by both approaches. We built the ML model using the gold standard on a first version of the two databases as a training dataset. This gold standard was obtained from PRL-derived matches verified by an exhaustive manual review. Results The Random Forest (RF) algorithm showed a highest recall (0.985) among six widely used ML algorithms: RF, Bagged trees, AdaBoost, Support Vector Machine, Neural Network. Therefore, RF was selected to build the ML model since our goal was to identify the maximum number of true matches. Our combined linkage PRL + ML showed a higher recall (range 0.988–0.992) than either PRL (range 0.916–0.991) or ML (0.981) alone. It identified 1995 individuals participating in both GEMO (6375 participants) and GENEPSO (4925 participants). Conclusions Our hybrid linkage process represents an efficient tool for linking GEMO and GENEPSO. It may be generalizable to other epidemiological studies involving other databases and registries.

scholarly journals Automatic detection and segmentation of adenomatous colorectal polyps during colonoscopy using Mask R-CNN

2020 ◽  
Vol 15 (1) ◽  
pp. 588-596 ◽  
Author(s):  
Jie Meng ◽  
Linyan Xue ◽  
Ying Chang ◽  
Jianguang Zhang ◽  
Shilong Chang ◽  
...  
Keyword(s):  
Deep Neural Networks ◽  
Alimentary Tract ◽  
Multiple Scale ◽  
Colorectal Polyps ◽  
Training Dataset ◽  
Cad System ◽  
Multicenter Trials ◽  
Testing Dataset ◽  
Adenoma Detection ◽  
Aided Diagnosis

AbstractColorectal cancer (CRC) is one of the main alimentary tract system malignancies affecting people worldwide. Adenomatous polyps are precursors of CRC, and therefore, preventing the development of these lesions may also prevent subsequent malignancy. However, the adenoma detection rate (ADR), a measure of the ability of a colonoscopist to identify and remove precancerous colorectal polyps, varies significantly among endoscopists. Here, we attempt to use a convolutional neural network (CNN) to generate a unique computer-aided diagnosis (CAD) system by exploring in detail the multiple-scale performance of deep neural networks. We applied this system to 3,375 hand-labeled images from the screening colonoscopies of 1,197 patients; of whom, 3,045 were assigned to the training dataset and 330 to the testing dataset. The images were diagnosed simply as either an adenomatous or non-adenomatous polyp. When applied to the testing dataset, our CNN-CAD system achieved a mean average precision of 89.5%. We conclude that the proposed framework could increase the ADR and decrease the incidence of interval CRCs, although further validation through large multicenter trials is required.

scholarly journals Estimating PQoS of Video Conferencing on Wi-Fi Networks Using Machine Learning

2021 ◽  
Vol 13 (3) ◽  
pp. 63
Author(s):  
Maghsoud Morshedi ◽  
Josef Noll
Keyword(s):  
Machine Learning ◽  
Network Performance ◽  
Service Providers ◽  
Video Conferencing ◽  
Machine Learning Algorithms ◽  
Performance Parameters ◽  
High Definition ◽  
Internet Service ◽  
Novel Approach ◽  
5 Ghz

Video conferencing services based on web real-time communication (WebRTC) protocol are growing in popularity among Internet users as multi-platform solutions enabling interactive communication from anywhere, especially during this pandemic era. Meanwhile, Internet service providers (ISPs) have deployed fiber links and customer premises equipment that operate according to recent 802.11ac/ax standards and promise users the ability to establish uninterrupted video conferencing calls with ultra-high-definition video and audio quality. However, the best-effort nature of 802.11 networks and the high variability of wireless medium conditions hinder users experiencing uninterrupted high-quality video conferencing. This paper presents a novel approach to estimate the perceived quality of service (PQoS) of video conferencing using only 802.11-specific network performance parameters collected from Wi-Fi access points (APs) on customer premises. This study produced datasets comprising 802.11-specific network performance parameters collected from off-the-shelf Wi-Fi APs operating at 802.11g/n/ac/ax standards on both 2.4 and 5 GHz frequency bands to train machine learning algorithms. In this way, we achieved classification accuracies of 92–98% in estimating the level of PQoS of video conferencing services on various Wi-Fi networks. To efficiently troubleshoot wireless issues, we further analyzed the machine learning model to correlate features in the model with the root cause of quality degradation. Thus, ISPs can utilize the approach presented in this study to provide predictable and measurable wireless quality by implementing a non-intrusive quality monitoring approach in the form of edge computing that preserves customers’ privacy while reducing the operational costs of monitoring and data analytics.

scholarly journals Reverse Scan Conversion and Efficient Deep Learning Network Architecture for Ultrasound Imaging on a Mobile Device

Sensors ◽  
2021 ◽  
Vol 21 (8) ◽  
pp. 2629
Author(s):  
Kunkyu Lee ◽  
Min Kim ◽  
Changhyun Lim ◽  
Tai-Kyong Song
Keyword(s):  
Deep Learning ◽  
Ultrasound Imaging ◽  
Mobile Device ◽  
Network Architecture ◽  
Point Of Care ◽  
Ultrasound Image ◽  
Training Dataset ◽  
Point Of Care Ultrasound ◽  
Reverse Scan ◽  
Scan Conversion

Point-of-care ultrasound (POCUS), realized by recent developments in portable ultrasound imaging systems for prompt diagnosis and treatment, has become a major tool in accidents or emergencies. Concomitantly, the number of untrained/unskilled staff not familiar with the operation of the ultrasound system for diagnosis is increasing. By providing an imaging guide to assist clinical decisions and support diagnosis, the risk brought by inexperienced users can be managed. Recently, deep learning has been employed to guide users in ultrasound scanning and diagnosis. However, in a cloud-based ultrasonic artificial intelligence system, the use of POCUS is limited due to information security, network integrity, and significant energy consumption. To address this, we propose (1) a structure that simultaneously provides ultrasound imaging and a mobile device-based ultrasound image guide using deep learning, and (2) a reverse scan conversion (RSC) method for building an ultrasound training dataset to increase the accuracy of the deep learning model. Experimental results show that the proposed structure can achieve ultrasound imaging and deep learning simultaneously at a maximum rate of 42.9 frames per second, and that the RSC method improves the image classification accuracy by more than 3%.

Sign in / Sign up

Export Citation Format

Share Document