Saint-Contest, France
Saint-Contest, France

Time filter

Source Type

Decenciere E.,MINES ParisTech | Zhang X.,MINES ParisTech | Cazuguel G.,Telecom Bretagne | Cazuguel G.,French Institute of Health and Medical Research | And 9 more authors.
Image Analysis and Stereology | Year: 2014

The Messidor database, which contains hundreds of eye fundus images, has been publicly distributed since 2008. It was created by the Messidor project in order to evaluate automatic lesion segmentation and diabetic retinopathy grading methods. Designing, producing and maintaining such a database entails significant costs. By publicly sharing it, one hopes to bring a valuable resource to the public research community. However, the real interest and benefit of the research community is not easy to quantify. We analyse here the feedback on the Messidor database, after more than 6 years of diffusion. This analysis should apply to other similar research databases.


Sudha N.,Coimbatore Institute of Technology | Santhiyakumari N.,Knowledge Institute of Technology | Brunolay,ADCIS
Proceedings of 2015 IEEE International Conference on Electrical, Computer and Communication Technologies, ICECCT 2015 | Year: 2015

Wireless Capsule Endoscopy (WCE) plays a wide role in the diagnosis of bowel diseases. Capsule Endoscopy produces large number of images and it has been analyzed manually by the physicians which is a tough task. Such large data set provide an opportunity for segmenting the bowel images. A segmentation method for extraction of bowel images based on threshold segmentation technique followed by the morphological technique is proposed in this paper. The segmentation of bowel image has been obtained using Aphelion Dev real time software. The resulting image has been implemented in Virtex Field Programmable Gate Array (FPGA) kit which provides minimum hardware resources and low power consumption and is desirable for real time medical applications and analysis. Total power consumption is 0.182W. © 2015 IEEE.


Watremez X.,TOTAL Exploration and Production | Labat N.,TOTAL Exploration and Production | Audouin G.,TOTAL Exploration and Production | Lejay B.,TOTAL Exploration and Production | And 8 more authors.
Proceedings - SPE Annual Technical Conference and Exhibition | Year: 2016

Hydrocarbon leaks in oil and gas installations present Health, Safety and Environmental risks. History of crisis management in oil and gas upstream has shown the value of efficient and accurate tools for quantifying the gas-leak rate and determining the perimeter of the hazardous areas. In this context, Total initiated a multi-year R&D collaborative project designed to develop remote sensing technologies and architectures for remote detection, identification, quantification and visualization of gas leaks in the event of a crisis. Total, the ONERA - The French Aerospace Lab - and ADCIS have developed a set of algorithms and software to measure, compute and visualize a methane plume using infrared optical imagers. Results are obtained in 3D and in real time. The following steps are involved: (1) Spectral images in the Long-Wavelength InfraRed (LWIR) region are captured by three hyper-spectral cameras located around a methane release point; (2) Concentrations of methane are measured linearly in ppm.m by comparing spectral images of the scene in the presence of gas and reference images acquired before the release; (3) An algorithm, drawing on tomography techniques, computes concentrations of methane in ppm from the linear concentrations; (4) Mass balance type equations finally help estimate the methane flowrates based on the set of concentrations and local wind data information. A one-week test campaign was organized in September 2015 and consisted of performing twenty-six methane gas releases of 1 g/s to 50 g/s. Three Telops Hyper-Cam cameras were connected as part of a network to a main server which ran the tomography and flowrate estimation code. The real-time remote detection and quantification worked fully. During the campaign, good accuracy was obtained at the low flowrates of 1 g/s and 10 g/s of methane. At the higher flowrate of 50 g/s, quantifications were underestimated due to an oversaturation phenomenon. Further works, the aim of which is to adapt the instrument sensing ranges to the maximum concentrations encountered, should help improve the accuracy of these quantifications. The innovation lies in the fact that a 3D visualization of the methane plume can be computed and created in real time and that flowrates and concentrations can be quantified, also in real time. This technology could be applied in environmental monitoring and crisis management. Copyright 2016, Society of Petroleum Engineers.


Quellec G.,French Institute of Health and Medical Research | Lamard M.,French Institute of Health and Medical Research | Cochener B.,French Institute of Health and Medical Research | Decenciere E.,MINES ParisTech | And 4 more authors.
Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS | Year: 2013

This paper presents TeleOphta, an automatic system for screening diabetic retinopathy in teleophthalmology networks. Its goal is to reduce the burden on ophthalmologists by automatically detecting non referable examination records, i.e. examination records presenting no image quality problems and no pathological signs related to diabetic retinopathy or any other retinal pathology. TeleOphta is an attempt to put into practice years of algorithmic developments from our groups. It combines image quality metrics, specific lesion detectors and a generic pathological pattern miner to process the visual content of eye fundus photographs. This visual information is further combined with contextual data in order to compute an abnormality risk for each examination record. The TeleOphta system was trained and tested on a large dataset of 25,702 examination records from the OPHDIAT screening network in Paris. It was able to automatically detect 68% of the non referable examination records while achieving the same sensitivity as a second ophthalmologist. This suggests that it could safely reduce the burden on ophthalmologists by 56%. © 2013 IEEE.


Zhang X.,MINES ParisTech | Thibault G.,MINES ParisTech | Decenciere E.,MINES ParisTech | Marcotegui B.,MINES ParisTech | And 11 more authors.
Medical Image Analysis | Year: 2014

The automatic detection of exudates in color eye fundus images is an important task in applications such as diabetic retinopathy screening. The presented work has been undertaken in the framework of the TeleOphta project, whose main objective is to automatically detect normal exams in a tele-ophthalmology network, thus reducing the burden on the readers. A new clinical database, e-ophtha EX, containing precisely manually contoured exudates, is introduced. As opposed to previously available databases, e-ophtha EX is very heterogeneous. It contains images gathered within the OPHDIAT telemedicine network for diabetic retinopathy screening. Image definition, quality, as well as patients condition or the retinograph used for the acquisition, for example, are subject to important changes between different examinations. The proposed exudate detection method has been designed for this complex situation. We propose new preprocessing methods, which perform not only normalization and denoising tasks, but also detect reflections and artifacts in the image. A new candidates segmentation method, based on mathematical morphology, is proposed. These candidates are characterized using classical features, but also novel contextual features. Finally, a random forest algorithm is used to detect the exudates among the candidates. The method has been validated on the e-ophtha EX database, obtaining an AUC of 0.95. It has been also validated on other databases, obtaining an AUC between 0.93 and 0.95, outperforming state-of-the-art methods. © 2014 Elsevier B.V.


Sindt C.W.,University of Iowa | Lay B.,ADCIS | Bouchard H.,ADCIS | Kern J.R.,Alcon
Cornea | Year: 2013

PURPOSE:: To develop rapid image processing techniques for the objective analysis of corneal in vivo confocal micrographs. METHODS:: Perpendicular central corneal volume scans from healthy volunteers were obtained via laser in vivo confocal microscopy. The layer in each volume scan that contained the nerve plexus was detected by applying software operators to analyze image features on the basis of their size, shape, and contrast. Dendritic immune cells were detected in the nerve image on the basis of cellular size, lack of elongation, and brightness relative to the nerves. Images that were 20 μm anterior to the best nerve layer images were used for the analysis of epithelial wing cells; wing cell detection was based on extended regional minima and a watershed transformation. RESULTS:: The software successfully detected the best nerve layer images in 15 scans from 15 eyes. Manual and automatic analyses were 81.8% in agreement for dendritic immune cells (for 11 cells in a representative image) and 94.4% in agreement for wing cells (for 466 cells in the image). Within 10 seconds per scan, the software calculated the number, mean length, and mean density of immune cells; the number, mean size, and mean density of wing cells; and the number and mean length of nerves. Factors defining the shape and position of cells and nerves also were available. CONCLUSIONS:: The software rapidly and accurately analyzed the in vivo confocal micrographs of the healthy central corneas, yielding quantitative results to describe the nerves, dendritic immune cells, and wing cells. Copyright © 2012 by Lippincott Williams Wilkins.


Hemalatha R.,Knowledge Institute of Technology | Santhiyakumari N.,Knowledge Institute of Technology | Lay B.J.,ADCIS
International Conference on Electrical, Electronics, Signals, Communication and Optimization, EESCO 2015 | Year: 2015

The diagnosis of ultrasound image becomes more difficult with the presence of speckle noise. It tends to reduce the resolution and contrast of image which minimizes the diagnostic values of an ultrasound image. An important step in the analysis of medical image is to improve the quality of ultrasound image by removing speckle noise. In this paper, the performance of Gaussian filtering in despeckling the ultrasound common carotid artery image has been compared with other speckle reduction filters like Box, Convolution, Median, Mode, Nagoa, Rank value and Weymouth filter. The statistical values such as mean, standard deviation, skewness and kurtosis are used to analyze the performance of different filters with the aid of Aphelion Dev software. The filtered image has been implemented in Unified Technology Learning Platform kit which expedites the process of image. This technique will helps to improve the appearance of ultrasound image and detects the presence of abnormalities in carotid artery earlier during diagnosis. © 2015 IEEE.


Decenciere E.,MINES ParisTech | Cazuguel G.,French Institute of Health and Medical Research | Cazuguel G.,Telecom Bretagne | Zhang X.,MINES ParisTech | And 13 more authors.
IRBM | Year: 2013

A complete prototype for the automatic detection of normal examinations on a teleophthalmology network for diabetic retinopathy screening is presented. The system combines pathological pattern mining methods, with specific lesion detection methods, to extract information from the images. This information, plus patient and other contextual data, is used by a classifier to compute an abnormality risk. Such a system should reduce the burden on readers on teleophthalmology networks. © 2013 Elsevier Masson SAS.


Quellec G.,French Institute of Health and Medical Research | Lamard M.,French Institute of Health and Medical Research | Lamard M.,University of Western Brittany | Abramoff M.D.,University of Iowa | And 8 more authors.
Medical Image Analysis | Year: 2012

A novel multiple-instance learning framework, for automated image classification, is presented in this paper. Given reference images marked by clinicians as relevant or irrelevant, the image classifier is trained to detect patterns, of arbitrary size, that only appear in relevant images. After training, similar patterns are sought in new images in order to classify them as either relevant or irrelevant images. Therefore, no manual segmentations are required. As a consequence, large image datasets are available for training. The proposed framework was applied to diabetic retinopathy screening in 2-D retinal image datasets: Messidor (1200 images) and e-ophtha, a dataset of 25,702 examination records from the Ophdiat screening network (107,799 images). In this application, an image (or an examination record) is relevant if the patient should be referred to an ophthalmologist. Trained on one half of Messidor, the classifier achieved high performance on the other half of Messidor (Az=0.881) and on e-ophtha (Az=0.761). We observed, in a subset of 273 manually segmented images from e-ophtha, that all eight types of diabetic retinopathy lesions are detected. © 2012 Elsevier B.V.


PubMed | ADCIS, Direction de la politique medicale, MINES ParisTech, French Institute of Health and Medical Research and 2 more.
Type: Journal Article | Journal: Medical image analysis | Year: 2014

The automatic detection of exudates in color eye fundus images is an important task in applications such as diabetic retinopathy screening. The presented work has been undertaken in the framework of the TeleOphta project, whose main objective is to automatically detect normal exams in a tele-ophthalmology network, thus reducing the burden on the readers. A new clinical database, e-ophtha EX, containing precisely manually contoured exudates, is introduced. As opposed to previously available databases, e-ophtha EX is very heterogeneous. It contains images gathered within the OPHDIAT telemedicine network for diabetic retinopathy screening. Image definition, quality, as well as patients condition or the retinograph used for the acquisition, for example, are subject to important changes between different examinations. The proposed exudate detection method has been designed for this complex situation. We propose new preprocessing methods, which perform not only normalization and denoising tasks, but also detect reflections and artifacts in the image. A new candidates segmentation method, based on mathematical morphology, is proposed. These candidates are characterized using classical features, but also novel contextual features. Finally, a random forest algorithm is used to detect the exudates among the candidates. The method has been validated on the e-ophtha EX database, obtaining an AUC of 0.95. It has been also validated on other databases, obtaining an AUC between 0.93 and 0.95, outperforming state-of-the-art methods.

Loading ADCIS collaborators
Loading ADCIS collaborators