Michel Paindavoine - Academia.edu (original) (raw)
Papers by Michel Paindavoine
Journal of imaging, Mar 5, 2024
This article is an open access article distributed under the terms and conditions of the Creative... more This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY
Nephrology Dialysis Transplantation
Background Interstitial inflammation and peritubular capillaritis are observed in many diseases o... more Background Interstitial inflammation and peritubular capillaritis are observed in many diseases on native and transplant kidney biopsies. A precise and automated evaluation of these histological criteria could help stratify patients’ kidney prognoses and facilitate therapeutic management. Methods We used a convolutional neural network to evaluate those criteria on kidney biopsies. A total of 423 kidney samples from various diseases were included; 83 kidney samples were used for the neural network training, 106 for comparing manual annotations on limited areas to automated predictions, and 234 to compare automated and visual gradings. Results The precision, recall and F-score for leukocyte detection were, respectively, 81%, 71% and 76%. Regarding peritubular capillaries detection the precision, recall and F-score were, respectively, 82%, 83% and 82%. There was a strong correlation between the predicted and observed grading of total inflammation, as for the grading of capillaritis (r ...
HAL (Le Centre pour la Communication Scientifique Directe), Jun 25, 2017
Nephrology Dialysis Transplantation
Background Although the MEST-C classification is among the best prognostic tools in immunoglobuli... more Background Although the MEST-C classification is among the best prognostic tools in immunoglobulin A nephropathy (IgAN), it has a wide interobserver variability between specialized pathologists and others. Therefore we trained and evaluated a tool using a neural network to automate the MEST-C grading. Methods Biopsies of patients with IgAN were divided into three independent groups: the Training cohort (n = 42) to train the network, the Test cohort (n = 66) to compare its pixel segmentation to that made by pathologists and the Application cohort (n = 88) to compare the MEST-C scores computed by the network or by pathologists. Results In the Test cohort, >73% of pixels were correctly identified by the network as M, E, S or C. In the Application cohort, the neural network area under the receiver operating characteristics curves were 0.88, 0.91, 0.88, 0.94, 0.96, 0.96 and 0.92 to predict M1, E1, S1, T1, T2, C1 and C2, respectively. The kappa coefficients between pathologists and the...
HAL (Le Centre pour la Communication Scientifique Directe), Jan 12, 2015
Nephrology Dialysis Transplantation, May 1, 2022
Characteristics Overall (N=1 735) Age, mean (SD) years 40.9 (16.2) Women, n (%) 830 (47.8) Self-r... more Characteristics Overall (N=1 735) Age, mean (SD) years 40.9 (16.2) Women, n (%) 830 (47.8) Self-reported race, n (%) White 559 (32.2) Black 58 (3.3) Asian or Pacific Islander 704 (40.6) American Indian/Alaska Native 5 (0.3) Multiracial 98 (5.6) Unknown or Hispanic only 311 (17.9) Hispanic ethnicity, n (%) 320 (18.4) Prior medical history, n (%) Acute myocardial infarction 3 (0.2) Heart failure 26 (1.5) Mitral or aortic valvular disease 27 (1.6) Atrial fibrillation or atrial flutter 46 (2.7) Ischemic stroke or transient ischemic attack 8 (0.5) Venous thromboembolism 20 (1.2) Other thromboembolic events 2 (0.1) Hospitalized bleed 13 (0.7) Diabetes mellitus 158 (9.1) Hypertension 707 (40.7) Dyslipidemia 694 (40.0) Chronic lung disease 294 (16.9) Chronic liver disease 97 (5.6) Hyperthyroidism 41 (2.4) Hypothyroidism 125 (7.2) Depression 123 (7.1) Dementia 6 (0.3) Current Smoker 90 (5.2) Vital signs, mean (SD) Body mass index, kg/m 2 28.0 (7.1) Missing, n (%) 399 (23.0) Systolic blood pressure, mmHg 124.9 (17.6) Missing, n (%) 359 (20.7) Medication use, n (%) Prednisone 174 (10.0) Angiotensin-converting enzyme inhibitor 427 (24.6) Angiotensin II receptor blocker 241 (13.9) Aldosterone receptor antagonist 14 (0.8) Diuretic 308 (17.8)
2022 IEEE International Conference on Omni-layer Intelligent Systems (COINS)
L'invention concerne un dispositif d'extraction d'informations (10) à partir d'un... more L'invention concerne un dispositif d'extraction d'informations (10) à partir d'une image (12), le dispositif comportant une alternance de deux couches simples (S1, S2) et de deux couches complexes (C1, C2), -la première couche simple (S1) comportant un banc de filtres (15) qui diffèrent par leurs tailles et/ou leurs orientations, chaque filtre (15) étant configuré pour filtrer l'image d'entrée (12, 14), -la première couche simple (S1) intégrant des filtres (15) dont chaque coefficient (16) est codé sur un nombre de bits inférieur ou égal à 8 bits
National audienceAs nano-scale devices such as OG-CNTFETs are under studies and may be used in a ... more National audienceAs nano-scale devices such as OG-CNTFETs are under studies and may be used in a near future, we choose to investigate in which application domain such components may be of the most interest. In this paper we present how neural networks can be used to implement functions on nano-scale components. This method has been tested in the image processing application field
Procédé, mis en œuvre par un processeur, de codage d'un signal réel, par exemple un signal im... more Procédé, mis en œuvre par un processeur, de codage d'un signal réel, par exemple un signal image, en un signal quantifié, ledit procédé comprenant les étapes suivantes appliquées à chaque échantillon réel dudit signal réel : • Convertir (101) l'échantillon réel dans une représentation numérique, • Sélectionner (102), dans la partie fractionnaire du nombre, un nombre N prédéterminé de bits non nuls les plus significatifs, • pour chaque bit i significatif non nul sélectionné, i variant de 1 à N, déterminer (103) sa distance P, par rapport au bit significatif non nul sélectionné voisin de rang supérieur ou, pour le premier bit significatif non nul sélectionné, par rapport à la virgule, • retrancher (104) de ladite distance P, la valeur minimale de distance entre deux bits non nuls, • coder (106) la distance modifiée P, sur un nombre M, de bits prédéterminé
2015 4th International Conference on Advancements in Nuclear Instrumentation Measurement Methods and their Applications (ANIMMA), 2015
This paper introduces a specific Multiple Program Multiple Data (MPMD) architecture designed to a... more This paper introduces a specific Multiple Program Multiple Data (MPMD) architecture designed to address the issues of nuclear instrumentation: pulse processing, real-time, multi-channel flexibility, dead-time management and programmability. The proposed architecture comprises a set of independent and programmable functional units. Their execution is driven by the pulses arrival. It is able to deal with non-deterministic events and program durations. The virtual prototype of the architecture is developed in cycle-accurate SystemC and shows promising results in terms of scalability while maintaining zero dead-time. This architecture paves the way for novel programmable embedded real-time pulse processing restricted until now to offline processing.
A high speed analog VLSI Image acquisition and pre-processing system has been designed and fabric... more A high speed analog VLSI Image acquisition and pre-processing system has been designed and fabricated in a a 0.35 μm standard CMOS process. The chip features a massively parallel architecture enabling the computation of programmable low level image processing in each pixel. Extraction of spatial gradients and convolutions such as Sobel filter or Laplacian are implemented on the circuit. For this purpose, each pixel of 35 μm x 35 μm includes a photodiode, an amplifier, two storage capacitors and an analog arithmetic unit based on a fourquadrants multipliers architecture. The retina provides addressevent coded output on three asynchronous buses, one output is dedicated to the gradient and both others to the pixel values. A proof-of-concept chip of 64 x 64 pixels was fabricated. A dedicated embedded platform including FPGA, ADCs has also been designed to evaluate the vision chip. Measured results show that the proposed sensor successfully captures raw images up to 10 000 frames per sec...
Clinical Journal of the American Society of Nephrology, 2021
Background and objectives The prognosis of patients undergoing kidney tumor resection or kidney d... more Background and objectives The prognosis of patients undergoing kidney tumor resection or kidney donation is linked to many histologic criteria. These criteria notably include glomerular density, glomerular volume, vascular luminal stenosis, and severity of interstitial fibrosis/tubular atrophy. Automated measurements through a deep-learning approach could save time and provide more precise data. This work aimed to develop a free tool to automatically obtain kidney histologic prognostic features. Design, setting, participants, & measurements In total, 241 samples of healthy kidney tissue were split into three independent cohorts. The “Training” cohort (n=65) was used to train two convolutional neural networks: one to detect the cortex and a second to segment the kidney structures. The “Test” cohort (n=50) assessed their performance by comparing manually outlined regions of interest to predicted ones. The “Application” cohort (n=126) compared prognostic histologic data obtained manual...
1996 8th European Signal Processing Conference (EUSIPCO 1996), 1996
One of the way to extract edges uses the fast wavelet transform algorithm. This technique allows ... more One of the way to extract edges uses the fast wavelet transform algorithm. This technique allows the detection of multiscale edge and is used to detect all the details which are in a picture by modifying the scale. The real time application for edge detection involves the implementation of the algorithm on an integrated circuit like an FPGA and the development of an appropriated board. This article deals about the implementation of a wavelet transform algorithm onto a FPGA and the development of an electronic board to detect multiscale edges.
2021 International Conference on Content-Based Multimedia Indexing (CBMI), 2021
The aim of this paper is to present a sub-0.3 V neuromorphic technology developed for spiking neu... more The aim of this paper is to present a sub-0.3 V neuromorphic technology developed for spiking neural network design and its potential application. The main properties of the developed ultra low power (ULP) artificial neuron are first recalled. A description of ULP synapses follows that includes the plasticity scheme. The neuromorphic toolbox is then used to design a basic circuit allowing oriented edges classification. The circuit is made of 40 neurons and 108 plastic synapses, its consumed silicon core area is 0.025 mm 2 and the overall power consumption of 5 nW. Finally, the deployment of the technology within an industrial context to fabricate highly energy efficient Spike-based visual sensor is discussed.
Journal of Real-Time Image Processing, 2018
The implementation of a video reconstruction pipeline is required to improve the quality of image... more The implementation of a video reconstruction pipeline is required to improve the quality of images delivered by highly constrained devices. These algorithms require high computing capacities-several dozens of GOPs for real-time HD 1080p video streams. Today's embedded design constraints impose limitations both in terms of silicon budget and power consumption-usually 2 mm 2 for half a Watt. This paper presents the eISP architecture that is able to reach 188 MOPs/mW with 94 GOPs/mm 2 and 378 GOPs/mW using TSMC 65-nm integration technology. This fully programmable and modular architecture, is based on an analysis of video-processing algorithms. Synthesizable VHDL is generated taking into account different parameters, which simplify the architecture sizing and characterization.
Journal of Materials Processing Technology, 2001
La mesure de l'endommagement par fluage est un enjeu économique pour les fabricants de machines f... more La mesure de l'endommagement par fluage est un enjeu économique pour les fabricants de machines fonctionnant à haute température. La durée de vie résiduelle est un critère de nombreuses fois recherché par les constructeurs. Cet article est consacré à la description et à la mise au point d'une méthode d'analyse d'images destinée à évaluer le taux d'endommagement par fluage d'éprouvettes de laboratoire par comptage de vacuoles. Il est essentiellement destiné à l'étude préliminaire de l'extraction des informations pertinentes pour la mesure. L'extraction des joints de grains, information à caractère directionnel, constitue la problématique majeure pour la mise en oeuvre de la méthode, à résoudre avant automatisation de la mesure. L'utilisation d'un détecteur optimal de ligne de crêtes conduit au développement d'un filtre par transformée en ondelettes directionnelles. Nous montrerons les résultats de cette méthode originale de filtrage et son application dans un processus industriel.
Journal of imaging, Mar 5, 2024
This article is an open access article distributed under the terms and conditions of the Creative... more This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY
Nephrology Dialysis Transplantation
Background Interstitial inflammation and peritubular capillaritis are observed in many diseases o... more Background Interstitial inflammation and peritubular capillaritis are observed in many diseases on native and transplant kidney biopsies. A precise and automated evaluation of these histological criteria could help stratify patients’ kidney prognoses and facilitate therapeutic management. Methods We used a convolutional neural network to evaluate those criteria on kidney biopsies. A total of 423 kidney samples from various diseases were included; 83 kidney samples were used for the neural network training, 106 for comparing manual annotations on limited areas to automated predictions, and 234 to compare automated and visual gradings. Results The precision, recall and F-score for leukocyte detection were, respectively, 81%, 71% and 76%. Regarding peritubular capillaries detection the precision, recall and F-score were, respectively, 82%, 83% and 82%. There was a strong correlation between the predicted and observed grading of total inflammation, as for the grading of capillaritis (r ...
HAL (Le Centre pour la Communication Scientifique Directe), Jun 25, 2017
Nephrology Dialysis Transplantation
Background Although the MEST-C classification is among the best prognostic tools in immunoglobuli... more Background Although the MEST-C classification is among the best prognostic tools in immunoglobulin A nephropathy (IgAN), it has a wide interobserver variability between specialized pathologists and others. Therefore we trained and evaluated a tool using a neural network to automate the MEST-C grading. Methods Biopsies of patients with IgAN were divided into three independent groups: the Training cohort (n = 42) to train the network, the Test cohort (n = 66) to compare its pixel segmentation to that made by pathologists and the Application cohort (n = 88) to compare the MEST-C scores computed by the network or by pathologists. Results In the Test cohort, >73% of pixels were correctly identified by the network as M, E, S or C. In the Application cohort, the neural network area under the receiver operating characteristics curves were 0.88, 0.91, 0.88, 0.94, 0.96, 0.96 and 0.92 to predict M1, E1, S1, T1, T2, C1 and C2, respectively. The kappa coefficients between pathologists and the...
HAL (Le Centre pour la Communication Scientifique Directe), Jan 12, 2015
Nephrology Dialysis Transplantation, May 1, 2022
Characteristics Overall (N=1 735) Age, mean (SD) years 40.9 (16.2) Women, n (%) 830 (47.8) Self-r... more Characteristics Overall (N=1 735) Age, mean (SD) years 40.9 (16.2) Women, n (%) 830 (47.8) Self-reported race, n (%) White 559 (32.2) Black 58 (3.3) Asian or Pacific Islander 704 (40.6) American Indian/Alaska Native 5 (0.3) Multiracial 98 (5.6) Unknown or Hispanic only 311 (17.9) Hispanic ethnicity, n (%) 320 (18.4) Prior medical history, n (%) Acute myocardial infarction 3 (0.2) Heart failure 26 (1.5) Mitral or aortic valvular disease 27 (1.6) Atrial fibrillation or atrial flutter 46 (2.7) Ischemic stroke or transient ischemic attack 8 (0.5) Venous thromboembolism 20 (1.2) Other thromboembolic events 2 (0.1) Hospitalized bleed 13 (0.7) Diabetes mellitus 158 (9.1) Hypertension 707 (40.7) Dyslipidemia 694 (40.0) Chronic lung disease 294 (16.9) Chronic liver disease 97 (5.6) Hyperthyroidism 41 (2.4) Hypothyroidism 125 (7.2) Depression 123 (7.1) Dementia 6 (0.3) Current Smoker 90 (5.2) Vital signs, mean (SD) Body mass index, kg/m 2 28.0 (7.1) Missing, n (%) 399 (23.0) Systolic blood pressure, mmHg 124.9 (17.6) Missing, n (%) 359 (20.7) Medication use, n (%) Prednisone 174 (10.0) Angiotensin-converting enzyme inhibitor 427 (24.6) Angiotensin II receptor blocker 241 (13.9) Aldosterone receptor antagonist 14 (0.8) Diuretic 308 (17.8)
2022 IEEE International Conference on Omni-layer Intelligent Systems (COINS)
L'invention concerne un dispositif d'extraction d'informations (10) à partir d'un... more L'invention concerne un dispositif d'extraction d'informations (10) à partir d'une image (12), le dispositif comportant une alternance de deux couches simples (S1, S2) et de deux couches complexes (C1, C2), -la première couche simple (S1) comportant un banc de filtres (15) qui diffèrent par leurs tailles et/ou leurs orientations, chaque filtre (15) étant configuré pour filtrer l'image d'entrée (12, 14), -la première couche simple (S1) intégrant des filtres (15) dont chaque coefficient (16) est codé sur un nombre de bits inférieur ou égal à 8 bits
National audienceAs nano-scale devices such as OG-CNTFETs are under studies and may be used in a ... more National audienceAs nano-scale devices such as OG-CNTFETs are under studies and may be used in a near future, we choose to investigate in which application domain such components may be of the most interest. In this paper we present how neural networks can be used to implement functions on nano-scale components. This method has been tested in the image processing application field
Procédé, mis en œuvre par un processeur, de codage d'un signal réel, par exemple un signal im... more Procédé, mis en œuvre par un processeur, de codage d'un signal réel, par exemple un signal image, en un signal quantifié, ledit procédé comprenant les étapes suivantes appliquées à chaque échantillon réel dudit signal réel : • Convertir (101) l'échantillon réel dans une représentation numérique, • Sélectionner (102), dans la partie fractionnaire du nombre, un nombre N prédéterminé de bits non nuls les plus significatifs, • pour chaque bit i significatif non nul sélectionné, i variant de 1 à N, déterminer (103) sa distance P, par rapport au bit significatif non nul sélectionné voisin de rang supérieur ou, pour le premier bit significatif non nul sélectionné, par rapport à la virgule, • retrancher (104) de ladite distance P, la valeur minimale de distance entre deux bits non nuls, • coder (106) la distance modifiée P, sur un nombre M, de bits prédéterminé
2015 4th International Conference on Advancements in Nuclear Instrumentation Measurement Methods and their Applications (ANIMMA), 2015
This paper introduces a specific Multiple Program Multiple Data (MPMD) architecture designed to a... more This paper introduces a specific Multiple Program Multiple Data (MPMD) architecture designed to address the issues of nuclear instrumentation: pulse processing, real-time, multi-channel flexibility, dead-time management and programmability. The proposed architecture comprises a set of independent and programmable functional units. Their execution is driven by the pulses arrival. It is able to deal with non-deterministic events and program durations. The virtual prototype of the architecture is developed in cycle-accurate SystemC and shows promising results in terms of scalability while maintaining zero dead-time. This architecture paves the way for novel programmable embedded real-time pulse processing restricted until now to offline processing.
A high speed analog VLSI Image acquisition and pre-processing system has been designed and fabric... more A high speed analog VLSI Image acquisition and pre-processing system has been designed and fabricated in a a 0.35 μm standard CMOS process. The chip features a massively parallel architecture enabling the computation of programmable low level image processing in each pixel. Extraction of spatial gradients and convolutions such as Sobel filter or Laplacian are implemented on the circuit. For this purpose, each pixel of 35 μm x 35 μm includes a photodiode, an amplifier, two storage capacitors and an analog arithmetic unit based on a fourquadrants multipliers architecture. The retina provides addressevent coded output on three asynchronous buses, one output is dedicated to the gradient and both others to the pixel values. A proof-of-concept chip of 64 x 64 pixels was fabricated. A dedicated embedded platform including FPGA, ADCs has also been designed to evaluate the vision chip. Measured results show that the proposed sensor successfully captures raw images up to 10 000 frames per sec...
Clinical Journal of the American Society of Nephrology, 2021
Background and objectives The prognosis of patients undergoing kidney tumor resection or kidney d... more Background and objectives The prognosis of patients undergoing kidney tumor resection or kidney donation is linked to many histologic criteria. These criteria notably include glomerular density, glomerular volume, vascular luminal stenosis, and severity of interstitial fibrosis/tubular atrophy. Automated measurements through a deep-learning approach could save time and provide more precise data. This work aimed to develop a free tool to automatically obtain kidney histologic prognostic features. Design, setting, participants, & measurements In total, 241 samples of healthy kidney tissue were split into three independent cohorts. The “Training” cohort (n=65) was used to train two convolutional neural networks: one to detect the cortex and a second to segment the kidney structures. The “Test” cohort (n=50) assessed their performance by comparing manually outlined regions of interest to predicted ones. The “Application” cohort (n=126) compared prognostic histologic data obtained manual...
1996 8th European Signal Processing Conference (EUSIPCO 1996), 1996
One of the way to extract edges uses the fast wavelet transform algorithm. This technique allows ... more One of the way to extract edges uses the fast wavelet transform algorithm. This technique allows the detection of multiscale edge and is used to detect all the details which are in a picture by modifying the scale. The real time application for edge detection involves the implementation of the algorithm on an integrated circuit like an FPGA and the development of an appropriated board. This article deals about the implementation of a wavelet transform algorithm onto a FPGA and the development of an electronic board to detect multiscale edges.
2021 International Conference on Content-Based Multimedia Indexing (CBMI), 2021
The aim of this paper is to present a sub-0.3 V neuromorphic technology developed for spiking neu... more The aim of this paper is to present a sub-0.3 V neuromorphic technology developed for spiking neural network design and its potential application. The main properties of the developed ultra low power (ULP) artificial neuron are first recalled. A description of ULP synapses follows that includes the plasticity scheme. The neuromorphic toolbox is then used to design a basic circuit allowing oriented edges classification. The circuit is made of 40 neurons and 108 plastic synapses, its consumed silicon core area is 0.025 mm 2 and the overall power consumption of 5 nW. Finally, the deployment of the technology within an industrial context to fabricate highly energy efficient Spike-based visual sensor is discussed.
Journal of Real-Time Image Processing, 2018
The implementation of a video reconstruction pipeline is required to improve the quality of image... more The implementation of a video reconstruction pipeline is required to improve the quality of images delivered by highly constrained devices. These algorithms require high computing capacities-several dozens of GOPs for real-time HD 1080p video streams. Today's embedded design constraints impose limitations both in terms of silicon budget and power consumption-usually 2 mm 2 for half a Watt. This paper presents the eISP architecture that is able to reach 188 MOPs/mW with 94 GOPs/mm 2 and 378 GOPs/mW using TSMC 65-nm integration technology. This fully programmable and modular architecture, is based on an analysis of video-processing algorithms. Synthesizable VHDL is generated taking into account different parameters, which simplify the architecture sizing and characterization.
Journal of Materials Processing Technology, 2001
La mesure de l'endommagement par fluage est un enjeu économique pour les fabricants de machines f... more La mesure de l'endommagement par fluage est un enjeu économique pour les fabricants de machines fonctionnant à haute température. La durée de vie résiduelle est un critère de nombreuses fois recherché par les constructeurs. Cet article est consacré à la description et à la mise au point d'une méthode d'analyse d'images destinée à évaluer le taux d'endommagement par fluage d'éprouvettes de laboratoire par comptage de vacuoles. Il est essentiellement destiné à l'étude préliminaire de l'extraction des informations pertinentes pour la mesure. L'extraction des joints de grains, information à caractère directionnel, constitue la problématique majeure pour la mise en oeuvre de la méthode, à résoudre avant automatisation de la mesure. L'utilisation d'un détecteur optimal de ligne de crêtes conduit au développement d'un filtre par transformée en ondelettes directionnelles. Nous montrerons les résultats de cette méthode originale de filtrage et son application dans un processus industriel.