A5059953137- Gastrointestinal Bleeding Diagnosis and Treatment
- Image and Object Detection Techniques
- Image Processing Techniques and Applications
- Robotics and Sensor-Based Localization
- Colorectal Cancer Screening and Detection
- Advanced Data Compression Techniques
- AI in cancer detection
- QR Code Applications and Technologies
- Gastric Cancer Management and Outcomes
- Spacecraft Design and Technology
- Vehicle License Plate Recognition
- Digital Image Processing Techniques
- Digital Rights Management and Security
- Image Retrieval and Classification Techniques
- Colorectal Cancer Surgical Treatments
- Domain Adaptation and Few-Shot Learning
- Data Stream Mining Techniques
- Advanced Image and Video Retrieval Techniques
- Machine Learning and ELM
Sorbonne Université
2018-2023
Laboratoire de Recherche en Informatique de Paris 6
2020-2022
Institut d'Électronique et des Systèmes
2017-2021
Sorbonne Paris Cité
2021
Centre National de la Recherche Scientifique
2019-2021
In this article, we present our work on classifier to realize a Wireless Capsule Endoscopy (WCE) including Smart Vision Chip (SVC). Our is based fuzzy tree and forest of trees. We obtain sensitivity 92.80% specificity 91.26% with false detection rate 8.74% large database, that have constructed, composed 18910 images containing 3895 polyps from 20 different video-colonoscopies.
Hough Transform is a widely used shape-based algorithm for object detection and localization [6], this technique can be generalized to parametric curves as circles. For real time execution embedded integration, several optimizations are necessary due the large memory computational requirements. This paper presents an efficient real-time pipelined architecture with FPGA implementation of our multi-circles detection. The computation center candidates was improved. A three stages pipeline...
In this paper, an image processing to detect polyps in intelligent Wireless Capsule Endoscopy (WCE) is presented. This will be integrated into the WCE. It a new screening method colorectal cancer (CRC). A motion estimation algorithm used follow detected polyp and improve pre-processing of our detection chain. With methodology, rate improved by up 40% from 53% 93.7%. The was validated with large database 20 video-colonoscopies (18,910 images).
We propose a new paradigm of smart wireless endoscopic capsule (WCE) that has the ability to select suspicious images containing polyp before sending them outside body. To do so, we have designed an image processing system with Regions Of Interest (ROI) polyp. The criterion used ROI is based on polyp's shape. use Hough Transform (HT), widely shape-based algorithm for object detection and localization, make this selection. In paper, present compute in real-time high definition (1920 x 1080...
A way to improve the early detection of colorectal cancer is screening. Polyps are a marker and best modality detect them image. In 2003 Wireless Capsule Endoscopy was introduced opened integrate automatic image processing realize screening tool. Moreover, capacity polyp with Convolutional Neural Network shown in many scientific studies, but one issue integration these networks. this article, we present our works CNN or based on inside WCE powerful We apply knowledge distillation method....
Accurate endoscopic characterization of colorectal lesions is essential for predicting histology and choosing the most appropriate resection technique, but it remains very difficult endoscopists [1]. Lesions are characterized in real time according to their macroscopic appearance, vascular pattern, pit white light virtual chromoendoscopy. Numerous classifications required fully characterize various lesions, few gastroenterologists familiar with or use them daily practice
To reduce the incidence of colorectal cancer (CRC), we propose a new paradigm Wireless Capsule Endoscopy (WCE) that recognizes polyps in-situ. We embed an image processing chain in System on Chip (SoC) uses Hough Transform as part to detect circles High Definition (HD, 1920×1080) images. A circle here is probable marker ; polyp. widely used shape-based algorithm for object detection and localization. This technique can be generalized circles. inside WCE, considering real time execution...
Online learning approach allows an Artificial Neural Network (ANN) to solve dynamic real-world problems. In this context, the objective of work is implement ANN-based voice recognition models with focus on class-incremental in real time, for low-cost implementations, such as electric wheelchair. paper, online incremental Supervised Growing Gas (oiSGNG) a feature extractor proposed command classifier. About model, two contributions are presented: (i) nodes inserted according exponential...