A5060441651- Medical Imaging Techniques and Applications
- Radiomics and Machine Learning in Medical Imaging
- Prostate Cancer Diagnosis and Treatment
- Medical Imaging and Analysis
- Advanced X-ray and CT Imaging
- Advanced Neural Network Applications
- Cancer-related molecular mechanisms research
- Cryptography and Data Security
- Blockchain Technology Applications and Security
- Advanced Radiotherapy Techniques
- Medical Image Segmentation Techniques
- Cell Image Analysis Techniques
- Cholangiocarcinoma and Gallbladder Cancer Studies
- Gastric Cancer Management and Outcomes
- Urological Disorders and Treatments
- Cancer Genomics and Diagnostics
- 3D Shape Modeling and Analysis
- Privacy-Preserving Technologies in Data
- AI in cancer detection
UCLouvain
2017-2021
Institute of Information and Communication Technologies
2021
ORCID
2019
In radiation therapy, a CT image is used to manually delineate the organs and plan treatment. During treatment, cone beam (CBCT) often acquired monitor anatomical modifications. For this purpose, automatic organ segmentation on CBCT crucial step. However, manual segmentations are scarce, models trained with data do not generalize well images. We investigate adversarial networks intensity-based augmentation, two strategies leveraging large databases of annotated CTs train neural for CBCT....
Distributed learning across coalitions is becoming popular for multi-centric implementation of deep models. However, the level trust between members a coalition can vary and requires different security architectures. Privacy training data has been largely described in distributed learning. In this paper, we present scalable architecture providing additional features such as validation on sources quality, confidentiality model within trusted or among untrusted partners inside coalition. More...
For prostate cancer patients, large organ deformations occurring between radiotherapy treatment sessions create uncertainty about the doses delivered to tumor and surrounding healthy organs. Segmenting those regions on cone beam CT (CBCT) scans acquired day would reduce such uncertainties. In this work, a 3D U-net deep-learning architecture was trained segment bladder, rectum, CBCT scans. Due scarcity of contoured scans, training set augmented with already in current clinical workflow. Our...
For prostate cancer patients, large organ deformations occurring between the sessions of a fractionated radiotherapy treatment lead to uncertainties in doses delivered tumour and surrounding organs at risk. The segmentation those structures cone beam CT (CBCT) volumes acquired before every session is desired reduce uncertainties. In this work, we perform fully automatic bladder CBCT with u-net, 3D convolutional neural network (FCN). Since annotations are hard collect for volumes, consider...
Purpose: Automation of organ segmentation, via convolutional neural networks (CNNs), is key to facilitate the work medical practitioners by ensuring that adequate radiation dose delivered target area while avoiding harmful exposure healthy organs. The issue with CNNs they require large amounts data transfer and storage which makes use image compression a necessity. Compression will affect quality in turn affects segmentation process. We address dilemma involved handling preserving accuracy....
External beam radiation therapy (EBRT) treats cancer by delivering daily fractions of to a target volume. For prostate cancer, the undergoes day-to-day variations in position, volume, and shape. stereotactic photon for proton EBRT, endorectal balloons (ERBs) can be used limit variations. To date, patterns non-rigid patients with ERB have not been modeled. We extracted modeled patient-specific variations, using regularly acquired CT-images, point cloud registration, principal component...
4121 Background: Intrahepatic cholangiocarcinoma (iCCA) is an aggressive malignancy and the second most common primary liver cancer. About a third of patients may benefit from surgery but recurrence overall survival low. Accurate prognosis modeling that can predict response to treatment outcome remains unmet clinical need. Deep learning methods provide new opportunity better by extracting distinguishing characteristics interrogating multiple data sources, which appears be superior unimodal...