Scientific publications - Median Technologies

Pierre Baudot, PhD [1], Mathieu Bernardi, MscI[1] -Affiliations: [1] Median Technologies, Valbonne, France, in col. with Daniel Bennequin, Monica Tapia, Jean-Marc Goaillard. Corresponding author e-mail: pierre.baudot@mediantechnologies.com

Poster presented at the Geometric Sciences of Information 2019 (GSI2019) conference | Aug 27-29, 2019 |Toulouse (France). Information cohomology is a branch of topological data analysis that allows us to quatify directly statistical dependencies and independences in a given dataset. Theorems establish Shannon entropy as a first cohomology class and mutual information as coboundaries on finite probability space endowed with a random variable chain complex structure. We present some simplicial subcase applications to supervised and unsupervised learning in different contexts: transcriptomic and digits and medical CT image classification.

Yan Liu [1], Corinne Ramos [1], Pierre Baudot [1], Johan Brag [1], Olivier Lucidarme [2] - Affiliations: [1] Median Technologies, Valbonne, France. [2] Radiology Unit, Pitié Salpétrière Hospital, APHP, Paris, France.

Current drug therapies in HCC remain limited because of substantial genomic, cellular and molecular heterogeneity of the liver tumor microenvironment (TME). This heterogeneity has implications for tumor development, immune response and cellular invasion and is compounded by multiple molecular pathways related to the various HCC etiologies. In this context, it is challenging to find biomarkers that are predictive of therapeutic response or outcome. A systems biology approach leveraging the iBiopsy phenotyping platform to automatically detect HCC and TME subtypes using non-invasive medical imaging could help identify specific disease pathways and accelerate HCC drug development.

Vladimir Groza [1], Johan Brag [1], Michael Auffret [1] - Affiliations: [1] Median Technologies, Valbonne, France

Poster presented at The IEEE International Symposium on Biomedical Imaging (ISBI) 2019 | Venice, Italy | April 8-11, 2019
In clinical trials, the evaluation of CT/MRI images is usually done manually or with the use of semi-automatic segmentation techniques. Liver segmentation, as well as segmentations of other organs such as prostate, lung, is a crucial step in computer-aided systems for cancer detection. In order to improve the quality and performance of diagnosis computer-aided systems became popular, where deep learning approach demonstrates its potential and strengths as robust and powerful tool in particular for medical image segmentation. This work presents a novel deep convolutional neural network architecture DualTail-Net in application for automatic liver segmentation on abdominal CT images.

Beaumont H [1], Evans T [2], Klifa C [1], Hons S [3], Chadjaa M [4], Monostori Z [5], Iannessi A [6] - Affiliations: [1] Median Technologies, Valbonne , France. [2] U. of Pennsylvania, USA. [3] Yonsei University of Medicine, Seoul, South Korea. [4] SANOFI, Vitry sur seine, France. [5] National Koranyi Institute, Budapest, Hungary. [6] Centre Antoine Lacassagne , Nice, France.

Oral presentation at the European Congress of Radiology (ECR) on Feb 27-March 3, 2019, Vienna, Austria

Beaumont H [1], Evans T [2], Klifa C [1], Guermazi A [3], Hong S [4], Chadjaa M [5], Monostori Z [6] - Affiliations: [1] Median Technologies, Valbonne, France. [2] Department of medicine, Hospital of the University of Pennsylvania, USA. [3] Quantitative Imaging Center (QIC) Boston University School of Medicine, Boston, USA. [4] Department of Radiology, Severance Hospital Yonsei University of Medicine, Seoul, South Korea. [5] Clinical Research, SANOFI, Paris, France. [6] Radiology, National Koranyi Institute of TB and pulmonology, Budapest, Hungary

In imaging-based clinical trials, it is common practice to perform double reads for each image, discrepant interpretations can result from these two different evaluations. In this study we analyzed discrepancies that occurred between local investigators (LI) and blinded independent central review (BICR) by comparing reader-selected imaging scans and lesions. Our goal was to identify the causes of discrepant declarations of progressive disease (PD) between LI and BICR in a clinical trial.