I am

Ludmilla Penarrubia

A signal and image processing PhD

I use artificial intelligence to extract features from data, in particular from images. I am particularly sensitive to projects with a high societal impact, especially health-related projects.

In short

I am currently a post-doctoral researcher in medical image processing at CREATIS laboratory. Daily, I code and use deep learning models in Python language with PyTorch library for 3D segmentation of computed tomography (CT) images. Here are a few skills and tools I use:

Python
PyTorch
Plotly
Dash
uv
Deep learning
SimpleITK

Experience

CREATIS
Capgemini Engineering
IRCAD
Seoul National University
Orange Caraïbe

Post-doctoral researcher - CREATIS

Sept 2024 - currently Lyon, France

I am currently a post-doctoral researcher at CREATIS lab in Lyon (France). My project entitled Artificial Intelligence for Image-Assisted Ventilation aims at improving robustness and generalisability of 3D segmentation models developped during my PhD thesis.

To achieve this, I incorporated the work of previous interns into the segmentation learning process. I developed a new data augmentation method based on re-calibration, which improved the generalizability of my segmentation models.

Activities :

3D segmentation, data augmentation through synthesis of pathological images
Management, description, and preprocessing of large medical image databases (over 1,000 patients, over 1,500 images)

Tools and skills used :

PhD student

Oct 2019 - Dec 2022 Lyon, France

My thesis topic, entitled Quantification of pulmonary aeration on CT images of patients with acute respiratory distress syndrome, aimed to automate the task of segmenting lungs on 3D pathological images. To do this, I used deep learning with convolutional neural networks on 2D or 3D images.

Activities :

Deep learning segmentation (U-Net, PyTorch, Python) on 2D and 3D medical images
Management of medical image databases (quality control, pre-processing)
Consideration of physiological elements in the cost function of the segmentation model
Integration of the segmentation model into software used in hospitals for preclinical research

Tools and skills used :

Data Scientist - Capgemini Engineering

Jan 2023 - Sept 2024 Lyon, France

I joined the Research and Innovation team (Auvergne-Rhône-Alpes) as a technical advisor on data science. I have been particularly involved in projects related to psychometrics, image processing, and natural language processing (NLP). All these projects have allowed me to apply these skills to different sectors such as the medical sector (treatment and/or diagnosis of endometriosis, early detection of schizophrenia), and the industrial sector (drafting and classification of incident reports).

Activities :

Detection of uterosacral ligaments on MRI images (2D)
Clustering of tabular and textual data
Automatic detection of bacteria on Petri dishes
Detection of doors on architectural plans
Technical supervision (1 M2 intern, 2 work-study students undergoing retraining)

Tools and skills used :

Research intern - IRCAD

March 2019 - August 2019 Strasbourg, France

I completed my final year project within the Research and Development team at the Digestive Cancer Research Institute (IRCAD). This project, entitled Real-time monitoring of structures of interest in transabdominal ultrasound imaging, allowed me to conduct a test bench and develop my skills in illustrating results.

Activities :

Image sequence processing for estimating movement caused by breathing on organs of interest
Test bench for various tracking, recalibration, and detection methods

Tools and skills used :

Research intern - Seoul National University

June 2018 - August 2018 Seoul, South Korea

I joined a research project at the Neurophysiology Laboratory of Seoul National University (SNU) led by a doctoral student and entitled In vivo calcium imaging of neuronal activity in the hippocampus of freely moving mice. I have mainly worked on behavioral observation videos of mice, in which I used segmentation and mathematical morphology to detect the position of the mice. From the information extracted from the video sequences, I was also able to estimate the activity of the mice.

Activities :

Image processing on videos
Assistance with the preparation, surgery, and postmortem analysis of mice
Proofreading of articles

Tools and skills used :

Software development intern - Orange Caraïbe

June 2017 - July 2017 Guadeloupe, France

This internship focusing on Infrastructure as code allowed me to develop my programming skills and learn more about telecommunications and virtualization.

Tools and skills used :

Education

2019 - 2022

PhD in signal and image processing

Université Claude Bernard Lyon 1

Lyon, France

Title : Quantification of pulmonary ventilation on CT images of patients with acute respiratory distress syndrome

Acquired skills :

Image processing
Deep learning
Reproductible science
Scientific article writing

My PhD thesis

2017 - 2019

MSc Imaging, Robotics and Engineering for Health

Télécom Physique Strasbourg

Strasbourg, France

Acquired skills :

Image processing
Biomechanics
Physics of medical imaging devices
Computer science
Finite element modeling

Individual research project (3 months - 2018) : Effects of a seat belt on the rib cage during a head-on car collision :

Finite element modeling (Altrair Hyperworks)
Returned in the form of a scientific article

2016 - 2019

General Engineering Degree (equivalent to MSc)

Télécom Physique Strasbourg

Strasbourg, France

Speciality: Engineering and Sciences for Health

Acquired skills :

Image processing
Computer science
Signal processing
Automatics

TOEIC test (2018) : 960 / 990

Engineering student project for CNES (Centre National d’Etudes Spatiales - National Center for Space Studies) (2017 - 2018) : Image processing applied to a portable bio-contamination monitoring device: