Post-doc researcher. Machine learning for medical imaging.

Posts by Collection

portfolio

research

Stain-free histology to validate quantitative MRI

Published in ISMRM, 2019

Quantitative MRI (qMRI) is reproducible but often lacks calibration and/or specificity to the underlying microstructure. Light transmission optical histology of stained tissue is a popular method for validation, however, it is hampered by calibration issues and inhomogeneous penetration of staining agents. We propose a method to validate quantitative MRI metrics using stainless histology by utilizing the innate autofluorescence spectra of tissues when excited with ultraviolet laser. We demonstrate a proof-of-concept application of a qMRI validation pipeline on a pig spinal cord section with in vivo and ex vivo qMRI followed by histological autofluorescence microscopy to quantify myelin content.

Gabriel Mangeat, Harris Nami, Nicolas Pinon, Alexandru Foias, Nikola Stikov, Tobias Granberg, Julien Cohen-Adad (2019). "Stain-free histology to validate quantitative MRI" International Society for Magnetic Resonance in Medicine (ISMRM)

Patch vs. global image-based unsupervised anomaly detection in MR brain scans of early Parkinsonian patients

Published in MLCN (in conjonction with MICCAI), 2021

Although neural networks have proven very successful in a number of medical image analysis applications, their use remains difficult when targeting subtle tasks such as the identification of barely visible brain lesions, especially given the lack of annotated datasets. Good candidate approaches are patch-based unsupervised pipelines which have both the advantage to increase the number of input data and to capture local and fine anomaly patterns distributed in the image, while potential inconveniences are the loss of global structural information. We illustrate this trade-off on Parkinson’s disease (PD) anomaly detection comparing the performance of two anomaly detection models based on a spatial auto-encoder (AE) and an adaptation of a patch-fed siamese auto-encoder (SAE). On average, the SAE model performs better, showing that patches may indeed be advantageous.

Verónica Muñoz-Ramírez, Nicolas Pinon, Florence Forbes, Carole Lartizen, Michel Dojat (2021). "Patch vs. global image-based unsupervised anomaly detection in MR brain scans of early Parkinsonian patients" MLCN 2021 (in conjonction with MICCAI).

Improving motion‐mask segmentation in thoracic CT with multiplanar U‐nets

Published in Medical Physics, 2022

Motion‐mask segmentation from thoracic computed tomography (CT) images is the process of extracting the region that encompasses lungs and viscera, where large displacements occur during breathing. It has been shown to help image registration between different respiratory phases. This registration step is, for example, useful for radiotherapy planning or calculating local lung ventilation. Knowing the location of motion discontinuity, that is, sliding motion near the pleura, allows a better control of the registration preventing unrealistic estimates. Nevertheless, existing methods for motion‐mask segmentation are not robust enough to be used in clinical routine. This article shows that it is feasible to overcome this lack of robustness by using a lightweight deep‐learning approach usable on a standard computer, and this even without data augmentation or advanced model design.

Ludmilla Penarrubia, Nicolas Pinon, Emmanuel Roux, Eduardo Enrique Dávila Serrano, Jean‐Christophe Richard, Maciej Orkisz, David Sarrut (2022). "Improving motion‐mask segmentation in thoracic CT with multiplanar U‐nets." Medical Physics.

Brain subtle anomaly detection based on auto-encoders latent space analysis: application to de novo parkinson patients

Published in ISBI, 2023

Neural network-based anomaly detection remains challenging in clinical applications with little or no supervised information and subtle anomalies such as hardly visible brain lesions. Among unsupervised methods, patch-based auto-encoders with their efficient representation power provided by their latent space, have shown good results for visible lesion detection. However, the commonly used reconstruction error criterion may limit their performance when facing less obvious lesions. In this work, we design two alternative detection criteria. They are derived from multivariate analysis and can more directly capture information from latent space representations. Their performance compares favorably with two additional supervised learning methods, on a difficult de novo Parkinson Disease (PD) classification task.

Nicolas Pinon, Geoffroy Oudoumanessah, Robin Trombetta, Michel Dojat, Florence Forbes, Carole Lartizien (2023). "Brain subtle anomaly detection based on auto-encoders latent space analysis: application to de novo parkinson patients." IEEE 20th International Symposium on Biomedical Imaging.

One-Class SVM on siamese neural network latent space for Unsupervised Anomaly Detection on brain MRI White Matter Hyperintensities

Published in MIDL, 2023

Anomaly detection remains a challenging task in neuroimaging when little to no supervision is available and when lesions can be very small or with subtle contrast. Patch-based representation learning has shown powerful representation capacities when applied to industrial or medical imaging and outlier detection methods have been applied successfully to these images. In this work, we propose an unsupervised anomaly detection (UAD) method based on a latent space constructed by a siamese patch-based auto-encoder and perform the outlier detection with a One-Class SVM training paradigm tailored to the lesion detection task in multi-modality neuroimaging. We evaluate performances of this model on a public database, the White Matter Hyperintensities (WMH) challenge and show in par performance with the two best performing state-of-the-art methods reported so far.

Nicolas Pinon, Robin Trombetta, Carole Lartizien (2023). "One-Class SVM on siamese neural network latent space for Unsupervised Anomaly Detection on brain MRI White Matter Hyperintensities." Medical Imaging with Deep Learning.

Anomaly detection in image or latent space of patch-based auto-encoders for industrial image analysis

Published in GRETSI, 2023

We study several methods for detecting anomalies in color images, constructed on patch-based auto-encoders. Wecompare the performance of three types of methods based, first, on the error between the original image and its reconstruction,second, on the support estimation of the normal image distribution in the latent space, and third, on the error between the originalimage and a restored version of the reconstructed image. These methods are evaluated on the industrial image database MVTecADand compared to two competitive state-of-the-art methods.

Nicolas Pinon, Robin Trombetta, Carole Lartizien (2023). "Anomaly detection in image or latent space of patch-based auto-encoders for industrial image analysis" XXIXème Colloque Francophone de Traitement du Signal et des Images.

Unsupervised anomaly detection in neuroimaging: contributions to representation learning and density support estimation in the latent space

Published in HAL (PhD Thesis), 2024

This PhD thesis covers the topic of deep unsupervised anomaly detection (UAD) in neuroimaging. This research is partially grounded on the UAD model that was proposed in [Alaverdyan, MEDIA 2020], whose novelty was to perform the detection step in the latent representation space by adjusting density support model of the normative distribution. This model developed was applied to the detection of subtle (MRI negative) epileptogenic zones in multiparametric MRI and evaluated on a private database. As a first part of this PhD, we optimize the architecture and hyperparameter setting of this UAD model, and evaluate its performance on different open datasets, including the non medical MVTec anomaly detection [Pinon, GRETSI 2023], the WMH challenge, and the Parkinson’s Progression Markers Initiative database [Ramirez, Pinon, MLCN 2021][Pinon, ISBI 2023]. This allows comparison with state of the art UAD methods, especially with the most common methods based on reconstruction error in the image space. The second main phase of this PhD work is to build on the limits of this model [Alaverdyan, MEDIA 2020] and propose original methodological contributions to 1) design patient specific models, relaxing the strong constraint to accurately coregister all control subjects and patients [Pinon, MIDL 2023], 2) provide a probabilistic detection framework to enable ensemble learning and score map uniformization, 3) fuse the representation learning step and the outlier detection step, by proposing a novel deep learning model.

Nicolas Pinon (2024). "Unsupervised anomaly detection in neuroimaging: Contributions to representation learning and density support estimation in the latent space" Doctoral dissertation, INSA Lyon.

GAN-based synthetic FDG PET images from T1 brain MRI can serve to improve performance of deep unsupervised anomaly detection models

Published in Computer Methods and Programs in Biomedicine, 2025

Research in the cross-modal medical image translation domain has been very productive over the past few years in tackling the scarce availability of large curated multi- modality datasets with the promising performance of GAN-based architectures. However, only a few of these studies assessed task-based related performance of these synthetic data, especially for the training of deep models. We design and compare different GAN-based frameworks for generating synthetic brain [18F]fluorodeoxyglucose (FDG) PET images from T1 weighted MRI data. We first perform standard qualitative and quantitative visual quality evaluation. Then, we explore further impact of using these fake PET data in the training of a deep unsupervised anomaly detection (UAD) model designed to detect subtle epilepsy lesions in T1 MRI and FDG PET images. We introduce novel diagnostic task-oriented quality metrics of the synthetic FDG PET data tailored to our unsupervised detection task, then use these fake data to train a use case UAD model combining a deep representation learning based on siamese autoencoders with a OC-SVM density support estimation model. This model is trained on normal subjects only and allows the detection of any variation from the pattern of the normal population. We compare the detection performance of models trained on 35 paired real MR T1 of normal subjects paired either on 35 true PET images or on 35 synthetic PET images generated from the best performing generative models. Performance analysis is conducted on 17 exams of epilepsy patients undergoing surgery. The best performing GAN-based models allow generating realistic fake PET images of control subject with SSIM and PSNR values around 0.9 and 23.8, respectively and in distribution (ID) with regard to the true control dataset. The best UAD model trained on these synthetic normative PET data allows reaching 74% sensitivity. Our results confirm that GAN-based models are the best suited for MR T1 to FDG PET translation, outperforming transformer or diffusion models. We also demonstrate the diagnostic value of these synthetic data for the training of UAD models and evaluation on clinical exams of epilepsy patients. Our code and the normative image dataset are available.

Daria Zotova, Nicolas Pinon, Robin Trombetta, Romain Bouet, Julien Jung, Carole Lartizien (2025). "GAN-based synthetic FDG PET images from T1 brain MRI can serve to improve performance of deep unsupervised anomaly detection models." Computer Methods and Programs in Biomedicine.

OCSVM-Guided Representation Learning for Unsupervised Anomaly Detection

Published in Preprint, submitted to IEEE Transaction on Image Processing, 2025

Unsupervised anomaly detection (UAD) aims to detect anomalies without labeled data, a necessity in many machine learning applications where anomalous samples are rare or not available. Most state-of-the-art methods fall into two categories: reconstruction-based approaches, which often reconstruct anomalies too well, and decoupled representation learning with density estimators, which can suffer from suboptimal feature spaces. While some recent methods attempt to couple feature learning and anomaly detection, they often rely on surrogate objectives, restrict kernel choices, or introduce approximations that limit their expressiveness and robustness. To address this challenge, we propose a novel method that tightly couples representation learning with an analytically solvable One-Class SVM (OCSVM), through a custom loss formulation that directly aligns latent features with the OCSVM decision boundary. The model is evaluated on two tasks: a new benchmark based on MNIST-C, and a challenging brain MRI subtle lesion detection task. Unlike most methods that focus on large, hyperintense lesions at the image level, our approach succeeds to target small, non-hyperintense lesions, while we evaluate voxel-wise metrics, addressing a more clinically relevant scenario. Both experiments evaluate a form of robustness to domain shifts, including corruption types in MNIST-C and scanner/age variations in MRI. Results demonstrate performance and robustness of our proposed model, highlighting its potential for general UAD and real-world medical imaging applications. The source code is available at https://github.com/Nicolas-Pinon/uad_ocsvm_guided_repr_learning.

Nicolas Pinon, Carole Lartizien (2025). "OCSVM-Guided Representation Learning for Unsupervised Anomaly Detection." Preprint, submitted to IEEE Transaction on Image Processing.

talks

teaching

Mentorship for middle school students (Maths, Physics)
[20h] Mentorat, Collège Victor Hugo, Cachan (Coordées de la réussite), 2017-2018

Mentorship of a small group of middle school students in mathematics and physics. Discovery of basic principle of ultrasounds, informations about higher education.

Deep learning for medical imaging
[12h] École de printemps, DLMI Spring school, 2020-2022

Practical work assistant for the pring DLMI spring school. Classification and segmentation of medical images using common convolutional neural networks (U-net, ResNet, VAE, etc.). Analysis and processing specific to medical imaging (spacing, modalities, contrast, etc.).

Electricity and electromagnetism
[23h] TP niveau L2, INSA Lyon FIMI, 2020-2022

Undergraduate practial course about electricity and electromagnetism. Analysis of basic electricity circuits, filters, Fourier analysis, oscilloscope usage. Analysis of electric and magnetic fields, modelisation with the Maxwell’s equations.

Electromagnetism and waves
[83h] TD niveau L2, INSA Lyon FIMI, 2020-2022

Undergraduate course about fundamental of electromagnetism: Maxwell’s equations, Gauss’ law, Ampère’s circuital law, electromagnetic waves, sound waves. Analytical resolution of basic electromagnetism problems.

Image processing projects
[16h] TP niveau L2, INSA Lyon FIMI, 2020-2022

Undergraduate practial projects using image processing.

Machine learning and Deep learning
[8h] TP niveau M2, CPE Lyon, 2022-2023

Graduate practial course about machine learning and deep learning. Basic machine learning practice (cross-validation, hyperparameter testing, etc.), SVM, random forests, classification, regression. Preliminary deep learning practice (CNN, ResNet, over/under-fitting, evaluation metrics).

Algorithms and programming
[35h] TD niveau L1, INSA Lyon FIMI, 2022-2023

Undergraduate course about algorithm and programming in Python: loops, lists, conditions, sorting, number representations, cellular automata, etc.

Image processing
[16h] Cours intégré niveau L2, INSA Lyon FIMI, 2024-2024

Undergraduate course about image processing. Histogram normalization, filters, Fourier transform, mathematical morphology, edge detection, etc.