Gwendoline de Bie

Sylvain Bodard, Pierre Baudot, Benjamin Renoust et al.

Early detection of malignant lung nodules remains limited by reliance on size- and growth-based screening criteria, which can delay diagnosis. We present an integrated AI system that - unlike conventional CADe or CADx approaches - jointly performs nodule detection and malignancy assessment directly at the nodule level from low-dose CT scans within a unified aided decision framework. To address limitations in dataset scale and explainability, we designed an ensemble of shallow deep learning and feature-based specialized models, trained and evaluated on 25,709 scans with 69,449 annotated nodules, with external validation on an independent cohort. The system achieves an area under the receiver operating characteristic curve (AUC) of 0.98 internally and 0.945 on an independent cohort, outperforming radiologists and leading AI models (Sybil, Brock, Google, Kaggle). With a sensitivity of 99.3 percent at 0.5 false positives per scan, it addresses key barriers to AI adoption and demonstrates improved performance relative to both Lung-RADS size-based triage and European volume- and VDT-based screening criteria. The model outperforms radiologists across all nodule sizes and cancer stages - excelling in stage I cancers - and across all growth-based metrics, including volume-doubling time. It also surpasses radiologists by up to one year in diagnosing indeterminate and slow-growing nodules.

Gwendoline De Bie, Herilalaina Rakotoarison, Gabriel Peyré et al.

Recent advances in deep learning from probability distributions successfully achieve classification or regression from distribution samples, thus invariant under permutation of the samples. The first contribution of the paper is to extend these neural architectures to achieve invariance under permutation of the features, too. The proposed architecture, called Dida, inherits the NN properties of universal approximation, and its robustness w.r.t. Lipschitz-bounded transformations of the input distribution is established. The second contribution is to empirically and comparatively demonstrate the merits of the approach on two tasks defined at the dataset level. On both tasks, Dida learns meta-features supporting the characterization of a (labelled) dataset. The first task consists of predicting whether two dataset patches are extracted from the same initial dataset. The second task consists of predicting whether the learning performance achieved by a hyper-parameter configuration under a fixed algorithm (ranging in k-NN, SVM, logistic regression and linear classifier with SGD) dominates that of another configuration, for a dataset extracted from the OpenML benchmarking suite. On both tasks, Dida outperforms the state of the art: DSS (Maron et al., 2020) and Dataset2Vec (Jomaa et al., 2019) architectures, as well as the models based on the hand-crafted meta-features of the literature.

Gwendoline de Bie, Gabriel Peyré, Marco Cuturi

Machine learning is increasingly targeting areas where input data cannot be accurately described by a single vector, but can be modeled instead using the more flexible concept of random vectors, namely probability measures or more simply point clouds of varying cardinality. Using deep architectures on measures poses, however, many challenging issues. Indeed, deep architectures are originally designed to handle fixedlength vectors, or, using recursive mechanisms, ordered sequences thereof. In sharp contrast, measures describe a varying number of weighted observations with no particular order. We propose in this work a deep framework designed to handle crucial aspects of measures, namely permutation invariances, variations in weights and cardinality. Architectures derived from this pipeline can (i) map measures to measures - using the concept of push-forward operators; (ii) bridge the gap between measures and Euclidean spaces - through integration steps. This allows to design discriminative networks (to classify or reduce the dimensionality of input measures), generative architectures (to synthesize measures) and recurrent pipelines (to predict measure dynamics). We provide a theoretical analysis of these building blocks, review our architectures' approximation abilities and robustness w.r.t. perturbation, and try them on various discriminative and generative tasks.