Christian Desrosiers

Mehrdad Noori, Milad Cheraghalikhani, Ali Bahri et al.

Standard deep learning models such as convolutional neural networks (CNNs) lack the ability of generalizing to domains which have not been seen during training. This problem is mainly due to the common but often wrong assumption of such models that the source and target data come from the same i.i.d. distribution. Recently, Vision Transformers (ViTs) have shown outstanding performance for a broad range of computer vision tasks. However, very few studies have investigated their ability to generalize to new domains. This paper presents a first Token-level Feature Stylization (TFS-ViT) approach for domain generalization, which improves the performance of ViTs to unseen data by synthesizing new domains. Our approach transforms token features by mixing the normalization statistics of images from different domains. We further improve this approach with a novel strategy for attention-aware stylization, which uses the attention maps of class (CLS) tokens to compute and mix normalization statistics of tokens corresponding to different image regions. The proposed method is flexible to the choice of backbone model and can be easily applied to any ViT-based architecture with a negligible increase in computational complexity. Comprehensive experiments show that our approach is able to achieve state-of-the-art performance on five challenging benchmarks for domain generalization, and demonstrate its ability to deal with different types of domain shifts. The implementation is available at: https://github.com/Mehrdad-Noori/TFS-ViT_Token-level_Feature_Stylization.

Mélanie Gaillochet, Christian Desrosiers, Hervé Lombaert

The performance of learning-based algorithms improves with the amount of labelled data used for training. Yet, manually annotating data is particularly difficult for medical image segmentation tasks because of the limited expert availability and intensive manual effort required. To reduce manual labelling, active learning (AL) targets the most informative samples from the unlabelled set to annotate and add to the labelled training set. On the one hand, most active learning works have focused on the classification or limited segmentation of natural images, despite active learning being highly desirable in the difficult task of medical image segmentation. On the other hand, uncertainty-based AL approaches notoriously offer sub-optimal batch-query strategies, while diversity-based methods tend to be computationally expensive. Over and above methodological hurdles, random sampling has proven an extremely difficult baseline to outperform when varying learning and sampling conditions. This work aims to take advantage of the diversity and speed offered by random sampling to improve the selection of uncertainty-based AL methods for segmenting medical images. More specifically, we propose to compute uncertainty at the level of batches instead of samples through an original use of stochastic batches (SB) during sampling in AL. Stochastic batch querying is a simple and effective add-on that can be used on top of any uncertainty-based metric. Extensive experiments on two medical image segmentation datasets show that our strategy consistently improves conventional uncertainty-based sampling methods. Our method can hence act as a strong baseline for medical image segmentation. The code is available on: https://github.com/Minimel/StochasticBatchAL.git.

Mélanie Gaillochet, Christian Desrosiers, Hervé Lombaert

Deep learning methods typically depend on the availability of labeled data, which is expensive and time-consuming to obtain. Active learning addresses such effort by prioritizing which samples are best to annotate in order to maximize the performance of the task model. While frameworks for active learning have been widely explored in the context of classification of natural images, they have been only sparsely used in medical image segmentation. The challenge resides in obtaining an uncertainty measure that reveals the best candidate data for annotation. This paper proposes Test-time Augmentation for Active Learning (TAAL), a novel semi-supervised active learning approach for segmentation that exploits the uncertainty information offered by data transformations. Our method applies cross-augmentation consistency during training and inference to both improve model learning in a semi-supervised fashion and identify the most relevant unlabeled samples to annotate next. In addition, our consistency loss uses a modified version of the JSD to further improve model performance. By relying on data transformations rather than on external modules or simple heuristics typically used in uncertainty-based strategies, TAAL emerges as a simple, yet powerful task-agnostic semi-supervised active learning approach applicable to the medical domain. Our results on a publicly-available dataset of cardiac images show that TAAL outperforms existing baseline methods in both fully-supervised and semi-supervised settings. Our implementation is publicly available on https://github.com/melinphd/TAAL.

Mehrdad Noori, Milad Cheraghalikhani, Ali Bahri et al.

Domain Generalization techniques aim to enhance model robustness by simulating novel data distributions during training, typically through various augmentation or stylization strategies. However, these methods frequently suffer from limited control over the diversity of generated images and lack assurance that these images span distinct distributions. To address these challenges, we propose FDS, Feedback-guided Domain Synthesis, a novel strategy that employs diffusion models to synthesize novel, pseudo-domains by training a single model on all source domains and performing domain mixing based on learned features. By incorporating images that pose classification challenges to models trained on original samples, alongside the original dataset, we ensure the generation of a training set that spans a broad distribution spectrum. Our comprehensive evaluations demonstrate that this methodology sets new benchmarks in domain generalization performance across a range of challenging datasets, effectively managing diverse types of domain shifts. The code can be found at: \url{https://github.com/Mehrdad-Noori/FDS.git}.

Djebril Mekhazni, Maximilien Dufau, Christian Desrosiers et al.

Systems for person re-identification (ReID) can achieve a high accuracy when trained on large fully-labeled image datasets. However, the domain shift typically associated with diverse operational capture conditions (e.g., camera viewpoints and lighting) may translate to a significant decline in performance. This paper focuses on unsupervised domain adaptation (UDA) for video-based ReID - a relevant scenario that is less explored in the literature. In this scenario, the ReID model must adapt to a complex target domain defined by a network of diverse video cameras based on tracklet information. State-of-art methods cluster unlabeled target data, yet domain shifts across target cameras (sub-domains) can lead to poor initialization of clustering methods that propagates noise across epochs, thus preventing the ReID model to accurately associate samples of same identity. In this paper, an UDA method is introduced for video person ReID that leverages knowledge on video tracklets, and on the distribution of frames captured over target cameras to improve the performance of CNN backbones trained using pseudo-labels. Our method relies on an adversarial approach, where a camera-discriminator network is introduced to extract discriminant camera-independent representations, facilitating the subsequent clustering. In addition, a weighted contrastive loss is proposed to leverage the confidence of clusters, and mitigate the risk of incorrect identity associations. Experimental results obtained on three challenging video-based person ReID datasets - PRID2011, iLIDS-VID, and MARS - indicate that our proposed method can outperform related state-of-the-art methods. Our code is available at: \url{https://github.com/dmekhazni/CAWCL-ReID}

Mélanie Gaillochet, Christian Desrosiers, Hervé Lombaert

Foundation models such as the recently introduced Segment Anything Model (SAM) have achieved remarkable results in image segmentation tasks. However, these models typically require user interaction through handcrafted prompts such as bounding boxes, which limits their deployment to downstream tasks. Adapting these models to a specific task with fully labeled data also demands expensive prior user interaction to obtain ground-truth annotations. This work proposes to replace conditioning on input prompts with a lightweight module that directly learns a prompt embedding from the image embedding, both of which are subsequently used by the foundation model to output a segmentation mask. Our foundation models with learnable prompts can automatically segment any specific region by 1) modifying the input through a prompt embedding predicted by a simple module, and 2) using weak labels (tight bounding boxes) and few-shot supervision (10 samples). Our approach is validated on MedSAM, a version of SAM fine-tuned for medical images, with results on three medical datasets in MR and ultrasound imaging. Our code is available on https://github.com/Minimel/MedSAMWeakFewShotPromptAutomation.

David Osowiechi, Gustavo A. Vargas Hakim, Mehrdad Noori et al.

A major problem of deep neural networks for image classification is their vulnerability to domain changes at test-time. Recent methods have proposed to address this problem with test-time training (TTT), where a two-branch model is trained to learn a main classification task and also a self-supervised task used to perform test-time adaptation. However, these techniques require defining a proxy task specific to the target application. To tackle this limitation, we propose TTTFlow: a Y-shaped architecture using an unsupervised head based on Normalizing Flows to learn the normal distribution of latent features and detect domain shifts in test examples. At inference, keeping the unsupervised head fixed, we adapt the model to domain-shifted examples by maximizing the log likelihood of the Normalizing Flow. Our results show that our method can significantly improve the accuracy with respect to previous works.

Gustavo A. Vargas Hakim, David Osowiechi, Mehrdad Noori et al.

Deep Learning models have shown remarkable performance in a broad range of vision tasks. However, they are often vulnerable against domain shifts at test-time. Test-time training (TTT) methods have been developed in an attempt to mitigate these vulnerabilities, where a secondary task is solved at training time simultaneously with the main task, to be later used as an self-supervised proxy task at test-time. In this work, we propose a novel unsupervised TTT technique based on the maximization of Mutual Information between multi-scale feature maps and a discrete latent representation, which can be integrated to the standard training as an auxiliary clustering task. Experimental results demonstrate competitive classification performance on different popular test-time adaptation benchmarks.

Ahmad Chaddad, Lama Hassan, Christian Desrosiers

This paper proposes to encode the distribution of features learned from a convolutional neural network using a Gaussian Mixture Model. These parametric features, called GMM-CNN, are derived from chest computed tomography and X-ray scans of patients with Coronavirus Disease 2019. We use the proposed GMM-CNN features as input to a robust classifier based on random forests to differentiate between COVID-19 and other pneumonia cases. Our experiments assess the advantage of GMM-CNN features compared to standard CNN classification on test images. Using a random forest classifier (80\% samples for training; 20\% samples for testing), GMM-CNN features encoded with two mixture components provided a significantly better performance than standard CNN classification (p\,$<$\,0.05). Specifically, our method achieved an accuracy in the range of 96.00\,--\,96.70\% and an area under the ROC curve in the range of 99.29\,--\,99.45\%, with the best performance obtained by combining GMM-CNN features from both computed tomography and X-ray images. Our results suggest that the proposed GMM-CNN features could improve the prediction of COVID-19 in chest computed tomography and X-ray scans.

Ahmad Chaddad, Paul Daniel Mingli Zhang, Saima Rathore et al.

Imaging biomarkers offer a non-invasive way to predict the response of immunotherapy prior to treatment. In this work, we propose a novel type of deep radiomic features (DRFs) computed from a convolutional neural network (CNN), which capture tumor characteristics related to immune cell markers and overall survival. Our study uses four MRI sequences (T1-weighted, T1-weighted post-contrast, T2-weighted and FLAIR) with corresponding immune cell markers of 151 patients with brain tumor. The proposed method extracts a total of 180 DRFs by aggregating the activation maps of a pre-trained 3D-CNN within labeled tumor regions of MRI scans. These features offer a compact, yet powerful representation of regional texture encoding tissue heterogeneity. A comprehensive set of experiments is performed to assess the relationship between the proposed DRFs and immune cell markers, and measure their association with overall survival. Results show a high correlation between DRFs and various markers, as well as significant differences between patients grouped based on these markers. Moreover, combining DRFs, clinical features and immune cell markers as input to a random forest classifier helps discriminate between short and long survival outcomes, with AUC of 72\% and p=2.36$\times$10$^{-5}$. These results demonstrate the usefulness of proposed DRFs as non-invasive biomarker for predicting treatment response in patients with brain tumors.

Farzad Beizaee, Christian Desrosiers, Gregory A. Lodygensky et al.

In this paper, we propose an unsupervised framework based on normalizing flows that harmonizes MR images to mimic the distribution of the source domain. The proposed framework consists of three steps. First, a shallow harmonizer network is trained to recover images of the source domain from their augmented versions. A normalizing flow network is then trained to learn the distribution of the source domain. Finally, at test time, a harmonizer network is modified so that the output images match the source domain's distribution learned by the normalizing flow model. Our unsupervised, source-free and task-independent approach is evaluated on cross-domain brain MRI segmentation using data from four different sites. Results demonstrate its superior performance compared to existing methods.

Julien Nicolas, Florent Chiaroni, Imtiaz Ziko et al.

Despite the recent progress in incremental learning, addressing catastrophic forgetting under distributional drift is still an open and important problem. Indeed, while state-of-the-art domain incremental learning (DIL) methods perform satisfactorily within known domains, their performance largely degrades in the presence of novel domains. This limitation hampers their generalizability, and restricts their scalability to more realistic settings where train and test data are drawn from different distributions. To address these limitations, we present a novel DIL approach based on a mixture of prompt-tuned CLIP models (MoP-CLIP), which generalizes the paradigm of S-Prompting to handle both in-distribution and out-of-distribution data at inference. In particular, at the training stage we model the features distribution of every class in each domain, learning individual text and visual prompts to adapt to a given domain. At inference, the learned distributions allow us to identify whether a given test sample belongs to a known domain, selecting the correct prompt for the classification task, or from an unseen domain, leveraging a mixture of the prompt-tuned CLIP models. Our empirical evaluation reveals the poor performance of existing DIL methods under domain shift, and suggests that the proposed MoP-CLIP performs competitively in the standard DIL settings while outperforming state-of-the-art methods in OOD scenarios. These results demonstrate the superiority of MoP-CLIP, offering a robust and general solution to the problem of domain incremental learning.

Ghazal Danaee, Mélanie Gaillochet, Christian Desrosiers et al.

Active learning (AL) has emerged as a crucial strategy for reducing the prohibitive costs associated with medical image segmentation. However, standard uncertainty-based AL methods typically focus on maximizing performance metrics, ignoring performance disparities or fairness across groups with sensitive attributes. While fair active learning has been explored in classification tasks, its intersection with medical image segmentation remains unaddressed. In this work, we introduced a fairness-aware active learning framework with a Weighted Entropy selection strategy that modulates uncertainty based on current group-specific performance estimates on the labeled set. To decouple true epistemic uncertainty from anatomical volume variances, we further utilized a masked, scaled entropy restricted to the region of interest. The framework was evaluated on synthetic T1-weighted brain MRIs with controlled left caudate bias in both strong and weak bias settings. A 3D U-Net was trained to segment the left caudate under several AL strategies, starting from both demographically balanced and strongly imbalanced initial labeled sets. Experiments demonstrated that our method markedly reduces performance disparities between groups compared to random sampling and standard uncertainty sampling. By prioritizing poorly segmented subgroups during the AL cycles, our method consistently achieved the highest equity-scaled performance and reduced the disparity metric by 75% (strong bias) and 86% (weak bias) relative to standard entropy at the final budget. Overall, this work is among the first studies on fair AL for medical image segmentation, offering an efficient strategy to train more equitable models in resource-constrained environments.

Kunal Mahatha, Jose Dolz, Christian Desrosiers

Despite recent advances in Open-Vocabulary Semantic Segmentation (OVSS), existing training-free methods face several limitations: use of computationally expensive affinity refinement strategies, ineffective fusion of transformer attention maps due to equal weighting or reliance on fixed-size Gaussian kernels to reinforce local spatial smoothness, enforcing isotropic neighborhoods. We propose a strong baseline for training-free OVSS termed as NERVE (Neighbourhood \& Entropy-guided Random-walk for open-Vocabulary sEgmentation), which uniquely integrates global and fine-grained local information, exploiting the neighbourhood structure from the self-attention layer of a stable diffusion model. We also introduce a stochastic random walk for refining the affinity rather than relying on fixed-size Gaussian kernels for local context. This spatial diffusion process encourages propagation across connected and semantically related areas, enabling it to effectively delineate objects with arbitrary shapes. Whereas most existing approaches treat self-attention maps from different transformer heads or layers equally, our method uses entropy-based uncertainty to select the most relevant maps. Notably, our method does not require any conventional post-processing techniques like Conditional Random Fields (CRF) or Pixel-Adaptive Mask Refinement (PAMR). Experiments are performed on 7 popular semantic segmentation benchmarks, yielding an overall state-of-the-art zero-shot segmentation performance, providing an effective approach to open-vocabulary semantic segmentation.

Gustavo Adolfo Vargas Hakim, David Osowiechi, Mehrdad Noori et al.

Pre-trained vision-language models (VLMs), exemplified by CLIP, demonstrate remarkable adaptability across zero-shot classification tasks without additional training. However, their performance diminishes in the presence of domain shifts. In this study, we introduce CLIP Adaptation duRing Test-Time (CLIPArTT), a fully test-time adaptation (TTA) approach for CLIP, which involves automatic text prompts construction during inference for their use as text supervision. Our method employs a unique, minimally invasive text prompt tuning process, wherein multiple predicted classes are aggregated into a single new text prompt, used as \emph{pseudo label} to re-classify inputs in a transductive manner. Additionally, we pioneer the standardization of TTA benchmarks (e.g., TENT) in the realm of VLMs. Our findings demonstrate that, without requiring additional transformations nor new trainable modules, CLIPArTT enhances performance dynamically across non-corrupted datasets such as CIFAR-100, corrupted datasets like CIFAR-100-C and ImageNet-C, alongside synthetic datasets such as VisDA-C. This research underscores the potential for improving VLMs' adaptability through novel test-time strategies, offering insights for robust performance across varied datasets and environments. The code can be found at: https://github.com/dosowiechi/CLIPArTT.git

Ahmad Chaddad, Yihang Wu, Yuchen Jiang et al.

Traditional machine learning assumes that training and test sets are derived from the same distribution; however, this assumption does not always hold in practical applications. This distribution disparity can lead to severe performance drops when the trained model is used in new data sets. Domain adaptation (DA) is a machine learning technique that aims to address this problem by reducing the differences between domains. This paper presents simulation-based algorithms of recent DA techniques, mainly related to unsupervised domain adaptation (UDA), where labels are available only in the source domain. Our study compares these techniques with public data sets and diverse characteristics, highlighting their respective strengths and drawbacks. For example, Safe Self-Refinement for Transformer-based DA (SSRT) achieved the highest accuracy (91.6\%) in the office-31 data set during our simulations, however, the accuracy dropped to 72.4\% in the Office-Home data set when using limited batch sizes. In addition to improving the reader's comprehension of recent techniques in DA, our study also highlights challenges and upcoming directions for research in this domain. The codes are available at https://github.com/AIPMLab/Domain_Adaptation.

Farzad Beizaee, Gregory A. Lodygensky, Christian Desrosiers et al.

Recent advances in diffusion models have spurred research into their application for Reconstruction-based unsupervised anomaly detection. However, these methods may struggle with maintaining structural integrity and recovering the anomaly-free content of abnormal regions, especially in multi-class scenarios. Furthermore, diffusion models are inherently designed to generate images from pure noise and struggle to selectively alter anomalous regions of an image while preserving normal ones. This leads to potential degradation of normal regions during reconstruction, hampering the effectiveness of anomaly detection. This paper introduces a reformulation of the standard diffusion model geared toward selective region alteration, allowing the accurate identification of anomalies. By modeling anomalies as noise in the latent space, our proposed Deviation correction diffusion (DeCo-Diff) model preserves the normal regions and encourages transformations exclusively on anomalous areas. This selective approach enhances the reconstruction quality, facilitating effective unsupervised detection and localization of anomaly regions. Comprehensive evaluations demonstrate the superiority of our method in accurately identifying and localizing anomalies in complex images, with pixel-level AUPRC improvements of 11-14% over state-of-the-art models on well known anomaly detection datasets. The code is available at https://github.com/farzad-bz/DeCo-Diff

Yihang Wu, Ahmad Chaddad, Christian Desrosiers et al.

Despite the remarkable performance of vision language models (VLMs) such as Contrastive Language Image Pre-training (CLIP), the large size of these models is a considerable obstacle to their use in federated learning (FL) systems where the parameters of local client models need to be transferred to a global server for aggregation. Another challenge in FL is the heterogeneity of data from different clients, which affects the generalization performance of the solution. In addition, natural pre-trained VLMs exhibit poor generalization ability in the medical datasets, suggests there exists a domain gap. To solve these issues, we introduce a novel method for the Federated Adversarial Adaptation (FAA) of CLIP. Our method, named FAA-CLIP, handles the large communication costs of CLIP using a light-weight feature adaptation module (FAM) for aggregation, effectively adapting this VLM to each client's data while greatly reducing the number of parameters to transfer. By keeping CLIP frozen and only updating the FAM parameters, our method is also computationally efficient. Unlike existing approaches, our FAA-CLIP method directly addresses the problem of domain shifts across clients via a domain adaptation (DA) module. This module employs a domain classifier to predict if a given sample is from the local client or the global server, allowing the model to learn domain-invariant representations. Extensive experiments on six different datasets containing both natural and medical images demonstrate that FAA-CLIP can generalize well on both natural and medical datasets compared to recent FL approaches. Our codes are available at https://github.com/AIPMLab/FAA-CLIP.

Mehrdad Noori, David Osowiechi, Gustavo Adolfo Vargas Hakim et al.

Recently, test-time adaptation has attracted wide interest in the context of vision-language models for image classification. However, to the best of our knowledge, the problem is completely overlooked in dense prediction tasks such as Open-Vocabulary Semantic Segmentation (OVSS). In response, we propose a novel TTA method tailored to adapting VLMs for segmentation during test time. Unlike TTA methods for image classification, our Multi-Level and Multi-Prompt (MLMP) entropy minimization integrates features from intermediate vision-encoder layers and is performed with different text-prompt templates at both the global CLS token and local pixel-wise levels. Our approach could be used as plug-and-play for any segmentation network, does not require additional training data or labels, and remains effective even with a single test sample. Furthermore, we introduce a comprehensive OVSS TTA benchmark suite, which integrates a rigorous evaluation protocol, nine segmentation datasets, 15 common synthetic corruptions, and additional real and rendered domain shifts, \textbf{with a total of 87 distinct test scenarios}, establishing a standardized and comprehensive testbed for future TTA research in open-vocabulary segmentation. Our experiments on this suite demonstrate that our segmentation-tailored method consistently delivers significant gains over direct adoption of TTA classification baselines. Code and data are available at https://github.com/dosowiechi/MLMP.

David Osowiechi, Gustavo A. Vargas Hakim, Mehrdad Noori et al.

Despite their exceptional performance in vision tasks, deep learning models often struggle when faced with domain shifts during testing. Test-Time Training (TTT) methods have recently gained popularity by their ability to enhance the robustness of models through the addition of an auxiliary objective that is jointly optimized with the main task. Being strictly unsupervised, this auxiliary objective is used at test time to adapt the model without any access to labels. In this work, we propose Noise-Contrastive Test-Time Training (NC-TTT), a novel unsupervised TTT technique based on the discrimination of noisy feature maps. By learning to classify noisy views of projected feature maps, and then adapting the model accordingly on new domains, classification performance can be recovered by an important margin. Experiments on several popular test-time adaptation baselines demonstrate the advantages of our method compared to recent approaches for this task. The code can be found at:https://github.com/GustavoVargasHakim/NCTTT.git

Farzad Beizaee, Gregory Lodygensky, Christian Desrosiers et al.

Unsupervised anomaly detection in brain images is crucial for identifying injuries and pathologies without access to labels. However, the accurate localization of anomalies in medical images remains challenging due to the inherent complexity and variability of brain structures and the scarcity of annotated abnormal data. To address this challenge, we propose a novel approach that incorporates masking within diffusion models, leveraging their generative capabilities to learn robust representations of normal brain anatomy. During training, our model processes only normal brain MRI scans and performs a forward diffusion process in the latent space that adds noise to the features of randomly-selected patches. Following a dual objective, the model learns to identify which patches are noisy and recover their original features. This strategy ensures that the model captures intricate patterns of normal brain structures while isolating potential anomalies as noise in the latent space. At inference, the model identifies noisy patches corresponding to anomalies and generates a normal counterpart for these patches by applying a reverse diffusion process. Our method surpasses existing unsupervised anomaly detection techniques, demonstrating superior performance in generating accurate normal counterparts and localizing anomalies. The code is available at hhttps://github.com/farzad-bz/MAD-AD.

Moslem Yazdanpanah, Ali Bahri, Mehrdad Noori et al.

Test-time adaptation (TTA) is crucial for mitigating performance degradation caused by distribution shifts in 3D point cloud classification. In this work, we introduce Token Purging (PG), a novel backpropagation-free approach that removes tokens highly affected by domain shifts before they reach attention layers. Unlike existing TTA methods, PG operates at the token level, ensuring robust adaptation without iterative updates. We propose two variants: PG-SP, which leverages source statistics, and PG-SF, a fully source-free version relying on CLS-token-driven adaptation. Extensive evaluations on ModelNet40-C, ShapeNet-C, and ScanObjectNN-C demonstrate that PG-SP achieves an average of +10.3\% higher accuracy than state-of-the-art backpropagation-free methods, while PG-SF sets new benchmarks for source-free adaptation. Moreover, PG is 12.4 times faster and 5.5 times more memory efficient than our baseline, making it suitable for real-world deployment. Code is available at \hyperlink{https://github.com/MosyMosy/Purge-Gate}{https://github.com/MosyMosy/Purge-Gate}

Sina Hajimiri, Farzad Beizaee, Fereshteh Shakeri et al.

Vision transformers have demonstrated remarkable success in classification by leveraging global self-attention to capture long-range dependencies. However, this same mechanism can obscure fine-grained spatial details crucial for tasks such as segmentation. In this work, we seek to enhance segmentation performance of vision transformers after standard image-level classification training. More specifically, we present a simple yet effective add-on that improves performance on segmentation tasks while retaining vision transformers' image-level recognition capabilities. In our approach, we modulate the self-attention with a learnable Gaussian kernel that biases the attention toward neighboring patches. We further refine the patch representations to learn better embeddings at patch positions. These modifications encourage tokens to focus on local surroundings and ensure meaningful representations at spatial positions, while still preserving the model's ability to incorporate global information. Experiments demonstrate the effectiveness of our modifications, evidenced by substantial segmentation gains on three benchmarks (e.g., over 6% and 4% on ADE20K for ViT Tiny and Base), without changing the training regime or sacrificing classification performance. The code is available at https://github.com/sinahmr/LocAtViT/.

Ali Bahri, Moslem Yazdanpanah, Sahar Dastani et al.

Test-Time Training (TTT) has emerged as a promising solution to address distribution shifts in 3D point cloud classification. However, existing methods often rely on computationally expensive backpropagation during adaptation, limiting their applicability in real-world, time-sensitive scenarios. In this paper, we introduce SMART-PC, a skeleton-based framework that enhances resilience to corruptions by leveraging the geometric structure of 3D point clouds. During pre-training, our method predicts skeletal representations, enabling the model to extract robust and meaningful geometric features that are less sensitive to corruptions, thereby improving adaptability to test-time distribution shifts. Unlike prior approaches, SMART-PC achieves real-time adaptation by eliminating backpropagation and updating only BatchNorm statistics, resulting in a lightweight and efficient framework capable of achieving high frame-per-second rates while maintaining superior classification performance. Extensive experiments on benchmark datasets, including ModelNet40-C, ShapeNet-C, and ScanObjectNN-C, demonstrate that SMART-PC achieves state-of-the-art results, outperforming existing methods such as MATE in terms of both accuracy and computational efficiency. The implementation is available at: https://github.com/AliBahri94/SMART-PC.

David Osowiechi, Mehrdad Noori, Gustavo Adolfo Vargas Hakim et al.

Vision-Language Models (VLMs) such as CLIP have yielded unprecedented performance for zero-shot image classification, yet their generalization capability may still be seriously challenged when confronted to domain shifts. In response, we present Weight Average Test-Time Adaptation (WATT) of CLIP, a pioneering approach facilitating full test-time adaptation (TTA) of this VLM. Our method employs a diverse set of templates for text prompts, augmenting the existing framework of CLIP. Predictions are utilized as pseudo labels for model updates, followed by weight averaging to consolidate the learned information globally. Furthermore, we introduce a text ensemble strategy, enhancing overall test performance by aggregating diverse textual cues. Our findings underscore the efficacy of WATT in enhancing performance across diverse datasets, including CIFAR-10-C, CIFAR-10.1, CIFAR-100-C, VisDA-C, and several other challenging datasets, effectively covering a wide range of domain shifts. Notably, these enhancements are achieved without necessitating additional model transformations or trainable modules. Moreover, compared to other Test-Time Adaptation methods, our approach can operate effectively with just a single image. Highlighting the potential of innovative test-time strategies, this research emphasizes their role in fortifying the adaptability of VLMs. The implementation is available at: \url{https://github.com/Mehrdad-Noori/WATT.git}.

Antoine Théberge, Christian Desrosiers, Maxime Descoteaux et al.

Recently, deep reinforcement learning (RL) has been proposed to learn the tractography procedure and train agents to reconstruct the structure of the white matter without manually curated reference streamlines. While the performances reported were competitive, the proposed framework is complex, and little is still known about the role and impact of its multiple parts. In this work, we thoroughly explore the different components of the proposed framework, such as the choice of the RL algorithm, seeding strategy, the input signal and reward function, and shed light on their impact. Approximately 7,400 models were trained for this work, totalling nearly 41,000 hours of GPU time. Our goal is to guide researchers eager to explore the possibilities of deep RL for tractography by exposing what works and what does not work with the category of approach. As such, we ultimately propose a series of recommendations concerning the choice of RL algorithm, the input to the agents, the reward function and more to help future work using reinforcement learning for tractography. We also release the open source codebase, trained models, and datasets for users and researchers wanting to explore reinforcement learning for tractography.

Bingyuan Liu, Christian Desrosiers, Ismail Ben Ayed et al.

Despite achieving promising results in a breadth of medical image segmentation tasks, deep neural networks require large training datasets with pixel-wise annotations. Obtaining these curated datasets is a cumbersome process which limits the applicability in scenarios. Mixed supervision is an appealing alternative for mitigating this obstacle. In this work, we propose a dual-branch architecture, where the upper branch (teacher) receives strong annotations, while the bottom one (student) is driven by limited supervision and guided by the upper branch. Combined with a standard cross-entropy loss over the labeled pixels, our novel formulation integrates two important terms: (i) a Shannon entropy loss defined over the less-supervised images, which encourages confident student predictions in the bottom branch; and (ii) a KL divergence term, which transfers the knowledge (i.e., predictions) of the strongly supervised branch to the less-supervised branch and guides the entropy (student-confidence) term to avoid trivial solutions. We show that the synergy between the entropy and KL divergence yields substantial improvements in performance. We also discuss an interesting link between Shannon-entropy minimization and standard pseudo-mask generation, and argue that the former should be preferred over the latter for leveraging information from unlabeled pixels. We evaluate the effectiveness of the proposed formulation through a series of quantitative and qualitative experiments using two publicly available datasets. Results demonstrate that our method significantly outperforms other strategies for semantic segmentation within a mixed-supervision framework, as well as recent semi-supervised approaches. Our code is publicly available: https://github.com/by-liu/ConfKD.

Hoel Kervadec, Jihene Bouchtiba, Christian Desrosiers et al.

Widely used loss functions for CNN segmentation, e.g., Dice or cross-entropy, are based on integrals over the segmentation regions. Unfortunately, for highly unbalanced segmentations, such regional summations have values that differ by several orders of magnitude across classes, which affects training performance and stability. We propose a boundary loss, which takes the form of a distance metric on the space of contours, not regions. This can mitigate the difficulties of highly unbalanced problems because it uses integrals over the interface between regions instead of unbalanced integrals over the regions. Furthermore, a boundary loss complements regional information. Inspired by graph-based optimization techniques for computing active-contour flows, we express a non-symmetric $L_2$ distance on the space of contours as a regional integral, which avoids completely local differential computations involving contour points. This yields a boundary loss expressed with the regional softmax probability outputs of the network, which can be easily combined with standard regional losses and implemented with any existing deep network architecture for N-D segmentation. We report comprehensive evaluations and comparisons on different unbalanced problems, showing that our boundary loss can yield significant increases in performances while improving training stability. Our code is publicly available: https://github.com/LIVIAETS/surface-loss .

Arnab Kumar Mondal, Jose Dolz, Christian Desrosiers

We address the problem of segmenting 3D multi-modal medical images in scenarios where very few labeled examples are available for training. Leveraging the recent success of adversarial learning for semi-supervised segmentation, we propose a novel method based on Generative Adversarial Networks (GANs) to train a segmentation model with both labeled and unlabeled images. The proposed method prevents over-fitting by learning to discriminate between true and fake patches obtained by a generator network. Our work extends current adversarial learning approaches, which focus on 2D single-modality images, to the more challenging context of 3D volumes of multiple modalities. The proposed method is evaluated on the problem of segmenting brain MRI from the iSEG-2017 and MRBrainS 2013 datasets. Significant performance improvement is reported, compared to state-of-art segmentation networks trained in a fully-supervised manner. In addition, our work presents a comprehensive analysis of different GAN architectures for semi-supervised segmentation, showing recent techniques like feature matching to yield a higher performance than conventional adversarial training approaches. Our code is publicly available at https://github.com/arnab39/FewShot_GAN-Unet3D

Jose Dolz, Karthik Gopinath, Jing Yuan et al.

Recently, dense connections have attracted substantial attention in computer vision because they facilitate gradient flow and implicit deep supervision during training. Particularly, DenseNet, which connects each layer to every other layer in a feed-forward fashion, has shown impressive performances in natural image classification tasks. We propose HyperDenseNet, a 3D fully convolutional neural network that extends the definition of dense connectivity to multi-modal segmentation problems. Each imaging modality has a path, and dense connections occur not only between the pairs of layers within the same path, but also between those across different paths. This contrasts with the existing multi-modal CNN approaches, in which modeling several modalities relies entirely on a single joint layer (or level of abstraction) for fusion, typically either at the input or at the output of the network. Therefore, the proposed network has total freedom to learn more complex combinations between the modalities, within and in-between all the levels of abstraction, which increases significantly the learning representation. We report extensive evaluations over two different and highly competitive multi-modal brain tissue segmentation challenges, iSEG 2017 and MRBrainS 2013, with the former focusing on 6-month infant data and the latter on adult images. HyperDenseNet yielded significant improvements over many state-of-the-art segmentation networks, ranking at the top on both benchmarks. We further provide a comprehensive experimental analysis of features re-use, which confirms the importance of hyper-dense connections in multi-modal representation learning. Our code is publicly available at https://www.github.com/josedolz/HyperDenseNet.

Ali Bahri, Moslem Yazdanpanah, Mehrdad Noori et al.

Test-Time Adaptation (TTA) addresses distribution shifts during testing by adapting a pretrained model without access to source data. In this work, we propose a novel TTA approach for 3D point cloud classification, combining sampling variation with weight averaging. Our method leverages Farthest Point Sampling (FPS) and K-Nearest Neighbors (KNN) to create multiple point cloud representations, adapting the model for each variation using the TENT algorithm. The final model parameters are obtained by averaging the adapted weights, leading to improved robustness against distribution shifts. Extensive experiments on ModelNet40-C, ShapeNet-C, and ScanObjectNN-C datasets, with different backbones (Point-MAE, PointNet, DGCNN), demonstrate that our approach consistently outperforms existing methods while maintaining minimal resource overhead. The proposed method effectively enhances model generalization and stability in challenging real-world conditions.

Marco Colussi, Sergio Mascetti, Jose Dolz et al.

The remarkable progress in deep learning (DL) showcases outstanding results in various computer vision tasks. However, adaptation to real-time variations in data distributions remains an important challenge. Test-Time Training (TTT) was proposed as an effective solution to this issue, which increases the generalization ability of trained models by adding an auxiliary task at train time and then using its loss at test time to adapt the model. Inspired by the recent achievements of contrastive representation learning in unsupervised tasks, we propose ReC-TTT, a test-time training technique that can adapt a DL model to new unseen domains by generating discriminative views of the input data. ReC-TTT uses cross-reconstruction as an auxiliary task between a frozen encoder and two trainable encoders, taking advantage of a single shared decoder. This enables, at test time, to adapt the encoders to extract features that will be correctly reconstructed by the decoder that, in this phase, is frozen on the source domain. Experimental results show that ReC-TTT achieves better results than other state-of-the-art techniques in most domain shift classification challenges.

Ali Bahri, Moslem Yazdanpanah, Mehrdad Noori et al.

We introduce a pioneering approach to self-supervised learning for point clouds, employing a geometrically informed mask selection strategy called GeoMask3D (GM3D) to boost the efficiency of Masked Auto Encoders (MAE). Unlike the conventional method of random masking, our technique utilizes a teacher-student model to focus on intricate areas within the data, guiding the model's focus toward regions with higher geometric complexity. This strategy is grounded in the hypothesis that concentrating on harder patches yields a more robust feature representation, as evidenced by the improved performance on downstream tasks. Our method also presents a complete-to-partial feature-level knowledge distillation technique designed to guide the prediction of geometric complexity utilizing a comprehensive context from feature-level information. Extensive experiments confirm our method's superiority over State-Of-The-Art (SOTA) baselines, demonstrating marked improvements in classification, and few-shot tasks.

Mélanie Gaillochet, Christian Desrosiers, Hervé Lombaert

The reliable deployment of deep learning in medical imaging requires uncertainty quantification that provides rigorous error guarantees while remaining anatomically meaningful. Conformal prediction (CP) is a powerful distribution-free framework for constructing statistically valid prediction intervals. However, standard applications in segmentation often ignore anatomical context, resulting in fragmented, spatially incoherent, and over-segmented prediction sets that limit clinical utility. To bridge this gap, this paper proposes Random-Walk Conformal Prediction (RW-CP), a model-agnostic framework which can be added on top of any segmentation method. RW-CP enforces spatial coherence to generate anatomically valid sets. Our method constructs a k-nearest neighbour graph from pre-trained vision foundation model features and applies a random walk to diffuse uncertainty. The random walk diffusion regularizes the non-conformity scores, making the prediction sets less sensitive to the conformal calibration parameter $λ$, ensuring more stable and continuous anatomical boundaries. RW-CP maintains rigorous marginal coverage while significantly improving segmentation quality. Evaluations on multi-modal public datasets show improvements of up to $35.4\%$ compared to standard CP baselines, given an allowable error rate of $α=0.1$.

Sahar Dastani, Ali Bahri, Gustavo Adolfo Vargas Hakim et al.

State Space Models (SSMs) have emerged as efficient alternatives to Vision Transformers (ViTs), with VMamba standing out as a pioneering architecture designed for vision tasks. However, their generalization performance degrades significantly under distribution shifts. To address this limitation, we propose TRUST (Test-Time Refinement using Uncertainty-Guided SSM Traverses), a novel test-time adaptation (TTA) method that leverages diverse traversal permutations to generate multiple causal perspectives of the input image. Model predictions serve as pseudo-labels to guide updates of the Mamba-specific parameters, and the adapted weights are averaged to integrate the learned information across traversal scans. Altogether, TRUST is the first approach that explicitly leverages the unique architectural properties of SSMs for adaptation. Experiments on seven benchmarks show that TRUST consistently improves robustness and outperforms existing TTA methods.

Ahmad Chaddad, Yihang Wu, Reem Kateb et al.

Domain adaptation (DA) techniques have the potential in machine learning to alleviate distribution differences between training and test sets by leveraging information from source domains. In image classification, most advances in DA have been made using natural images rather than medical data, which are harder to work with. Moreover, even for natural images, the use of mainstream datasets can lead to performance bias. {With the aim of better understanding the benefits of DA for both natural and medical images, this study performs 557 simulation studies using seven widely-used DA techniques for image classification in five natural and eight medical datasets that cover various scenarios, such as out-of-distribution, dynamic data streams, and limited training samples.} Our experiments yield detailed results and insightful observations highlighting the performance and medical applicability of these techniques. Notably, our results have shown the outstanding performance of the Deep Subdomain Adaptation Network (DSAN) algorithm. This algorithm achieved feasible classification accuracy (91.2\%) in the COVID-19 dataset using Resnet50 and showed an important accuracy improvement in the dynamic data stream DA scenario (+6.7\%) compared to the baseline. Our results also demonstrate that DSAN exhibits remarkable level of explainability when evaluated on COVID-19 and skin cancer datasets. These results contribute to the understanding of DA techniques and offer valuable insight into the effective adaptation of models to medical data.

Sedigheh Dargahi, Sylvain Bouix, Christian Desrosiers

Diffusion MRI (dMRI) is a valuable tool to map brain microstructure and connectivity by analyzing water molecule diffusion in tissue. However, acquiring dMRI data requires to capture multiple 3D brain volumes in a short time, often leading to trade-offs in image quality. One challenging artifact is susceptibility-induced distortion, which introduces significant geometric and intensity deformations. Traditional correction methods, such as topup, rely on having access to blip-up and blip-down image pairs, limiting their applicability to retrospective data acquired with a single phase encoding direction. In this work, we propose a deep learning-based approach to correct susceptibility distortions using only a single acquisition (either blip-up or blip-down), eliminating the need for paired acquisitions. Experimental results show that our method achieves performance comparable to topup, demonstrating its potential as an efficient and practical alternative for susceptibility distortion correction in dMRI.

Samuel Barbeau, Pedram Fekri, David Osowiechi et al.

Deep learning models have demonstrated exceptional performance across a wide range of computer vision tasks. However, their performance often degrades significantly when faced with distribution shifts, such as domain or dataset changes. Test-Time Training (TTT) has emerged as an effective method to enhance model robustness by incorporating an auxiliary unsupervised task during training and leveraging it for model updates at test time. In this work, we introduce CTA (Cross-Task Alignment), a novel approach for improving TTT. Unlike existing TTT methods, CTA does not require a specialized model architecture and instead takes inspiration from the success of multi-modal contrastive learning to align a supervised encoder with a self-supervised one. This process enforces alignment between the learned representations of both models, thereby mitigating the risk of gradient interference, preserving the intrinsic robustness of self-supervised learning and enabling more semantically meaningful updates at test-time. Experimental results demonstrate substantial improvements in robustness and generalization over the state-of-the-art on several benchmark datasets.

Rajarshi Bhattacharya, Shakeeb Murtaza, Christian Desrosiers et al.

Person re-identification (ReID) models are known to suffer from camera bias, where learned representations cluster according to camera viewpoints rather than identity, leading to significant performance degradation under (inter-camera) domain shifts in real-world surveillance systems when new cameras are added to camera networks. State-of-the-art test-time adaptation (TTA) methods, largely designed for classification tasks, rely on classification entropy-based objectives that fail to generalize well to ReID, thus making them unsuitable for tackling camera bias. In this paper, we introduce DART$^3$, a TTA framework specifically designed to mitigate camera-induced domain shifts in person ReID. DART$^3$ (Distance-Aware Retrieval Tuning at Test Time) leverages a distance-based objective that aligns better with image retrieval tasks like ReID by exploiting the correlation between nearest-neighbor distance and prediction error. Unlike prior ReID-specific domain adaptation methods, DART$^3$ requires no source data, architectural modifications, or retraining, and can be deployed in both fully black-box and hybrid settings. Empirical evaluations on multiple ReID benchmarks indicate that DART$^3$ and DART$^3$ LITE, a lightweight alternative to the approach, consistently outperforms state-of-the-art TTA baselines, making for a viable option to online learning to mitigate the adverse effects of camera bias.

Bach Kim, Jose Dolz, Pierre-Marc Jodoin et al.

Privacy protection in medical data is a legitimate obstacle for centralized machine learning applications. Here, we propose a client-server image segmentation system which allows for the analysis of multi-centric medical images while preserving patient privacy. In this approach, the client protects the to-be-segmented patient image by mixing it to a reference image. As shown in our work, it is challenging to separate the image mixture to exact original content, thus making the data unworkable and unrecognizable for an unauthorized person. This proxy image is sent to a server for processing. The server then returns the mixture of segmentation maps, which the client can revert to a correct target segmentation. Our system has two components: 1) a segmentation network on the server side which processes the image mixture, and 2) a segmentation unmixing network which recovers the correct segmentation map from the segmentation mixture. Furthermore, the whole system is trained end-to-end. The proposed method is validated on the task of MRI brain segmentation using images from two different datasets. Results show that the segmentation accuracy of our method is comparable to a system trained on raw images, and outperforms other privacy-preserving methods with little computational overhead.

Jizong Peng, Ping Wang, Marco Pedersoli et al.

Unsupervised pre-training has been proven as an effective approach to boost various downstream tasks given limited labeled data. Among various methods, contrastive learning learns a discriminative representation by constructing positive and negative pairs. However, it is not trivial to build reasonable pairs for a segmentation task in an unsupervised way. In this work, we propose a novel unsupervised pre-training framework that avoids the drawback of contrastive learning. Our framework consists of two principles: unsupervised over-segmentation as a pre-train task using mutual information maximization and boundary-aware preserving learning. Experimental results on two benchmark medical segmentation datasets reveal our method's effectiveness in improving segmentation performance when few annotated images are available.

Jizong Peng, Christian Desrosiers, Marco Pedersoli

This work considers semi-supervised segmentation as a dense prediction problem based on prototype vector correlation and proposes a simple way to represent each segmentation class with multiple prototypes. To avoid degenerate solutions, two regularization strategies are applied on unlabeled images. The first one leverages mutual information maximization to ensure that all prototype vectors are considered by the network. The second explicitly enforces prototypes to be orthogonal by minimizing their cosine distance. Experimental results on two benchmark medical segmentation datasets reveal our method's effectiveness in improving segmentation performance when few annotated images are available.

Athanasios N. Nikolakopoulos, Xia Ning, Christian Desrosiers et al.

Collaborative recommendation approaches based on nearest-neighbors are still highly popular today due to their simplicity, their efficiency, and their ability to produce accurate and personalized recommendations. This chapter offers a comprehensive survey of neighborhood-based methods for the item recommendation problem. It presents the main characteristics and benefits of such methods, describes key design choices for implementing a neighborhood-based recommender system, and gives practical information on how to make these choices. A broad range of methods is covered in the chapter, including traditional algorithms like k-nearest neighbors as well as advanced approaches based on matrix factorization, sparse coding and random walks.

Benoit Anctil-Robitaille, Antoine Théberge, Pierre-Marc Jodoin et al.

The physical and clinical constraints surrounding diffusion-weighted imaging (DWI) often limit the spatial resolution of the produced images to voxels up to 8 times larger than those of T1w images. Thus, the detailed information contained in T1w imagescould help in the synthesis of diffusion images in higher resolution. However, the non-Euclidean nature of diffusion imaging hinders current deep generative models from synthesizing physically plausible images. In this work, we propose the first Riemannian network architecture for the direct generation of diffusion tensors (DT) and diffusion orientation distribution functions (dODFs) from high-resolution T1w images. Our integration of the Log-Euclidean Metric into a learning objective guarantees, unlike standard Euclidean networks, the mathematically-valid synthesis of diffusion. Furthermore, our approach improves the fractional anisotropy mean squared error (FA MSE) between the synthesized diffusion and the ground-truth by more than 23% and the cosine similarity between principal directions by almost 5% when compared to our baselines. We validate our generated diffusion by comparing the resulting tractograms to our expected real data. We observe similar fiber bundles with streamlines having less than 3% difference in length, less than 1% difference in volume, and a visually close shape. While our method is able to generate high-resolution diffusion images from structural inputs in less than 15 seconds, we acknowledge and discuss the limits of diffusion inference solely relying on T1w images. Our results nonetheless suggest a relationship between the high-level geometry of the brain and the overall white matter architecture.

Ping Wang, Jizong Peng, Marco Pedersoli et al.

Despite their outstanding accuracy, semi-supervised segmentation methods based on deep neural networks can still yield predictions that are considered anatomically impossible by clinicians, for instance, containing holes or disconnected regions. To solve this problem, we present a Context-aware Virtual Adversarial Training (CaVAT) method for generating anatomically plausible segmentation. Unlike approaches focusing solely on accuracy, our method also considers complex topological constraints like connectivity which cannot be easily modeled in a differentiable loss function. We use adversarial training to generate examples violating the constraints, so the network can learn to avoid making such incorrect predictions on new examples, and employ the Reinforce algorithm to handle non-differentiable segmentation constraints. The proposed method offers a generic and efficient way to add any constraint on top of any segmentation network. Experiments on two clinically-relevant datasets show our method to produce segmentations that are both accurate and anatomically-plausible in terms of region connectivity.

Laurent Chauvin, Kuldeep Kumar, Christian Desrosiers et al.

We propose a novel pairwise distance measure between image keypoint sets, for the purpose of large-scale medical image indexing. Our measure generalizes the Jaccard index to account for soft set equivalence (SSE) between keypoint elements, via an adaptive kernel framework modeling uncertainty in keypoint appearance and geometry. A new kernel is proposed to quantify the variability of keypoint geometry in location and scale. Our distance measure may be estimated between $O(N^2)$ image pairs in $O(N~\log~N)$ operations via keypoint indexing. Experiments report the first results for the task of predicting family relationships from medical images, using 1010 T1-weighted MRI brain volumes of 434 families including monozygotic and dizygotic twins, siblings and half-siblings sharing 100%-25% of their polymorphic genes. Soft set equivalence and the keypoint geometry kernel improve upon standard hard set equivalence (HSE) and appearance kernels alone in predicting family relationships. Monozygotic twin identification is near 100%, and three subjects with uncertain genotyping are automatically paired with their self-reported families, the first reported practical application of image-based family identification. Our distance measure can also be used to predict group categories, sex is predicted with an AUC=0.97. Software is provided for efficient fine-grained curation of large, generic image datasets.

Jizong Peng, Marco Pedersoli, Christian Desrosiers

The scarcity of labeled data often impedes the application of deep learning to the segmentation of medical images. Semi-supervised learning seeks to overcome this limitation by exploiting unlabeled examples in the learning process. In this paper, we present a novel semi-supervised segmentation method that leverages mutual information (MI) on categorical distributions to achieve both global representation invariance and local smoothness. In this method, we maximize the MI for intermediate feature embeddings that are taken from both the encoder and decoder of a segmentation network. We first propose a global MI loss constraining the encoder to learn an image representation that is invariant to geometric transformations. Instead of resorting to computationally-expensive techniques for estimating the MI on continuous feature embeddings, we use projection heads to map them to a discrete cluster assignment where MI can be computed efficiently. Our method also includes a local MI loss to promote spatial consistency in the feature maps of the decoder and provide a smoother segmentation. Since mutual information does not require a strict ordering of clusters in two different assignments, we incorporate a final consistency regularization loss on the output which helps align the cluster labels throughout the network. We evaluate the method on four challenging publicly-available datasets for medical image segmentation. Experimental results show our method to outperform recently-proposed approaches for semi-supervised segmentation and provide an accuracy near to full supervision while training with very few annotated images.

Jose Dolz, Christian Desrosiers, Ismail Ben Ayed

Deep segmentation neural networks require large training datasets with pixel-wise segmentations, which are expensive to obtain in practice. Mixed supervision could mitigate this difficulty, with a small fraction of the data containing complete pixel-wise annotations, while the rest being less supervised, e.g., only a handful of pixels are labeled. In this work, we propose a dual-branch architecture, where the upper branch (teacher) receives strong annotations, while the bottom one (student) is driven by limited supervision and guided by the upper branch. In conjunction with a standard cross-entropy over the labeled pixels, our novel formulation integrates two important terms: (i) a Shannon entropy loss defined over the less-supervised images, which encourages confident student predictions at the bottom branch; and (ii) a Kullback-Leibler (KL) divergence, which transfers the knowledge from the predictions generated by the strongly supervised branch to the less-supervised branch, and guides the entropy (student-confidence) term to avoid trivial solutions. Very interestingly, we show that the synergy between the entropy and KL divergence yields substantial improvements in performances. Furthermore, we discuss an interesting link between Shannon-entropy minimization and standard pseudo-mask generation and argue that the former should be preferred over the latter for leveraging information from unlabeled pixels. Through a series of quantitative and qualitative experiments, we show the effectiveness of the proposed formulation in segmenting the left-ventricle endocardium in MRI images. We demonstrate that our method significantly outperforms other strategies to tackle semantic segmentation within a mixed-supervision framework. More interestingly, and in line with recent observations in classification, we show that the branch trained with reduced supervision largely outperforms the teacher.

Bach Ngoc Kim, Jose Dolz, Christian Desrosiers et al.

We propose a client-server system which allows for the analysis of multi-centric medical images while preserving patient identity. In our approach, the client protects the patient identity by applying a pseudo-random non-linear deformation to the input image. This results into a proxy image which is sent to the server for processing. The server then returns back the deformed processed image which the client reverts to a canonical form. Our system has three components: 1) a flow-field generator which produces a pseudo-random deformation function, 2) a Siamese discriminator that learns the patient identity from the processed image, 3) a medical image processing network that analyzes the content of the proxy images. The system is trained end-to-end in an adversarial manner. By fooling the discriminator, the flow-field generator learns to produce a bi-directional non-linear deformation which allows to remove and recover the identity of the subject from both the input image and output result. After end-to-end training, the flow-field generator is deployed on the client side and the segmentation network is deployed on the server side. The proposed method is validated on the task of MRI brain segmentation using images from two different datasets. Results show that the segmentation accuracy of our method is similar to a system trained on non-encoded images, while considerably reducing the ability to recover subject identity.

Ping Wang, Jizong Peng, Marco Pedersoli et al.

Deep co-training has recently been proposed as an effective approach for image segmentation when annotated data is scarce. In this paper, we improve existing approaches for semi-supervised segmentation with a self-paced and self-consistent co-training method. To help distillate information from unlabeled images, we first design a self-paced learning strategy for co-training that lets jointly-trained neural networks focus on easier-to-segment regions first, and then gradually consider harder ones.This is achieved via an end-to-end differentiable loss inthe form of a generalized Jensen Shannon Divergence(JSD). Moreover, to encourage predictions from different networks to be both consistent and confident, we enhance this generalized JSD loss with an uncertainty regularizer based on entropy. The robustness of individual models is further improved using a self-ensembling loss that enforces their prediction to be consistent across different training iterations. We demonstrate the potential of our method on three challenging image segmentation problems with different image modalities, using small fraction of labeled data. Results show clear advantages in terms of performance compared to the standard co-training baselines and recently proposed state-of-the-art approaches for semi-supervised segmentation