Prostate cancer is one of the main diseases affecting men worldwide. The gold standard for diagnosis and prognosis is the Gleason grading system. In this process, pathologists manually analyze prostate histology slides under microscope, in a high time-consuming and subjective task. In the last years, computer-aided-diagnosis (CAD) systems have emerged as a promising tool that could support pathologists in the daily clinical practice. Nevertheless, these systems are usually trained using tedious and prone-to-error pixel-level annotations of Gleason grades in the tissue. To alleviate the need of manual pixel-wise labeling, just a handful of works have been presented in the literature. Motivated by this, we propose a novel weakly-supervised deep-learning model, based on self-learning CNNs, that leverages only the global Gleason score of gigapixel whole slide images during training to accurately perform both, grading of patch-level patterns and biopsy-level scoring. To evaluate the performance of the proposed method, we perform extensive experiments on three different external datasets for the patch-level Gleason grading, and on two different test sets for global Grade Group prediction. We empirically demonstrate that our approach outperforms its supervised counterpart on patch-level Gleason grading by a large margin, as well as state-of-the-art methods on global biopsy-level scoring. Particularly, the proposed model brings an average improvement on the Cohen's quadratic kappa (k) score of nearly 18% compared to full-supervision for the patch-level Gleason grading task.
Standard losses for training deep segmentation networks could be seen as individual classifications of pixels, instead of supervising the global shape of the predicted segmentations. While effective, they require exact knowledge of the label of each pixel in an image. This study investigates how effective global geometric shape descriptors could be, when used on their own as segmentation losses for training deep networks. Not only interesting theoretically, there exist deeper motivations to posing segmentation problems as a reconstruction of shape descriptors: Annotations to obtain approximations of low-order shape moments could be much less cumbersome than their full-mask counterparts, and anatomical priors could be readily encoded into invariant shape descriptions, which might alleviate the annotation burden. Also, and most importantly, we hypothesize that, given a task, certain shape descriptions might be invariant across image acquisition protocols/modalities and subject populations, which might open interesting research avenues for generalization in medical image segmentation. We introduce and formulate a few shape descriptors in the context of deep segmentation, and evaluate their potential as standalone losses on two different challenging tasks. Inspired by recent works in constrained optimization for deep networks, we propose a way to use those descriptors to supervise segmentation, without any pixel-level label. Very surprisingly, as little as 4 descriptors values per class can approach the performance of a segmentation mask with 65k individual discrete labels. We also found that shape descriptors can be a valid way to encode anatomical priors about the task, enabling to leverage expert knowledge without additional annotations. Our implementation is publicly available and can be easily extended to other tasks and descriptors: https://github.com/hkervadec/shape_descriptors
Techniques for multi-target domain adaptation (MTDA) seek to adapt a recognition model such that it can generalize well across multiple target domains. While several successful techniques have been proposed for unsupervised single-target domain adaptation (STDA) in object detection, adapting a model to multiple target domains using unlabeled image data remains a challenging and largely unexplored problem. Key challenges include the lack of bounding box annotations for target data, knowledge corruption, and the growing resource requirements needed to train accurate deep detection models. The later requirements are augmented by the need to retraining a model with previous-learned target data when adapting to each new target domain. Currently, the only MTDA technique in literature for object detection relies on distillation with a duplicated model to avoid knowledge corruption but does not leverage the source-target feature alignment after UDA. To address these challenges, we propose a new Incremental MTDA technique for object detection that can adapt a detector to multiple target domains, one at a time, without having to retain data of previously-learned target domains. Instead of distillation, our technique efficiently transfers source images to a joint target domains' space, on the fly, thereby preserving knowledge during incremental MTDA. Using adversarial training, our Domain Transfer Module (DTM) is optimized to trick the domain classifiers into classifying source images as though transferred into the target domain, thus allowing the DTM to generate samples close to a joint distribution of target domains. Our proposed technique is validated on different MTDA detection benchmarks, and results show it improving accuracy across multiple domains, despite the considerable reduction in complexity.
Most segmentation losses are arguably variants of the Cross-Entropy (CE) or Dice loss. In the literature, there is no clear consensus as to which of these losses is a better choice, with varying performances for each across different benchmarks and applications. We develop a theoretical analysis that links these two types of losses, exposing their advantages and weaknesses. First, we explicitly demonstrate that CE and Dice share a much deeper connection than previously thought: CE is an upper bound on both logarithmic and linear Dice losses. Furthermore, we provide an information-theoretic analysis, which highlights hidden label-marginal biases : Dice has an intrinsic bias towards imbalanced solutions, whereas CE implicitly encourages the ground-truth region proportions. Our theoretical results explain the wide experimental evidence in the medical-imaging literature, whereby Dice losses bring improvements for imbalanced segmentation. It also explains why CE dominates natural-image problems with diverse class proportions, in which case Dice might have difficulty adapting to different label-marginal distributions. Based on our theoretical analysis, we propose a principled and simple solution, which enables to control explicitly the label-marginal bias. Our loss integrates CE with explicit ${\cal L}_1$ regularization, which encourages label marginals to match target class proportions, thereby mitigating class imbalance but without losing generality. Comprehensive experiments and ablation studies over different losses and applications validate our theoretical analysis, as well as the effectiveness of our explicit label-marginal regularizers.
Weakly supervised learning has emerged as an appealing alternative to alleviate the need for large labeled datasets in semantic segmentation. Most current approaches exploit class activation maps (CAMs), which can be generated from image-level annotations. Nevertheless, resulting maps have been demonstrated to be highly discriminant, failing to serve as optimal proxy pixel-level labels. We present a novel learning strategy that leverages self-supervision in a multi-modal image scenario to significantly enhance original CAMs. In particular, the proposed method is based on two observations. First, the learning of fully-supervised segmentation networks implicitly imposes equivariance by means of data augmentation, whereas this implicit constraint disappears on CAMs generated with image tags. And second, the commonalities between image modalities can be employed as an efficient self-supervisory signal, correcting the inconsistency shown by CAMs obtained across multiple modalities. To effectively train our model, we integrate a novel loss function that includes a within-modality and a cross-modality equivariant term to explicitly impose these constraints during training. In addition, we add a KL-divergence on the class prediction distributions to facilitate the information exchange between modalities, which, combined with the equivariant regularizers further improves the performance of our model. Exhaustive experiments on the popular multi-modal BRATS dataset demonstrate that our approach outperforms relevant recent literature under the same learning conditions.
Beyond the complexity of CNNs that require training on large annotated datasets, the domain shift between design and operational data has limited the adoption of CNNs in many real-world applications. For instance, in person re-identification, videos are captured over a distributed set of cameras with non-overlapping viewpoints. The shift between the source (e.g. lab setting) and target (e.g. cameras) domains may lead to a significant decline in recognition accuracy. Additionally, state-of-the-art CNNs may not be suitable for such real-time applications given their computational requirements. Although several techniques have recently been proposed to address domain shift problems through unsupervised domain adaptation (UDA), or to accelerate/compress CNNs through knowledge distillation (KD), we seek to simultaneously adapt and compress CNNs to generalize well across multiple target domains. In this paper, we propose a progressive KD approach for unsupervised single-target DA (STDA) and multi-target DA (MTDA) of CNNs. Our method for KD-STDA adapts a CNN to a single target domain by distilling from a larger teacher CNN, trained on both target and source domain data in order to maintain its consistency with a common representation. Our proposed approach is compared against state-of-the-art methods for compression and STDA of CNNs on the Office31 and ImageClef-DA image classification datasets. It is also compared against state-of-the-art methods for MTDA on Digits, Office31, and OfficeHome. In both settings -- KD-STDA and KD-MTDA -- results indicate that our approach can achieve the highest level of accuracy across target domains, while requiring a comparable or lower CNN complexity.
Precise determination and assessment of bladder cancer (BC) extent of muscle invasion involvement guides proper risk stratification and personalized therapy selection. In this context, segmentation of both bladder walls and cancer are of pivotal importance, as it provides invaluable information to stage the primary tumour. Hence, multi region segmentation on patients presenting with symptoms of bladder tumours using deep learning heralds a new level of staging accuracy and prediction of the biologic behaviour of the tumour. Nevertheless, despite the success of these models in other medical problems, progress in multi region bladder segmentation is still at a nascent stage, with just a handful of works tackling a multi region scenario. Furthermore, most existing approaches systematically follow prior literature in other clinical problems, without casting a doubt on the validity of these methods on bladder segmentation, which may present different challenges. Inspired by this, we provide an in-depth look at bladder cancer segmentation using deep learning models. The critical determinants for accurate differentiation of muscle invasive disease, current status of deep learning based bladder segmentation, lessons and limitations of prior work are highlighted.
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.
Few-shot segmentation has recently attracted substantial interest, with the popular meta-learning paradigm widely dominating the literature. We show that the way inference is performed for a given few-shot segmentation task has a substantial effect on performances, an aspect that has been overlooked in the literature. We introduce a transductive inference, which leverages the statistics of the unlabeled pixels of a task by optimizing a new loss containing three complementary terms: (i) a standard cross-entropy on the labeled pixels; (ii) the entropy of posteriors on the unlabeled query pixels; and (iii) a global KL-divergence regularizer based on the proportion of the predicted foreground region. Our inference uses a simple linear classifier of the extracted features, has a computational load comparable to inductive inference and can be used on top of any base training. Using standard cross-entropy training on the base classes, our inference yields highly competitive performances on well-known few-shot segmentation benchmarks. On PASCAL-5i, it brings about 5% improvement over the best performing state-of-the-art method in the 5-shot scenario, while being on par in the 1-shot setting. Even more surprisingly, this gap widens as the number of support samples increases, reaching up to 6% in the 10-shot scenario. Furthermore, we introduce a more realistic setting with domain shift, where the base and novel classes are drawn from different datasets. In this setting, we found that our method achieves the best performances.