Camila González

John Kalkhof, Camila González, Anirban Mukhopadhyay

Access to the proper infrastructure is critical when performing medical image segmentation with Deep Learning. This requirement makes it difficult to run state-of-the-art segmentation models in resource-constrained scenarios like primary care facilities in rural areas and during crises. The recently emerging field of Neural Cellular Automata (NCA) has shown that locally interacting one-cell models can achieve competitive results in tasks such as image generation or segmentations in low-resolution inputs. However, they are constrained by high VRAM requirements and the difficulty of reaching convergence for high-resolution images. To counteract these limitations we propose Med-NCA, an end-to-end NCA training pipeline for high-resolution image segmentation. Our method follows a two-step process. Global knowledge is first communicated between cells across the downscaled image. Following that, patch-based segmentation is performed. Our proposed Med-NCA outperforms the classic UNet by 2% and 3% Dice for hippocampus and prostate segmentation, respectively, while also being 500 times smaller. We also show that Med-NCA is by design invariant with respect to image scale, shape and translation, experiencing only slight performance degradation even with strong shifts; and is robust against MRI acquisition artefacts. Med-NCA enables high-resolution medical image segmentation even on a Raspberry Pi B+, arguably the smallest device able to run PyTorch and that can be powered by a standard power bank.

Amin Ranem, Camila González, Anirban Mukhopadhyay

In clinical settings, where acquisition conditions and patient populations change over time, continual learning is key for ensuring the safe use of deep neural networks. Yet most existing work focuses on convolutional architectures and image classification. Instead, radiologists prefer to work with segmentation models that outline specific regions-of-interest, for which Transformer-based architectures are gaining traction. The self-attention mechanism of Transformers could potentially mitigate catastrophic forgetting, opening the way for more robust medical image segmentation. In this work, we explore how recently-proposed Transformer mechanisms for semantic segmentation behave in sequential learning scenarios, and analyse how best to adapt continual learning strategies for this setting. Our evaluation on hippocampus segmentation shows that Transformer mechanisms mitigate catastrophic forgetting for medical image segmentation compared to purely convolutional architectures, and demonstrates that regularising ViT modules should be done with caution.

Camila Gonzalez, Amin Ranem, Ahmed Othman et al.

Most continual learning methods are validated in settings where task boundaries are clearly defined and task identity information is available during training and testing. We explore how such methods perform in a task-agnostic setting that more closely resembles dynamic clinical environments with gradual population shifts. We propose ODEx, a holistic solution that combines out-of-distribution detection with continual learning techniques. Validation on two scenarios of hippocampus segmentation shows that our proposed method reliably maintains performance on earlier tasks without losing plasticity.

Amelia Jiménez-Sánchez, Natalia-Rozalia Avlona, Sarah de Boer et al.

Datasets play a critical role in medical imaging research, yet issues such as label quality, shortcuts, and metadata are often overlooked. This lack of attention may harm the generalizability of algorithms and, consequently, negatively impact patient outcomes. While existing medical imaging literature reviews mostly focus on machine learning (ML) methods, with only a few focusing on datasets for specific applications, these reviews remain static -- they are published once and not updated thereafter. This fails to account for emerging evidence, such as biases, shortcuts, and additional annotations that other researchers may contribute after the dataset is published. We refer to these newly discovered findings of datasets as research artifacts. To address this gap, we propose a living review that continuously tracks public datasets and their associated research artifacts across multiple medical imaging applications. Our approach includes a framework for the living review to monitor data documentation artifacts, and an SQL database to visualize the citation relationships between research artifact and dataset. Lastly, we discuss key considerations for creating medical imaging datasets, review best practices for data annotation, discuss the significance of shortcuts and demographic diversity, and emphasize the importance of managing datasets throughout their entire lifecycle. Our demo is publicly available at http://inthepicture.itu.dk/.

Camila Gonzalez, Karol Gotkowski, Moritz Fuchs et al.

Automatic segmentation of ground glass opacities and consolidations in chest computer tomography (CT) scans can potentially ease the burden of radiologists during times of high resource utilisation. However, deep learning models are not trusted in the clinical routine due to failing silently on out-of-distribution (OOD) data. We propose a lightweight OOD detection method that leverages the Mahalanobis distance in the feature space and seamlessly integrates into state-of-the-art segmentation pipelines. The simple approach can even augment pre-trained models with clinically relevant uncertainty quantification. We validate our method across four chest CT distribution shifts and two magnetic resonance imaging applications, namely segmentation of the hippocampus and the prostate. Our results show that the proposed method effectively detects far- and near-OOD samples across all explored scenarios.

Hendrik Alexander Mehrtens, Camila González, Anirban Mukhopadhyay

Calibration and uncertainty estimation are crucial topics in high-risk environments. We introduce a new diversity regularizer for classification tasks that uses out-of-distribution samples and increases the overall accuracy, calibration and out-of-distribution detection capabilities of ensembles. Following the recent interest in the diversity of ensembles, we systematically evaluate the viability of explicitly regularizing ensemble diversity to improve calibration on in-distribution data as well as under dataset shift. We demonstrate that diversity regularization is highly beneficial in architectures, where weights are partially shared between the individual members and even allows to use fewer ensemble members to reach the same level of robustness. Experiments on CIFAR-10, CIFAR-100, and SVHN show that regularizing diversity can have a significant impact on calibration and robustness, as well as out-of-distribution detection.

John Kalkhof, Camila González, Anirban Mukhopadhyay

Limited amount of labelled training data are a common problem in medical imaging. This makes it difficult to train a well-generalised model and therefore often leads to failure in unknown domains. Hippocampus segmentation from magnetic resonance imaging (MRI) scans is critical for the diagnosis and treatment of neuropsychatric disorders. Domain differences in contrast or shape can significantly affect segmentation. We address this issue by disentangling a T1-weighted MRI image into its content and domain. This separation enables us to perform a domain transfer and thus convert data from new sources into the training domain. This step thus simplifies the segmentation problem, resulting in higher quality segmentations. We achieve the disentanglement with the proposed novel methodology 'Content Domain Disentanglement GAN', and we propose to retrain the UNet on the transformed outputs to deal with GAN-specific artefacts. With these changes, we are able to improve performance on unseen domains by 6-13% and outperform state-of-the-art domain transfer methods.

Camila Gonzalez, Christian Harder, Amin Ranem et al.

Federated Learning is the most promising way to train robust Deep Learning models for the segmentation of Covid-19-related findings in chest CTs. By learning in a decentralized fashion, heterogeneous data can be leveraged from a variety of sources and acquisition protocols whilst ensuring patient privacy. It is, however, crucial to continuously monitor the performance of the model. Yet when it comes to the segmentation of diffuse lung lesions, a quick visual inspection is not enough to assess the quality, and thorough monitoring of all network outputs by expert radiologists is not feasible. In this work, we present an array of lightweight metrics that can be calculated locally in each hospital and then aggregated for central monitoring of a federated system. Our linear model detects over 70% of low-quality segmentations on an out-of-distribution dataset and thus reliably signals a decline in model performance.

Antoine Sanner, Camila Gonzalez, Anirban Mukhopadhyay

The recent achievements of Deep Learning rely on the test data being similar in distribution to the training data. In an ideal case, Deep Learning models would achieve Out-of-Distribution (OoD) Generalization, i.e. reliably make predictions on out-of-distribution data. Yet in practice, models usually fail to generalize well when facing a shift in distribution. Several methods were thereby designed to improve the robustness of the features learned by a model through Regularization- or Domain-Prediction-based schemes. Segmenting medical images such as MRIs of the hippocampus is essential for the diagnosis and treatment of neuropsychiatric disorders. But these brain images often suffer from distribution shift due to the patient's age and various pathologies affecting the shape of the organ. In this work, we evaluate OoD Generalization solutions for the problem of hippocampus segmentation in MR data using both fully- and semi-supervised training. We find that no method performs reliably in all experiments. Only the V-REx loss stands out as it remains easy to tune, while it outperforms a standard U-Net in most cases.

Camila Gonzalez, Karol Gotkowski, Andreas Bucher et al.

Automatic segmentation of lung lesions in computer tomography has the potential to ease the burden of clinicians during the Covid-19 pandemic. Yet predictive deep learning models are not trusted in the clinical routine due to failing silently in out-of-distribution (OOD) data. We propose a lightweight OOD detection method that exploits the Mahalanobis distance in the feature space. The proposed approach can be seamlessly integrated into state-of-the-art segmentation pipelines without requiring changes in model architecture or training procedure, and can therefore be used to assess the suitability of pre-trained models to new data. We validate our method with a patch-based nnU-Net architecture trained with a multi-institutional dataset and find that it effectively detects samples that the model segments incorrectly.

Camila Gonzalez, Nick Lemke, Georgios Sakas et al.

Continual learning protocols are attracting increasing attention from the medical imaging community. In continual environments, datasets acquired under different conditions arrive sequentially; and each is only available for a limited period of time. Given the inherent privacy risks associated with medical data, this setup reflects the reality of deployment for deep learning diagnostic radiology systems. Many techniques exist to learn continuously for image classification, and several have been adapted to semantic segmentation. Yet most struggle to accumulate knowledge in a meaningful manner. Instead, they focus on preventing the problem of catastrophic forgetting, even when this reduces model plasticity and thereon burdens the training process. This puts into question whether the additional overhead of knowledge preservation is worth it - particularly for medical image segmentation, where computation requirements are already high - or if maintaining separate models would be a better solution. We propose UNEG, a simple and widely applicable multi-model benchmark that maintains separate segmentation and autoencoder networks for each training stage. The autoencoder is built from the same architecture as the segmentation network, which in our case is a full-resolution nnU-Net, to bypass any additional design decisions. During inference, the reconstruction error is used to select the most appropriate segmenter for each test image. Open this concept, we develop a fair evaluation scheme for different continual learning settings that moves beyond the prevention of catastrophic forgetting. Our results across three regions of interest (prostate, hippocampus, and right ventricle) show that UNEG outperforms several continual learning methods, reinforcing the need for strong baselines in continual learning research.