Confidence-Aware Paced-Curriculum Learning by Label Smoothing for Surgical Scene Understanding

Curriculum learning and self-paced learning are the training strategies that gradually feed the samples from easy to more complex. They have captivated increasing attention due to their excellent performance in robotic vision. Most recent works focus on designing curricula based on difficulty levels in input samples or smoothing the feature maps. However, smoothing labels to control the learning utility in a curriculum manner is still unexplored. In this work, we design a paced curriculum by label smoothing (P-CBLS) using paced learning with uniform label smoothing (ULS) for classification tasks and fuse uniform and spatially varying label smoothing (SVLS) for semantic segmentation tasks in a curriculum manner. In ULS and SVLS, a bigger smoothing factor value enforces a heavy smoothing penalty in the true label and limits learning less information. Therefore, we design the curriculum by label smoothing (CBLS). We set a bigger smoothing value at the beginning of training and gradually decreased it to zero to control the model learning utility from lower to higher. We also designed a confidence-aware pacing function and combined it with our CBLS to investigate the benefits of various curricula. The proposed techniques are validated on four robotic surgery datasets of multi-class, multi-label classification, captioning, and segmentation tasks. We also investigate the robustness of our method by corrupting validation data into different severity levels. Our extensive analysis shows that the proposed method improves prediction accuracy and robustness.

Context Label Learning: Improving Background Class Representations in Semantic Segmentation

Background samples provide key contextual information for segmenting regions of interest (ROIs). However, they always cover a diverse set of structures, causing difficulties for the segmentation model to learn good decision boundaries with high sensitivity and precision. The issue concerns the highly heterogeneous nature of the background class, resulting in multi-modal distributions. Empirically, we find that neural networks trained with heterogeneous background struggle to map the corresponding contextual samples to compact clusters in feature space. As a result, the distribution over background logit activations may shift across the decision boundary, leading to systematic over-segmentation across different datasets and tasks. In this study, we propose context label learning (CoLab) to improve the context representations by decomposing the background class into several subclasses. Specifically, we train an auxiliary network as a task generator, along with the primary segmentation model, to automatically generate context labels that positively affect the ROI segmentation accuracy. Extensive experiments are conducted on several challenging segmentation tasks and datasets. The results demonstrate that CoLab can guide the segmentation model to map the logits of background samples away from the decision boundary, resulting in significantly improved segmentation accuracy. Code is available.

Analysing the effectiveness of a generative model for semi-supervised medical image segmentation

Image segmentation is important in medical imaging, providing valuable, quantitative information for clinical decision-making in diagnosis, therapy, and intervention. The state-of-the-art in automated segmentation remains supervised learning, employing discriminative models such as U-Net. However, training these models requires access to large amounts of manually labelled data which is often difficult to obtain in real medical applications. In such settings, semi-supervised learning (SSL) attempts to leverage the abundance of unlabelled data to obtain more robust and reliable models. Recently, generative models have been proposed for semantic segmentation, as they make an attractive choice for SSL. Their ability to capture the joint distribution over input images and output label maps provides a natural way to incorporate information from unlabelled images. This paper analyses whether deep generative models such as the SemanticGAN are truly viable alternatives to tackle challenging medical image segmentation problems. To that end, we thoroughly evaluate the segmentation performance, robustness, and potential subgroup disparities of discriminative and generative segmentation methods when applied to large-scale, publicly available chest X-ray datasets.

A Comparative Study of Graph Neural Networks for Shape Classification in Neuroimaging

Graph neural networks have emerged as a promising approach for the analysis of non-Euclidean data such as meshes. In medical imaging, mesh-like data plays an important role for modelling anatomical structures, and shape classification can be used in computer aided diagnosis and disease detection. However, with a plethora of options, the best architectural choices for medical shape analysis using GNNs remain unclear. We conduct a comparative analysis to provide practitioners with an overview of the current state-of-the-art in geometric deep learning for shape classification in neuroimaging. Using biological sex classification as a proof-of-concept task, we find that using FPFH as node features substantially improves GNN performance and generalisation to out-of-distribution data; we compare the performance of three alternative convolutional layers; and we reinforce the importance of data augmentation for graph based learning. We then confirm these results hold for a clinically relevant task, using the classification of Alzheimer's disease.

Visual Debates

The natural way of obtaining different perspectives on any given topic is by conducting a debate, where participants argue for and against the topic. Here, we propose a novel debate framework for understanding the classifier's reasoning for making a particular prediction by modelling it as a multiplayer sequential zero-sum game. The players aim to maximise their utilities by adjusting their arguments with respect to other players' counterarguments. The contrastive nature of our framework encourages players to put forward diverse arguments, picking up the reasoning trails missed by their opponents. Thus, our framework answers the question: why did the classifier make a certain prediction?, by allowing players to argue for and against the classifier's decision. In the proposed setup, given the question and the classifier's latent knowledge, both agents take turns in proposing arguments to support or contradict the classifier's decision; arguments here correspond to the selection of specific features from the discretised latent space of the continuous classifier. By the end of the debate, we collect sets of supportive and manipulative features, serving as an explanation depicting the internal reasoning of the classifier. We demonstrate our Visual Debates on the geometric SHAPE and MNIST datasets for subjective validation, followed by the high-resolution AFHQ dataset. For further investigation, our framework is available at \url{https://github.com/koriavinash1/VisualDebates}.

Frequency Dropout: Feature-Level Regularization via Randomized Filtering

Deep convolutional neural networks have shown remarkable performance on various computer vision tasks, and yet, they are susceptible to picking up spurious correlations from the training signal. So called `shortcuts' can occur during learning, for example, when there are specific frequencies present in the image data that correlate with the output predictions. Both high and low frequencies can be characteristic of the underlying noise distribution caused by the image acquisition rather than in relation to the task-relevant information about the image content. Models that learn features related to this characteristic noise will not generalize well to new data. In this work, we propose a simple yet effective training strategy, Frequency Dropout, to prevent convolutional neural networks from learning frequency-specific imaging features. We employ randomized filtering of feature maps during training which acts as a feature-level regularization. In this study, we consider common image processing filters such as Gaussian smoothing, Laplacian of Gaussian, and Gabor filtering. Our training strategy is model-agnostic and can be used for any computer vision task. We demonstrate the effectiveness of Frequency Dropout on a range of popular architectures and multiple tasks including image classification, domain adaptation, and semantic segmentation using both computer vision and medical imaging datasets. Our results suggest that the proposed approach does not only improve predictive accuracy but also improves robustness against domain shift.

Risk of Bias in Chest X-ray Foundation Models

Foundation models are considered a breakthrough in all applications of AI, promising robust and reusable mechanisms for feature extraction, alleviating the need for large amounts of high quality training data for task-specific prediction models. However, foundation models may potentially encode and even reinforce existing biases present in historic datasets. Given the limited ability to scrutinize foundation models, it remains unclear whether the opportunities outweigh the risks in safety critical applications such as clinical decision making. In our statistical bias analysis of a recently published, and publicly available chest X-ray foundation model, we found reasons for concern as the model seems to encode protected characteristics including biological sex and racial identity, which may lead to disparate performance across subgroups in downstream applications. While research into foundation models for healthcare applications is in an early stage, we believe it is important to make the community aware of these risks to avoid harm.

Deep Structural Causal Shape Models

Causal reasoning provides a language to ask important interventional and counterfactual questions beyond purely statistical association. In medical imaging, for example, we may want to study the causal effect of genetic, environmental, or lifestyle factors on the normal and pathological variation of anatomical phenotypes. However, while anatomical shape models of 3D surface meshes, extracted from automated image segmentation, can be reliably constructed, there is a lack of computational tooling to enable causal reasoning about morphological variations. To tackle this problem, we propose deep structural causal shape models (CSMs), which utilise high-quality mesh generation techniques, from geometric deep learning, within the expressive framework of deep structural causal models. CSMs enable subject-specific prognoses through counterfactual mesh generation ("How would this patient's brain structure change if they were ten years older?"), which is in contrast to most current works on purely population-level statistical shape modelling. We demonstrate the capabilities of CSMs at all levels of Pearl's causal hierarchy through a number of qualitative and quantitative experiments leveraging a large dataset of 3D brain structures.

Evaluation of 3D GANs for Lung Tissue Modelling in Pulmonary CT

GANs are able to model accurately the distribution of complex, high-dimensional datasets, e.g. images. This makes high-quality GANs useful for unsupervised anomaly detection in medical imaging. However, differences in training datasets such as output image dimensionality and appearance of semantically meaningful features mean that GAN models from the natural image domain may not work `out-of-the-box' for medical imaging, necessitating re-implementation and re-evaluation. In this work we adapt and evaluate three GAN models to the task of modelling 3D healthy image patches for pulmonary CT. To the best of our knowledge, this is the first time that such an evaluation has been performed. The DCGAN, styleGAN and the bigGAN architectures were investigated due to their ubiquity and high performance in natural image processing. We train different variants of these methods and assess their performance using the FID score. In addition, the quality of the generated images was evaluated by a human observer study, the ability of the networks to model 3D domain-specific features was investigated, and the structure of the GAN latent spaces was analysed. Results show that the 3D styleGAN produces realistic-looking images with meaningful 3D structure, but suffer from mode collapse which must be addressed during training to obtain samples diversity. Conversely, the 3D DCGAN models show a greater capacity for image variability, but at the cost of poor-quality images. The 3D bigGAN models provide an intermediate level of image quality, but most accurately model the distribution of selected semantically meaningful features. The results suggest that future development is required to realise a 3D GAN with sufficient capacity for patch-based lung CT anomaly detection and we offer recommendations for future areas of research, such as experimenting with other architectures and incorporation of position-encoding.

Automatic lesion analysis for increased efficiency in outcome prediction of traumatic brain injury

The accurate prognosis for traumatic brain injury (TBI) patients is difficult yet essential to inform therapy, patient management, and long-term after-care. Patient characteristics such as age, motor and pupil responsiveness, hypoxia and hypotension, and radiological findings on computed tomography (CT), have been identified as important variables for TBI outcome prediction. CT is the acute imaging modality of choice in clinical practice because of its acquisition speed and widespread availability. However, this modality is mainly used for qualitative and semi-quantitative assessment, such as the Marshall scoring system, which is prone to subjectivity and human errors. This work explores the predictive power of imaging biomarkers extracted from routinely-acquired hospital admission CT scans using a state-of-the-art, deep learning TBI lesion segmentation method. We use lesion volumes and corresponding lesion statistics as inputs for an extended TBI outcome prediction model. We compare the predictive power of our proposed features to the Marshall score, independently and when paired with classic TBI biomarkers. We find that automatically extracted quantitative CT features perform similarly or better than the Marshall score in predicting unfavourable TBI outcomes. Leveraging automatic atlas alignment, we also identify frontal extra-axial lesions as important indicators of poor outcome. Our work may contribute to a better understanding of TBI, and provides new insights into how automated neuroimaging analysis can be used to improve prognostication after TBI.