Image and Video Processing (eess.IV)

Abstract: Deep generative models have emerged as influential instruments for data generation and manipulation. Enhancing the controllability of these models by selectively modifying data attributes has been a recent focus. Variational Autoencoders (VAEs) have shown promise in capturing hidden attributes but often produce blurry reconstructions. Controlling these attributes through different imaging domains is difficult in medical imaging. Recently, Soft Introspective VAE leverage the benefits of both VAEs and Generative Adversarial Networks (GANs), which have demonstrated impressive image synthesis capabilities, by incorporating an adversarial loss into VAE training. In this work, we propose the Attributed Soft Introspective VAE (Attri-SIVAE) by incorporating an attribute regularized loss, into the Soft-Intro VAE framework. We evaluate experimentally the proposed method on cardiac MRI data from different domains, such as various scanner vendors and acquisition centers. The proposed method achieves similar performance in terms of reconstruction and regularization compared to the state-of-the-art Attributed regularized VAE but additionally also succeeds in keeping the same regularization level when tested on a different dataset, unlike the compared method.

Abstract: Short axis cardiac MRI segmentation is a well-researched topic, with excellent results achieved by state-of-the-art models in a supervised setting. However, annotating MRI volumes is time-consuming and expensive. Many different approaches (e.g. transfer learning, data augmentation, few-shot learning, etc.) have emerged in an effort to use fewer annotated data and still achieve similar performance as a fully supervised model. Nevertheless, to the best of our knowledge, none of these works focus on which slices of MRI volumes are most important to annotate for yielding the best segmentation results. In this paper, we investigate the effects of training with sparse volumes, i.e. reducing the number of cases annotated, and sparse annotations, i.e. reducing the number of slices annotated per case. We evaluate the segmentation performance using the state-of-the-art nnU-Net model on two public datasets to identify which slices are the most important to annotate. We have shown that training on a significantly reduced dataset (48 annotated volumes) can give a Dice score greater than 0.85 and results comparable to using the full dataset (160 and 240 volumes for each dataset respectively). In general, training on more slice annotations provides more valuable information compared to training on more volumes. Further, annotating slices from the middle of volumes yields the most beneficial results in terms of segmentation performance, and the apical region the worst. When evaluating the trade-off between annotating volumes against slices, annotating as many slices as possible instead of annotating more volumes is a better strategy.

Abstract: The automatic lung lobe segmentation algorithm is of great significance for the diagnosis and treatment of lung diseases, however, which has great challenges due to the incompleteness of pulmonary fissures in lung CT images and the large variability of pathological features. Therefore, we propose a new automatic lung lobe segmentation framework, in which we urge the model to pay attention to the area around the pulmonary fissure during the training process, which is realized by a task-specific loss function. In addition, we introduce an end-to-end pulmonary fissure generation method in the auxiliary pulmonary fissure segmentation task, without any additional network branch. Finally, we propose a registration-based loss function to alleviate the convergence difficulty of the Dice loss supervised pulmonary fissure segmentation task. We achieve 97.83% and 94.75% dice scores on our private dataset STLB and public LUNA16 dataset respectively.

Abstract: Remote Photoplethysmography (rPPG) is a technology that utilizes the light absorption properties of hemoglobin, captured via camera, to analyze and measure blood volume pulse (BVP). By analyzing the measured BVP, various physiological signals such as heart rate, stress levels, and blood pressure can be derived, enabling applications such as the early prediction of cardiovascular diseases. rPPG is a rapidly evolving field as it allows the measurement of vital signals using camera-equipped devices without the need for additional devices such as blood pressure monitors or pulse oximeters, and without the assistance of medical experts. Despite extensive efforts and advances in this field, serious challenges remain, including issues related to skin color, camera characteristics, ambient lighting, and other sources of noise, which degrade performance accuracy. We argue that fair and evaluable benchmarking is urgently required to overcome these challenges and make any meaningful progress from both academic and commercial perspectives. In most existing work, models are trained, tested, and validated only on limited datasets. Worse still, some studies lack available code or reproducibility, making it difficult to fairly evaluate and compare performance. Therefore, the purpose of this study is to provide a benchmarking framework to evaluate various rPPG techniques across a wide range of datasets for fair evaluation and comparison, including both conventional non-deep neural network (non-DNN) and deep neural network (DNN) methods. GitHub URL: https://github.com/remotebiosensing/rppg.

Abstract: In dynamic Magnetic Resonance Imaging (MRI), k-space is typically undersampled due to limited scan time, resulting in aliasing artifacts in the image domain. Hence, dynamic MR reconstruction requires not only modeling spatial frequency components in the x and y directions of k-space but also considering temporal redundancy. Most previous works rely on image-domain regularizers (priors) to conduct MR reconstruction. In contrast, we focus on interpolating the undersampled k-space before obtaining images with Fourier transform. In this work, we connect masked image modeling with k-space interpolation and propose a novel Transformer-based k-space Global Interpolation Network, termed k-GIN. Our k-GIN learns global dependencies among low- and high-frequency components of 2D+t k-space and uses it to interpolate unsampled data. Further, we propose a novel k-space Iterative Refinement Module (k-IRM) to enhance the high-frequency components learning. We evaluate our approach on 92 in-house 2D+t cardiac MR subjects and compare it to MR reconstruction methods with image-domain regularizers. Experiments show that our proposed k-space interpolation method quantitatively and qualitatively outperforms baseline methods. Importantly, the proposed approach achieves substantially higher robustness and generalizability in cases of highly-undersampled MR data.

Abstract: Deploying 3D single-photon Lidar imaging in real world applications faces several challenges due to imaging in high noise environments and with sensors having limited resolution. This paper presents a deep learning algorithm based on unrolling a Bayesian model for the reconstruction and super-resolution of 3D single-photon Lidar. The resulting algorithm benefits from the advantages of both statistical and learning based frameworks, providing best estimates with improved network interpretability. Compared to existing learning-based solutions, the proposed architecture requires a reduced number of trainable parameters, is more robust to noise and mismodelling of the system impulse response function, and provides richer information about the estimates including uncertainty measures. Results on synthetic and real data show competitive results regarding the quality of the inference and computational complexity when compared to state-of-the-art algorithms. This short paper is based on contributions published in [1] and [2].

Abstract: We propose a neural image codec at reduced complexity which overfits the decoder parameters to each input image. While autoencoders perform up to a million multiplications per decoded pixel, the proposed approach only requires 2300 multiplications per pixel. Albeit low-complexity, the method rivals autoencoder performance and surpasses HEVC performance under various coding conditions. Additional lightweight modules and an improved training process provide a 14% rate reduction with respect to previous overfitted codecs, while offering a similar complexity. This work is made open-source at https://orange-opensource.github.io/Cool-Chic/

Abstract: Single image super-resolution (SISR) is a challenging ill-posed problem that aims to up-sample a given low-resolution (LR) image to a high-resolution (HR) counterpart. Due to the difficulty in obtaining real LR-HR training pairs, recent approaches are trained on simulated LR images degraded by simplified down-sampling operators, e.g., bicubic. Such an approach can be problematic in practice because of the large gap between the synthesized and real-world LR images. To alleviate the issue, we propose a novel Invertible scale-Conditional Function (ICF), which can scale an input image and then restore the original input with different scale conditions. By leveraging the proposed ICF, we construct a novel self-supervised SISR framework (ICF-SRSR) to handle the real-world SR task without using any paired/unpaired training data. Furthermore, our ICF-SRSR can generate realistic and feasible LR-HR pairs, which can make existing supervised SISR networks more robust. Extensive experiments demonstrate the effectiveness of the proposed method in handling SISR in a fully self-supervised manner. Our ICF-SRSR demonstrates superior performance compared to the existing methods trained on synthetic paired images in real-world scenarios and exhibits comparable performance compared to state-of-the-art supervised/unsupervised methods on public benchmark datasets.

Abstract: Graph-based neural network models are gaining traction in the field of representation learning due to their ability to uncover latent topological relationships between entities that are otherwise challenging to identify. These models have been employed across a diverse range of domains, encompassing drug discovery, protein interactions, semantic segmentation, and fluid dynamics research. In this study, we investigate the potential of Graph Neural Networks (GNNs) for medical image classification. We introduce a novel model that combines GNNs and edge convolution, leveraging the interconnectedness of RGB channel feature values to strongly represent connections between crucial graph nodes. Our proposed model not only performs on par with state-of-the-art Deep Neural Networks (DNNs) but does so with 1000 times fewer parameters, resulting in reduced training time and data requirements. We compare our Graph Convolutional Neural Network (GCNN) to pre-trained DNNs for classifying MedMNIST dataset classes, revealing promising prospects for GNNs in medical image analysis. Our results also encourage further exploration of advanced graph-based models such as Graph Attention Networks (GAT) and Graph Auto-Encoders in the medical imaging domain. The proposed model yields more reliable, interpretable, and accurate outcomes for tasks like semantic segmentation and image classification compared to simpler GCNNs

Abstract: Hyperspectral imaging shows promise for surgical applications to non-invasively provide spatially-resolved, spectral information. For calibration purposes, a white reference image of a highly-reflective Lambertian surface should be obtained under the same imaging conditions. Standard white references are not sterilizable, and so are unsuitable for surgical environments. We demonstrate the necessity for in situ white references and address this by proposing a novel, sterile, synthetic reference construction algorithm. The use of references obtained at different distances and lighting conditions to the subject were examined. Spectral and color reconstructions were compared with standard measurements qualitatively and quantitatively, using $\Delta E$ and normalised RMSE respectively. The algorithm forms a composite image from a video of a standard sterile ruler, whose imperfect reflectivity is compensated for. The reference is modelled as the product of independent spatial and spectral components, and a scalar factor accounting for gain, exposure, and light intensity. Evaluation of synthetic references against ideal but non-sterile references is performed using the same metrics alongside pixel-by-pixel errors. Finally, intraoperative integration is assessed though cadaveric experiments. Improper white balancing leads to increases in all quantitative and qualitative errors. Synthetic references achieve median pixel-by-pixel errors lower than 6.5% and produce similar reconstructions and errors to an ideal reference. The algorithm integrated well into surgical workflow, achieving median pixel-by-pixel errors of 4.77%, while maintaining good spectral and color reconstruction.

Abstract: In this work, we investigate laparoscopic camera motion automation through imitation learning from retrospective videos of laparoscopic interventions. A novel method is introduced that learns to augment a surgeon's behavior in image space through object motion invariant image registration via homographies. Contrary to existing approaches, no geometric assumptions are made and no depth information is necessary, enabling immediate translation to a robotic setup. Deviating from the dominant approach in the literature which consist of following a surgical tool, we do not handcraft the objective and no priors are imposed on the surgical scene, allowing the method to discover unbiased policies. In this new research field, significant improvements are demonstrated over two baselines on the Cholec80 and HeiChole datasets, showcasing an improvement of 47% over camera motion continuation. The method is further shown to indeed predict camera motion correctly on the public motion classification labels of the AutoLaparo dataset. All code is made accessible on GitHub.

Abstract: Accurately localizing and identifying vertebrae from CT images is crucial for various clinical applications. However, most existing efforts are performed on 3D with cropping patch operation, suffering from the large computation costs and limited global information. In this paper, we propose a multi-view vertebra localization and identification from CT images, converting the 3D problem into a 2D localization and identification task on different views. Without the limitation of the 3D cropped patch, our method can learn the multi-view global information naturally. Moreover, to better capture the anatomical structure information from different view perspectives, a multi-view contrastive learning strategy is developed to pre-train the backbone. Additionally, we further propose a Sequence Loss to maintain the sequential structure embedded along the vertebrae. Evaluation results demonstrate that, with only two 2D networks, our method can localize and identify vertebrae in CT images accurately, and outperforms the state-of-the-art methods consistently. Our code is available at https://github.com/ShanghaiTech-IMPACT/Multi-View-Vertebra-Localization-and-Identification-from-CT-Images.

Abstract: As a fundamental part of computational healthcare, Computer Tomography (CT) and Magnetic Resonance Imaging (MRI) provide volumetric data, making the development of algorithms for 3D image analysis a necessity. Despite being computationally cheap, 2D Convolutional Neural Networks can only extract spatial information. In contrast, 3D CNNs can extract three-dimensional features, but they have higher computational costs and latency, which is a limitation for clinical practice that requires fast and efficient models. Inspired by the field of video action recognition we propose a new 2D-based model dubbed Slice SHift UNet (SSH-UNet) which encodes three-dimensional features at 2D CNN's complexity. More precisely multi-view features are collaboratively learned by performing 2D convolutions along the three orthogonal planes of a volume and imposing a weights-sharing mechanism. The third dimension, which is neglected by the 2D convolution, is reincorporated by shifting a portion of the feature maps along the slices' axis. The effectiveness of our approach is validated in Multi-Modality Abdominal Multi-Organ Segmentation (AMOS) and Multi-Atlas Labeling Beyond the Cranial Vault (BTCV) datasets, showing that SSH-UNet is more efficient while on par in performance with state-of-the-art architectures.

Abstract: Conditional coding is a new video coding paradigm enabled by neural-network-based compression. It can be shown that conditional coding is in theory better than the traditional residual coding, which is widely used in video compression standards like HEVC or VVC. However, on closer inspection, it becomes clear that conditional coders can suffer from information bottlenecks in the prediction path, i.e., that due to the data processing inequality not all information from the prediction signal can be passed to the reconstructed signal, thereby impairing the coder performance. In this paper we propose the conditional residual coding concept, which we derive from information theoretical properties of the conditional coder. This coder significantly reduces the influence of bottlenecks, while maintaining the theoretical performance of the conditional coder. We provide a theoretical analysis of the coding paradigm and demonstrate the performance of the conditional residual coder in a practical example. We show that conditional residual coders alleviate the disadvantages of conditional coders while being able to maintain their advantages over residual coders. In the spectrum of residual and conditional coding, we can therefore consider them as ``the best from both worlds''.