Medical image synthesis is an important topic for both clinical and research applications. Recently, diffusion models have become a leading approach in this area. Despite their strengths, many existing methods struggle with (1) limited generalizability, only working for specific body regions or voxel spacings, (2) slow inference, which is a common issue for diffusion models, and (3) weak alignment with input conditions, which is a critical issue for medical imaging. MAISI, a previously proposed framework, addresses generalizability issues but still suffers from slow inference and limited condition consistency. In this work, we present MAISI-v2, the first accelerated 3D medical image synthesis framework that integrates rectified flow to enable fast and high-quality generation. To further enhance condition fidelity, we introduce a novel region-specific contrastive loss to improve sensitivity to the region of interest. Our experiments show that MAISI-v2 can achieve state-of-the-art image quality with 33× acceleration for latent diffusion models. We also conducted a downstream segmentation experiment to show that the synthetic images can be used for data augmentation. We release our code, training details, model weights, and a GUI demo to facilitate reproducibility and promote further development within the community.

Tumor synthesis can generate challenging cases that AI often misses or over-detects. Training on these cases improves AI performance. However, most existing synthesis methods are either unconditional- generating images from random variables-or conditioned only on tumor shape. As a result, they lack control over clinically important tumor characteristics, such as texture, heterogeneity, boundary, and pathology. The generated tumors are therefore overly similar or duplicates of existing training cases, failing to effectively address AI's weaknesses. We propose a new text-driven tumor synthesis approach, termed TextoMorph, that provides textual control over tumor characteristics in conjunction with mask control. This approach is particularly beneficial for examples that confuse the AI the most, such as early tumor detection (improving Sensitivity by + 6.5%), tumor segmentation for precise radiotherapy (improving NSD by + 3.1%), and classification between benign and malignant tumors (improving Sensitivity by + 8.2%). By incorporating text mined from radiology reports into the synthesis process, we increase the variability and controllability of the synthetic tumors to target AI's failure cases more precisely. Moreover, TextoMorph uses contrastive learning across different texts and CT scans, significantly reducing dependence on scarce image-report pairs (only 141 pairs used in this study) by leveraging a large corpus of 34,035 radiology reports. Finally, we have developed rigorous tests to evaluate synthetic tumors, showing that our synthetic tumors is realistic and diverse in texture, heterogeneity, boundary, and pathology. Code and models are available at https://github.com/MrGiovanni/TextoMorph.

Foundation models for interactive segmentation in 2D natural images and videos have sparked significant interest in building 3D foundation models for medical imaging. However, the domain gaps and clinical use cases for 3D medical imaging require a dedicated model that diverges from existing 2D solutions. Specifically, such foundation models should support a full workflow that can actually reduce human effort. Treating 3D medical images as sequences of 2D slices and reusing interactive 2D foundation models seems straightforward, but 2D annotation is too time-consuming for 3D tasks. Moreover, for large cohort analysis, it’s the highly accurate automatic segmentation models that reduce the most human effort. However, these models lack support for interactive corrections and lack zero-shot ability for novel structures, which is a key feature of "foundation". While reusing pre-trained 2D backbones in 3D enhances zero-shot potential, their performance on complex 3D structures still lags behind leading 3D models. To address these issues, we present VISTA3D, Versatile Imaging SegmenTation and Annotation model, that targets to solve all these challenges and requirements with one unified foundation model. VISTA3D is built on top of the well-established 3D segmentation pipeline, and it is the first model to achieve state-of-the-art performance in both 3D automatic (supporting 127 classes) and 3D interactive segmentation, even when compared with top 3D expert models on large and diverse benchmarks. Additionally, VISTA3D’s 3D interactive design allows efficient human correction, and a novel 3D supervoxel method that distills 2D pre-trained backbones grants VISTA3D top 3D zero-shot performance. We believe the model, recipe, and insights represent a promising step towards a clinically useful 3D foundation model. Code and weights are publicly available at https://github.com/Project-MONAI/VISTA.

Generalist vision language models (VLMs) have made significant strides in computer vision, but they fall short in specialized fields like healthcare, where expert knowledge is essential. Current large multimodal models like Gemini and GPT-4o are insufficient for medical tasks due to their reliance on memorized internet knowledge rather than the nuanced expertise required in healthcare. Meanwhile, existing medical VLMs (e.g. Med-Gemini) often lack expert consultation as part of their design, and many rely on outdated, static datasets that were not created with modern, large deep learning models in mind. VLMs are usually trained in three stages: vision pre-training, vision-language pre-training, and instruction fine-tuning (IFT). IFT has been typically applied using a mixture of generic and healthcare data. In contrast, we propose that for medical VLMs, a fourth stage of specialized IFT is necessary, which focuses on medical data and includes information from domain expert models. Domain expert models developed for medical use are crucial because they are specifically trained for certain clinical tasks, e.g. to detect tumors and classify abnormalities through segmentation and classification, which learn fine-grained features of medical data−features that are often too intricate for a VLM to capture effectively. This paper introduces a new framework, VILA-M3, for medical VLMs that utilizes domain knowledge via expert models. We argue that generic VLM architectures alone are not viable for real-world clinical applications and on-demand usage of domain-specialized expert model knowledge is critical for advancing AI in healthcare. Through our experiments, we show an improved state-of-the-art (SOTA) performance with an average improvement of ~9% over the prior SOTA model Med-Gemini and ~6% over models trained on the specific tasks. Our approach emphasizes the importance of domain expertise in creating precise, reliable VLMs for medical applications.

Recent progress in vision-language modeling for 3D medical imaging has been fueled by large-scale computed tomography (CT) corpora with paired free-text reports, stronger architectures, and powerful pretrained models. This has enabled applications such as automated report generation and text-conditioned 3D image synthesis. Yet, current approaches struggle with high-resolution, long-sequence volumes: contrastive pretraining often yields vision encoders that are misaligned with clinical language, and slice-wise tokenization blurs fine anatomy, reducing diagnostic performance on downstream tasks. We introduce BTB3D (Better Tokens for Better 3D), a causal convolutional encoder-decoder that unifies 2D and 3D training and inference while producing compact, frequency-aware volumetric tokens. A three-stage training curriculum enables (i) local reconstruction, (ii) overlapping-window tiling, and (iii) long-context decoder refinement, during which the model learns from short slice excerpts yet generalizes to scans exceeding $300$ slices without additional memory overhead. BTB3D sets a new state-of-the-art on two key tasks: it improves BLEU scores and increases clinical F1 by 40\% over CT2Rep, CT-CHAT, and Merlin for report generation; and it reduces FID by 75\% and halves FVD compared to GenerateCT and MedSyn for text-to-CT synthesis, producing anatomically consistent $512\times512\times241$ volumes. These results confirm that precise three-dimensional tokenization, rather than larger language backbones alone, is essential for scalable vision-language modeling in 3D medical imaging. The codebase is available at: https://github.com/ibrahimethemhamamci/BTB3D

Medical imaging analysis faces challenges such as data scarcity, high annotation costs, and privacy concerns. This paper introduces the Medical AI for Synthetic Imaging (MAISI), an innovative approach using the diffusion model to generate synthetic 3D computed tomography (CT) images to address those challenges. MAISI leverages the foundation volume compression network and the latent diffusion model to produce high-resolution CT images (up to a landmark volume dimension of 512 × 512 × 768) with flexible volume dimensions and voxel spacing. By incorporating ControlNet, MAISI can process organ segmentation, including 127 anatomical structures, as additional conditions and enables the generation of accurately annotated synthetic images that can be used for various downstream tasks. Our experiment results show that MAISI's capabilities in generating realistic, anatomically accurate images for diverse regions and conditions reveal its promising potential to mitigate challenges using synthetic data.

Pre-trained language models (PLM) have revolutionized the NLP landscape, achieving stellar performances across diverse tasks. These models, while benefiting from vast training data, often require fine-tuning on specific data to cater to distinct downstream tasks. However, this data adaptation process has inherent security and privacy concerns, primarily when leveraging user-generated, device-residing data. Federated learning (FL) provides a solution, allowing collaborative model fine-tuning without centralized data collection. However, applying FL to finetune PLMs is hampered by challenges, including restricted model parameter access due to the high encapsulation, high computational requirements, and communication overheads. This paper introduces Federated Black-box Prompt Tuning (FedBPT), a framework designed to address these challenges. FedBPT allows the clients to treat the model as a black-box inference API. By focusing on training optimal prompts and utilizing gradient-free optimization methods, FedBPT reduces the number of exchanged variables, boosts communication efficiency, and minimizes computational and storage costs. Experiments highlight the framework’s ability to drastically cut communication and memory costs while maintaining competitive performance. Ultimately, FedBPT presents a promising solution for efficient, privacy-preserving fine-tuning of PLM in the age of large language models.

Multi-modal registration between tagged and untagged cardiac cine magnetic resonance (MR) images remains difficult, due to the domain gap and large deformations between the two modalities. Recent work using an image-to-image translation (I2I) module to overcome the domain gap can convert the multi-modal into a mono-modal registration task and take advantage of advanced mono-modal registration architectures. However, they often ignore two issues: the sample-specific style of each image to be registered during I2I and large hybrid rigid and non-rigid deformations between modalities. We first propose an exemplar-based I2I module capable of unsupervised cross-domain correspondence learning to enforce the style consistency between the fake image and the image to be registered. Then we propose an efficient cascaded vision transformer-based registration network to predict both the affine and non-rigid deformations, in which a single feature embedding subnetwork is shared by the two stages of deformation prediction. We validated our method on a clinical cardiac MR dataset with paired but unaligned untagged and tagged MR images. The results show that our method outperforms traditional methods significantly in terms of the I2I quality and multi-modal image registration accuracy.

Supervised learning paradigms largely benefit from the tremendous amount of annotated data. However, the quality of the annotations often varies among labelers. Multi-observer studies have been conducted to examine the annotation variances (by labeling the same data multiple times) to see how it affects critical applications like medical image analysis. In this paper, we demonstrate how multiple sets of annotations (either hand-labeled or algorithm-generated) can be utilized together and mutually benefit the learning of classification tasks. A scheme of learning-to-vote is introduced to sample quality label sets for each data entry on-the-fly during the training. Specifically, a label-sampling module is designed to achieve refined labels (weighted sum of attended ones) that benefit the model learning the most through additional back-propagations. We apply the learning-to-vote scheme on the classification task of a synthetic noisy CIFAR-10 to prove the concept and then demonstrate superior results (3-5% increase on average in multiple disease classification AUCs) on the chest x-ray images from a hospital-scale dataset (MIMIC-CXR) and hand-labeled dataset (OpenI) in comparison to regular training paradigms.

How to ensure fairness is an important topic in federated learning (FL). Recent studies have investigated how to reward clients based on their contribution (collaboration fairness), and how to achieve uniformity of performance across clients (performance fairness). Despite achieving progress on either one, we argue that it is critical to consider them together, in order to engage and motivate more diverse clients joining FL to derive a high-quality global model. In this work, we propose a novel method to optimize both types of fairness simultaneously. Specifically, we propose to estimate client contribution in gradient and data space. In gradient space, we monitor the gradient direction differences of each client with respect to others. And in data space, we measure the prediction error on client data using an auxiliary model. Based on this contribution estimation, we propose a FL method, federated training via contribution estimation (FedCE), i.e., using estimation as global model aggregation weights. We have theoretically analyzed our method and empirically evaluated it on two real-world medical datasets. The effectiveness of our approach has been validated with significant performance improvements, better collaboration fairness, better performance fairness, and comprehensive analytical studies. Code is available at https://nvidia.github.io/NVFlare/research/fed-ce

Federated learning is a popular collaborative learning approach that enables clients to train a global model without sharing their local data. Vertical federated learning (VFL) deals with scenarios in which the data on clients have different feature spaces but share some overlapping samples. Existing VFL approaches suffer from high communication costs and cannot deal efficiently with limited overlapping samples commonly seen in the real world. We propose a practical VFL framework called one-shot VFL that can solve the communication bottleneck and the problem of limited overlapping samples simultaneously based on semi-supervised learning. We also propose few-shot VFL to improve the accuracy further with just one more communication round between the server and the clients. In our proposed framework, the clients only need to communicate with the server once or only a few times. We evaluate the proposed VFL framework on both image and tabular datasets. Our methods can improve the accuracy by more than 46.5% and reduce the communication cost by more than 330× compared with state-of-the-art VFL methods when evaluated on CIFAR-10. Our code is available at https://nvidia.github.io/NVFlare/research/one-shot-vfl.

We propose a novel neural deformable model (NDM) targeting at the reconstruction and modeling of 3D bi-ventricular shape of the heart from 2D sparse cardiac magnetic resonance (CMR) imaging data. We model the bi-ventricular shape using blended deformable superquadrics, which are parameterized by a set of geometric parameter functions and are capable of deforming globally and locally. While global geometric parameter functions and deformations capture gross shape features from visual data, local deformations, parameterized as neural diffeomorphic point flows, can be learned to recover the detailed heart shape. Different from iterative optimization methods used in conventional deformable model formulations, NDMs can be trained to learn such geometric parameter functions, global and local deformations from a shape distribution manifold. Our NDM can learn to densify a sparse cardiac point cloud with arbitrary scales and generate high-quality triangular meshes automatically. It also enables the implicit learning of dense correspondences among different heart shape instances for accurate cardiac shape registration. Furthermore, the parameters of NDM are intuitive, and can be used by a physician without sophisticated post-processing. Experimental results on a large CMR dataset demonstrate the improved performance of NDM over conventional methods.

Modern medical imaging techniques, such as ultrasound (US) and cardiac magnetic resonance (MR) imaging, have enabled the evaluation of myocardial deformation directly from an image sequence. While many traditional cardiac motion tracking methods have been developed for the automated estimation of the myocardial wall deformation, they are not widely used in clinical diagnosis, due to their lack of accuracy and efficiency. In this paper, we propose a novel deep learning-based fully unsupervised method, SequenceMorph, for in vivo motion tracking in cardiac image sequences. In our method, we introduce the concept of motion decomposition and recomposition. We first estimate the inter-frame (INF) motion field between any two consecutive frames, by a bi-directional generative diffeomorphic registration neural network. Using this result, we then estimate the Lagrangian motion field between the reference frame and any other frame, through a differentiable composition layer. Our framework can be extended to incorporate another registration network, to further reduce the accumulated errors introduced in the INF motion tracking step, and to refine the Lagrangian motion estimation. By utilizing temporal information to perform reasonable estimations of spatio-temporal motion fields, this novel method provides a useful solution for image sequence motion tracking. Our method has been applied to US (echocardiographic) and cardiac MR (untagged and tagged cine) image sequences; the results show that SequenceMorph is significantly superior to conventional motion tracking methods, in terms of the cardiac motion tracking accuracy and inference efficiency.

Semantic segmentation of 3D medical images is a challenging task due to the high variability of the shape and pattern of objects (such as organs or tumors). Given the recent success of deep learning in medical image segmentation, Neural Architecture Search (NAS) has been introduced to find high-performance 3D segmentation network architectures. However, because of the massive computational requirements of 3D data and the discrete optimization nature of architecture search, previous NAS methods require a long search time or necessary continuous relaxation, and commonly lead to sub-optimal network architectures. While one-shot NAS can potentially address these disadvantages, its application in the segmentation domain has not been well studied in the expansive multi-scale multi-path search space. To enable one-shot NAS for medical image segmentation, our method, named HyperSegNAS, introduces a HyperNet to assist super-net training by incorporating architecture topology information. Such a HyperNet can be removed once the super-net is trained and introduces no overhead during architecture search. We show that HyperSegNAS yields better performing and more intuitive architectures compared to the previous state-of-the-art (SOTA) segmentation networks; furthermore, it can quickly and accurately find good architecture candidates under different computing constraints. Our method is evaluated on public datasets from the Medical Segmentation Decathlon (MSD) challenge, and achieves SOTA performances.

Vision Transformers (ViT)s have shown great performance in self-supervised learning of global and local representations that can be transferred to downstream applications. Inspired by these results, we introduce a novel self-supervised learning framework with tailored proxy tasks for medical image analysis. Specifically, we propose: (i) a new 3D transformer-based model, dubbed Swin UNEt TRansformers (Swin UNETR), with a hierarchical encoder for self-supervised pretraining; (ii) tailored proxy tasks for learning the underlying pattern of human anatomy. We demonstrate successful pre-training of the proposed model on 5,050 publicly available computed tomography (CT) images from various body organs. The effectiveness of our approach is validated by fine-tuning the pre-trained models on the Beyond the Cranial Vault (BTCV) Segmentation Challenge with 13 abdominal organs and segmentation tasks from the Medical Segmentation Decathlon (MSD) dataset. Our model is currently the state-of-the-art on the public test leaderboards of both MSD11https://decathlon-10.grand-challenge.org/evaluation/challenge/leaderboard/ and BTCV22https://www.synapse.org/#!Synapse:syn3193805/wiki/217785/ datasets. Code: https://monai.io/research/swin-unetr.

Cross-silo federated learning (FL) has attracted much attention in medical imaging analysis with deep learning in recent years as it can resolve the critical issues of insufficient data, data privacy, and training efficiency. However, there can be a generalization gap between the model trained from FL and the one from centralized training. This important issue comes from the non-iid data distribution of the local data in the participating clients and is well-known as client drift. In this work, we propose a novel training frame-work FedSM to avoid the client drift issue and successfully close the generalization gap compared with the centralized training for medical image segmentation tasks for the first time. We also propose a novel personalized FL objective formulation and a new method SoftPull to solve it in our proposed framework FedSM. We conduct rigorous theoretical analysis to guarantee its convergence for optimizing the non-convex smooth objective function. Real-world medical image segmentation experiments using deep FL validate the motivations and effectiveness of our proposed method.

Fully Convolutional Neural Networks (FCNNs) with contracting and expanding paths have shown prominence for the majority of medical image segmentation applications since the past decade. In FCNNs, the encoder plays an integral role by learning both global and local features and contextual representations which can be utilized for semantic output prediction by the decoder. Despite their success, the locality of convolutional layers in FCNNs, limits the capability of learning long-range spatial dependencies. Inspired by the recent success of transformers for Natural Language Processing (NLP) in long-range sequence learning, we reformulate the task of volumetric (3D) medical image segmentation as a sequence-to-sequence prediction problem. We introduce a novel architecture, dubbed as UNEt TRansformers (UNETR), that utilizes a transformer as the encoder to learn sequence representations of the input volume and effectively capture the global multi-scale information, while also following the successful "U-shaped" network design for the encoder and decoder. The transformer encoder is directly connected to a decoder via skip connections at different resolutions to compute the final semantic segmentation output. We have validated the performance of our method on the Multi Atlas Labeling Beyond The Cranial Vault (BTCV) dataset for multi-organ segmentation and the Medical Segmentation Decathlon (MSD) dataset for brain tumor and spleen segmentation tasks. Our benchmarks demonstrate new state-of-the-art performance on the BTCV leaderboard.

Recently, neural architecture search (NAS) has been applied to automatically search high-performance networks for medical image segmentation. The NAS search space usually contains a network topology level (controlling connections among cells with different spatial scales) and a cell level (operations within each cell). Existing methods either require long searching time for large-scale 3D image datasets, or are limited to pre-defined topologies (such as U-shaped or single-path) . In this work, we focus on three important aspects of NAS in 3D medical image segmentation: flexible multi-path network topology, high search efficiency, and budgeted GPU memory usage. A novel differentiable search framework is proposed to support fast gradient-based search within a highly flexible network topology search space. The discretization of the searched optimal continuous model in differentiable scheme may produce a sub-optimal final discrete model (discretization gap). Therefore, we propose a topology loss to alleviate this problem. In addition, the GPU memory usage for the searched 3D model is limited with budget constraints during search. Our Differentiable Network Topology Search scheme (DiNTS) is evaluated on the Medical Segmentation Decathlon (MSD) challenge, which contains ten challenging segmentation tasks. Our method achieves the state-of-the-art performance and the top ranking on the MSD challenge leaderboard.

Cardiac tagging magnetic resonance imaging (t-MRI) is the gold standard for regional myocardium deformation and cardiac strain estimation. However, this technique has not been widely used in clinical diagnosis, as a result of the difficulty of motion tracking encountered with t-MRI images. In this paper, we propose a novel deep learning-based fully unsupervised method for in vivo motion tracking on t-MRI images. We first estimate the motion field (INF) between any two consecutive t-MRI frames by a bi-directional generative diffeomorphic registration neural network. Using this result, we then estimate the Lagrangian motion field between the reference frame and any other frame through a differentiable composition layer. By utilizing temporal information to perform reasonable estimations on spatiotemporal motion fields, this novel method provides a useful solution for motion tracking and image registration in dynamic medical imaging. Our method has been validated on a representative clinical t-MRI dataset; the experimental results show that our method is superior to conventional motion tracking methods in terms of landmark tracking accuracy and inference efficiency. Project page is at: https://github.com/DeepTag/cardiac_tagging_motion_estimation.

Lesion segmentation in medical imaging has been an important topic in clinical research. Researchers have proposed various detection and segmentation algorithms to address this task. Recently, deep learning-based approaches have significantly improved the performance over conventional methods. However, most state-of-the-art deep learning methods require the manual design of multiple network components and training strategies. In this paper, we propose a new automated machine learning algorithm, T-AutoML, which not only searches for the best neural architecture, but also finds the best combination of hyper-parameters and data augmentation strategies simultaneously. The proposed method utilizes the modern transformer model, which is introduced to adapt to the dynamic length of the search space embedding and can significantly improve the ability of the search. We validate T-AutoML on several large-scale public lesion segmentation data-sets and achieve state-of-the-art performance.

Active learning is a unique abstraction of machine learning techniques where the model/algorithm could guide users for annotation of a set of data points that would be beneficial to the model, unlike passive machine learning. The primary advantage being that active learning frameworks select data points that can accelerate the learning process of a model and can reduce the amount of data needed to achieve full accuracy as compared to a model trained on a randomly acquired data set. Multiple frameworks for active learning combined with deep learning have been proposed, and the majority of them are dedicated to classification tasks. Herein, we explore active learning for the task of segmentation of medical imaging data sets. We investigate our proposed framework using two datasets: 1.) MRI scans of the hippocampus, 2.) CT scans of pancreas and tumors. This work presents a query-by-committee approach for active learning where a joint optimizer is used for the committee. At the same time, we propose three new strategies for active learning: 1.) increasing frequency of uncertain data to bias the training data set; 2.) Using mutual information among the input images as a regularizer for acquisition to ensure diversity in the training dataset; 3.) adaptation of Dice log-likelihood for Stein variational gradient descent (SVGD). The results indicate an improvement in terms of data reduction by achieving full accuracy while only using 22.69% and 48.85% of the available data for each dataset, respectively.

3D convolution neural networks (CNN) have been proved very successful in parsing organs or tumours in 3D medical images, but it remains sophisticated and time-consuming to choose or design proper 3D networks given different task contexts. Recently, Neural Architecture Search (NAS) is proposed to solve this problem by searching for the best network architecture automatically. However, the inconsistency between search stage and deployment stage often exists in NAS algorithms due to memory constraints and large search space, which could become more serious when applying NAS to some memory and time-consuming tasks, such as 3D medical image segmentation. In this paper, we propose a coarse-to-fine neural architecture search (C2FNAS) to automatically search a 3D segmentation network from scratch without inconsistency on network size or input size. Specifically, we divide the search procedure into two stages: 1) the coarse stage, where we search the macro-level topology of the network, i.e. how each convolution module is connected to other modules; 2) the fine stage, where we search at micro-level for operations in each cell based on previous searched macro-level topology. The coarse-to-fine manner divides the search procedure into two consecutive stages and meanwhile resolves the inconsistency. We evaluate our method on 10 public datasets from Medical Segmentation Decalthon (MSD) challenge, and achieve state-of-the-art performance with the network searched using one dataset, which demonstrates the effectiveness and generalization of our searched models.

While making a tremendous impact in various fields, deep neural networks usually require large amounts of labeled data for training which are expensive to collect in many applications, especially in the medical domain. Un-labeled data, on the other hand, is much more abundant. Semi-supervised learning techniques, such as co-training, could provide a powerful tool to leverage unlabeled data. In this paper, we propose a novel framework, uncertainty-aware multi-view co-training (UMCT), to address semi-supervised learning on 3D data, such as volumetric data from medical imaging. In our work, co-training is achieved by exploiting multi-viewpoint consistency of 3D data. We generate different views by rotating or permuting the 3D data and utilize asymmetrical 3D kernels to encourage diversified features in different sub-networks. In addition, we propose an uncertainty-weighted label fusion mechanism to estimate the reliability of each view's prediction with Bayesian deep learning. As one view requires the supervision from other views in co-training, our self-adaptive approach computes a confidence score for the prediction of each unlabeled sample in order to assign a reliable pseudo label. Thus, our approach can take advantage of unlabeled data during training. We show the effectiveness of our proposed semi-supervised method on several public datasets from medical image segmentation tasks (NIH pancreas & LiTS liver tumor dataset). Meanwhile, a fully-supervised method based on our approach achieved state-of-the-art performances on both the LiTS liver tumor segmentation and the Medical Segmentation Decathlon (MSD) challenge, demonstrating the robustness and value of our framework, even when fully supervised training is feasible.

Recent advances in deep learning for medical image segmentation demonstrate expert-level accuracy. However, application of these models in clinically realistic environments can result in poor generalization and decreased accuracy, mainly due to the domain shift across different hospitals, scanner vendors, imaging protocols, and patient populations etc. Common transfer learning and domain adaptation techniques are proposed to address this bottleneck. However, these solutions require data (and annotations) from the target domain to retrain the model, and is therefore restrictive in practice for widespread model deployment. Ideally, we wish to have a trained (locked) model that can work uniformly well across unseen domains without further training. In this paper, we propose a deep stacked transformation approach for domain generalization. Specifically, a series of n stacked transformations are applied to each image during network training. The underlying assumption is that the "expected" domain shift for a specific medical imaging modality could be simulated by applying extensive data augmentation on a single source domain, and consequently, a deep model trained on the augmented "big" data (BigAug) could generalize well on unseen domains. We exploit four surprisingly effective, but previously understudied, image-based characteristics for data augmentation to overcome the domain generalization problem. We train and evaluate the BigAug model (with n=9 transformations) on three different 3D segmentation tasks (prostate gland, left atrial, left ventricle) covering two medical imaging modalities (MRI and ultrasound) involving eight publicly available challenge datasets. The results show that when training on relatively small dataset (n = 10~32 volumes, depending on the size of the available datasets) from a single source domain: (i) BigAug models degrade an average of 11%(Dice score change) from source to unseen domain, substantially better than conventional augmentation (degrading 39%) and CycleGAN-based domain adaptation method (degrading 25%), (ii) BigAug is better than "shallower" stacked transforms (i.e. those with fewer transforms) on unseen domains and demonstrates modest improvement to conventional augmentation on the source domain, (iii) after training with BigAug on one source domain, performance on an unseen domain is similar to training a model from scratch on that domain when using the same number of training samples. When training on large datasets (n = 465 volumes) with BigAug, (iv) application to unseen domains reaches the performance of state-of-the-art fully supervised models that are trained and tested on their source domains. These findings establish a strong benchmark for the study of domain generalization in medical imaging, and can be generalized to the design of highly robust deep segmentation models for clinical deployment.

Deep learning algorithms, in particular 2D and 3D fully convolutional neural networks (FCNs), have rapidly become the mainstream methodology for volumetric medical image segmentation. However, 2D convolutions cannot fully leverage the rich spatial information along the third axis, while 3D convolutions suffer from the demanding computation and high GPU memory consumption. In this paper, we propose to automatically search the network architecture tailoring to volumetric medical image segmentation problem. Concretely, we formulate the structure learning as differentiable neural architecture search, and let the network itself choose between 2D, 3D or Pseudo-3D (P3D) convolutions at each layer. We evaluate our method on 3 public datasets, i.e., the NIH Pancreas dataset, the Lung and Pancreas dataset from the Medical Segmentation Decathlon (MSD) Challenge. Our method, named V-NAS, consistently outperforms other state-of-the-arts on the segmentation tasks of both normal organ (NIH Pancreas) and abnormal organs (MSD Lung tumors and MSD Pancreas tumors), which shows the power of chosen architecture. Moreover, the searched architecture on one dataset can be well generalized to other datasets, which demonstrates the robustness and practical use of our proposed method.

It is usually hard for a learning system to predict correctly on rare events that never occur in the training data, and there is no exception for segmentation algorithms. Meanwhile, manual inspection of each case to locate the failures becomes infeasible due to the trend of large data scale and limited human resource. Therefore, we build an alarm system that will set off alerts when the segmentation result is possibly unsatisfactory, assuming no corresponding ground truth mask is provided. One plausible solution is to project the segmentation results into a low dimensional feature space; then learn classifiers/regressors to predict their qualities. Motivated by this, in this paper, we learn a feature space using the shape information which is a strong prior shared among different datasets and robust to the appearance variation of input data. The shape feature is captured using a Variational Auto-Encoder (VAE) network that trained with only the ground truth masks. During testing, the segmentation results with bad shapes shall not fit the shape prior well, resulting in large loss values. Thus, the VAE is able to evaluate the quality of segmentation result on unseen data, without using ground truth. Finally, we learn a regressor in the one-dimensional feature space to predict the qualities of segmentation results. Our alarm system is evaluated on several recent state-of-art segmentation algorithms for 3D medical segmentation tasks. Compared with other standard quality assessment methods, our system consistently provides more reliable prediction on the qualities of segmentation results.

As a minimally invasive surgery to treat atrial fibrillation (AF), catheter based ablation uses high radio-frequency energy to eliminate potential sources of abnormal electrical events, especially around the ostia of pulmonary veins (PV). Fusing a patient-specific left atrium (LA) model (including LA chamber, appendage, and PVs) with electro-anatomical maps or overlaying the model onto 2-D real-time fluoroscopic images provides valuable visual guidance during the intervention. In this work, we present a fully automatic LA segmentation system on nongated C-arm computed tomography (C-arm CT) data, where thin boundaries between the LA and surrounding tissues are often blurred due to the cardiac motion artifacts. To avoid segmentation leakage, the shape prior should be exploited to guide the segmentation. A single holistic shape model is often not accurate enough to represent the whole LA shape population under anatomical variations, e.g., the left common PVs vs. separate left PVs. Instead, a part based LA model is proposed, which includes the chamber, appendage, four major PVs, and right middle PVs. Each part is a much simpler anatomical structure compared to the holistic one and can be segmented using a model-based approach (except the right middle PVs). After segmenting the LA parts, the gaps and overlaps among the parts are resolved and segmentation of the ostia region is further refined. As a common anatomical variation, some patients may contain extra right middle PVs, which are segmented using a graph cuts algorithm under the constraints from the already extracted major right PVs. Our approach is computationally efficient, taking about 2.6 s to process a volume with 256 × 256 × 245 voxels. Experiments on 687 C-arm CT datasets demonstrate its robustness and state-of-the-art segmentation accuracy.