Multi-Source Domain Adaptation for Medical Image Segmentation.

Unsupervised domain adaptation(UDA) aims to mitigate the performance drop of models tested on the target domain, due to the domain shift from the target to sources. Most UDA segmentation methods focus on the scenario of solely single source domain. However, in practical situations data with gold standard could be available from multiple sources (domains), and the multi-source training data could provide more information for knowledge transfer. How to utilize them to achieve better domain adaptation yet remains to be further explored. This work investigates multi-source UDA and proposes a new framework for medical image segmentation. Firstly, we employ a multi-level adversarial learning scheme to adapt features at different levels between each of the source domains and the target, to improve the segmentation performance. Then, we propose a multi-model consistency loss to transfer the learned multi-source knowledge to the target domain simultaneously. Finally, we validated the proposed framework on two applications, i.e., multi-modality cardiac segmentation and cross-modality liver segmentation. The results showed our method delivered promising performance and compared favorably to state-of-the-art approaches.

Multi-site, Multi-domain Airway Tree Modeling.

Open international challenges are becoming the de facto standard for assessing computer vision and image analysis algorithms. In recent years, new methods have extended the reach of pulmonary airway segmentation that is closer to the limit of image resolution. Since EXACT'09 pulmonary airway segmentation, limited effort has been directed to the quantitative comparison of newly emerged algorithms driven by the maturity of deep learning based approaches and extensive clinical efforts for resolving finer details of distal airways for early intervention of pulmonary diseases. Thus far, public annotated datasets are extremely limited, hindering the development of data-driven methods and detailed performance evaluation of new algorithms. To provide a benchmark for the medical imaging community, we organized the Multi-site, Multi-domain Airway Tree Modeling (ATM'22), which was held as an official challenge event during the MICCAI 2022 conference. ATM'22 provides large-scale CT scans with detailed pulmonary airway annotation, including 500 CT scans (300 for training, 50 for validation, and 150 for testing). The dataset was collected from different sites and it further included a portion of noisy COVID-19 CTs with ground-glass opacity and consolidation. Twenty-three teams participated in the entire phase of the challenge and the algorithms for the top ten teams are reviewed in this paper. Both quantitative and qualitative results revealed that deep learning models embedded with the topological continuity enhancement achieved superior performance in general. ATM'22 challenge holds as an open-call design, the training data and the gold standard evaluation are available upon successful registration via its homepage (https://atm22.grand-challenge.org/).

Distribution of cracks in an anchored cavern under blast load based on cohesive elements.

To explore the distribution of cracks in anchored caverns under the blast load, cohesive elements with zero thickness were employed to simulate crack propagation through numerical analysis based on a similar model test. Furthermore, the crack propagation process in anchored caverns under top explosion was analyzed. The crack propagation modes and distributions in anchored caverns with different dip angles fractures in the vault were thoroughly discussed. With the propagation of the explosive stress waves, cracks successively occur at the arch foot, the floor of the anchored caverns, and the boundary of the anchored zone of the vault. Tensile cracks are preliminarily found in rocks that surround the caverns. In the scenario of a pre-fabricated fracture in the upper part of the vault, the number of cracks at the boundary of the anchored zone of the vault first decreases then increases with the increasing dip angle of the pre-fabricated fracture. When the dip angle of the pre-fabricated fracture is 45°, the fewest cracks occur at the boundary of the anchored zone. The wing cracks deflected to the vault are formed at the tip of the pre-fabricated fracture, around which are synchronous formed tensile and shear cracks. Under top explosion, the peak displacement and the peak particle velocity in surrounding rocks of anchored caverns both reach their maximum values at the vault, successively followed by the sidewall and the floor. In addition, with the different dip angles of the pre-fabricated fracture, asymmetry could be found between the peak displacement and the peak particle velocity. The vault displacement of anchored caverns is mainly attributed to tensile cracks at the boundary of the anchored zone, which are generated due to the tensile waves reflected from the free face of the vault. When a fracture occurs in the vault, the peak displacement of the vault gradually decreases while the residual displacement increases.

Interpretative computer-aided lung cancer diagnosis: From radiology analysis to malignancy evaluation.

BACKGROUND AND OBJECTIVE: Computer-aided diagnosis (CAD) systems promote accurate diagnosis and reduce the burden of radiologists. A CAD system for lung cancer diagnosis includes nodule candidate detection and nodule malignancy evaluation. Recently, deep learning-based pulmonary nodule detection has reached satisfactory performance ready for clinical application. However, deep learning-based nodule malignancy evaluation depends on heuristic inference from low-dose computed tomography (LDCT) volume to malignant probability, and lacks clinical cognition. METHODS: In this paper, we propose a joint radiology analysis and malignancy evaluation network called R2MNet to evaluate pulmonary nodule malignancy via the analysis of radiological characteristics. Radiological features are extracted as channel descriptor to highlight specific regions of the input volume that are critical for nodule malignancy evaluation. In addition, for model explanations, we propose channel-dependent activation mapping (CDAM) to visualize features and shed light on the decision process of deep neural networks (DNNs). RESULTS: Experimental results on the lung image database consortium image collection (LIDC-IDRI) dataset demonstrate that the proposed method achieved an area under curve (AUC) of 96.27% and 97.52% on nodule radiology analysis and nodule malignancy evaluation, respectively. In addition, explanations of CDAM features proved that the shape and density of nodule regions are two critical factors that influence a nodule to be inferred as malignant. This process conforms to the diagnosis cognition of experienced radiologists. CONCLUSION: The network inference process conforms to the diagnostic procedure of radiologists and increases the confidence of evaluation results by incorporating radiology analysis with nodule malignancy evaluation. Besides, model interpretation with CDAM features shed light on the focus regions of DNNs during the estimation of nodule malignancy probabilities.

Disentangle domain features for cross-modality cardiac image segmentation.

Unsupervised domain adaptation (UDA) generally learns a mapping to align the distribution of the source domain and target domain. The learned mapping can boost the performance of the model on the target data, of which the labels are unavailable for model training. Previous UDA methods mainly focus on domain-invariant features (DIFs) without considering the domain-specific features (DSFs), which could be used as complementary information to constrain the model. In this work, we propose a new UDA framework for cross-modality image segmentation. The framework first disentangles each domain into the DIFs and DSFs. To enhance the representation of DIFs, self-attention modules are used in the encoder which allows attention-driven, long-range dependency modeling for image generation tasks. Furthermore, a zero loss is minimized to enforce the information of target (source) DSFs, contained in the source (target) images, to be as close to zero as possible. These features are then iteratively decoded and encoded twice to maintain the consistency of the anatomical structure. To improve the quality of the generated images and segmentation results, several discriminators are introduced for adversarial learning. Finally, with the source data and their DIFs, we train a segmentation network, which can be applicable to target images. We validated the proposed framework for cross-modality cardiac segmentation using two public datasets, and the results showed our method delivered promising performance and compared favorably to state-of-the-art approaches in terms of segmentation accuracies. The source code of this work will be released via https://zmiclab.github.io/projects.html, once this manuscript is accepted for publication.