Medical image registration is a crucial process for aligning anatomical structures, enabling applications such as atlas mapping, longitudinal analysis, and multimodal data fusion. This paper introduces DINO-Reg, an adaptation-free registration method leveraging the vision foundation model, DINOv2, to extract features for deformable 3D medical image alignment. Although DINOv2 was originally trained on natural images, our study links the vision foundation model with medical image registration and demonstrates that the generic image encoder could readily generalize to medical images with state-of-the-art performance. We further propose DINO-Reg-Eco, a knowledge-distilled version using a UNet-structured 3D convolutional neural network (CNN) for feature extraction. The Eco model reduces encoding time by 99% while maintaining state-of-the-art performance, which is essential for resource-limited settings and significantly lowers the carbon footprint associated with intensive computational demands. Benchmarking across diverse datasets shows that both methods outperform existing supervised and unsupervised approaches without fine-tuning, demonstrating the transformative potential of foundation models in medical image registration. Our code is open-sourced at https://github.com/RPIDIAL/DINO-Reg.

Chest X-ray (CXR) is the most frequently ordered imaging test, supporting diverse clinical tasks from thoracic disease detection to postoperative monitoring. However, task-specific classification models are limited in scope, require costly labeled data, and lack generalizability to out-of-distribution datasets. To address these challenges, we introduce CheXFound, a self-supervised vision foundation model that learns robust CXR representations and generalizes effectively across a wide range of downstream tasks. We pretrained CheXFound on a curated CXR-987K dataset, comprising over approximately 987K unique CXRs from 12 publicly available sources. We propose a Global and Local Representations Integration (GLoRI) head for downstream adaptations, by incorporating fine- and coarse-grained disease-specific local features with global image features for enhanced performance in multilabel classification. Our experimental results showed that CheXFound outperformed state-of-the-art models in classifying 40 disease findings across different prevalence levels on the CXR-LT 24 dataset and exhibited superior label efficiency on downstream tasks with limited training data. Additionally, CheXFound achieved significant improvements on downstream tasks with out-of-distribution datasets, including opportunistic cardiovascular disease risk estimation, mortality prediction, malpositioned tube detection, and anatomical structure segmentation. The above results demonstrate CheXFound's strong generalization capabilities, which will enable diverse downstream adaptations with improved label efficiency in future applications. The project source code is publicly available at https://github.com/RPIDIAL/CheXFound.

Adjuvant and salvage radiotherapy after radical prostatectomy requires precise delineations of prostate bed (PB), i.e., the clinical target volume, and surrounding organs at risk (OARs) to optimize radiotherapy planning. Segmenting PB is particularly challenging even for clinicians, e.g., from the planning computed tomography (CT) images, as it is an invisible/virtual target after the operative removal of the cancerous prostate gland. Very recently, a few deep learning-based methods have been proposed to automatically contour non-contrast PB by leveraging its spatial reliance on adjacent OARs (i.e., the bladder and rectum) with much more clear boundaries, mimicking the clinical workflow of experienced clinicians. Although achieving state-of-the-art results from both the clinical and technical aspects, these existing methods improperly ignore the gap between the hierarchical feature representations needed for segmenting those fundamentally different clinical targets (i.e., PB and OARs), which in turn limits their delineation accuracy. This paper proposes an asymmetric multi-task network integrating dynamic cross-task representation adaptation (i.e., DyAdapt) for accurate and efficient co-segmentation of PB and OARs in one-pass from CT images. In the learning-to-learn framework, the DyAdapt modules adaptively transfer the hierarchical feature representations from the source task of OARs segmentation to match up with the target (and more challenging) task of PB segmentation, conditioned on the dynamic inter-task associations learned from the learning states of the feed-forward path. On a real-patient dataset, our method led to state-of-the-art results of PB and OARs co-segmentation. Code is available at https://github.com/ladderlab-xjtu/DyAdapt.

Federated learning is an emerging paradigm allowing large-scale decentralized learning without sharing data across different data owners, which helps address the concern of data privacy in medical image analysis. However, the requirement for label consistency across clients by the existing methods largely narrows its application scope. In practice, each clinical site may only annotate certain organs of interest with partial or no overlap with other sites. Incorporating such partially labeled data into a unified federation is an unexplored problem with clinical significance and urgency. This work tackles the challenge by using a novel federated multi-encoding U-Net (Fed-MENU) method for multi-organ segmentation. In our method, a multi-encoding U-Net (MENU-Net) is proposed to extract organ-specific features through different encoding sub-networks. Each sub-network can be seen as an expert of a specific organ and trained for that client. Moreover, to encourage the organ-specific features extracted by different sub-networks to be informative and distinctive, we regularize the training of the MENU-Net by designing an auxiliary generic decoder (AGD). Extensive experiments on six public abdominal CT datasets show that our Fed-MENU method can effectively obtain a federated learning model using the partially labeled datasets with superior performance to other models trained by either localized or centralized learning methods. Source code is publicly available at https://github.com/DIAL-RPI/Fed-MENU.

Prostate segmentation in transrectal ultrasound (TRUS) image is an essential prerequisite for many prostate-related clinical procedures, which, however, is also a long-standing problem due to the challenges caused by the low image quality and shadow artifacts. In this paper, we propose a Shadow-consistent Semi-supervised Learning (SCO-SSL) method with two novel mechanisms, namely shadow augmentation (Shadow-AUG) and shadow dropout (Shadow-DROP), to tackle this challenging problem. Specifically, Shadow-AUG enriches training samples by adding simulated shadow artifacts to the images to make the network robust to the shadow patterns. Shadow-DROP enforces the segmentation network to infer the prostate boundary using the neighboring shadow-free pixels. Extensive experiments are conducted on two large clinical datasets (a public dataset containing 1,761 TRUS volumes and an in-house dataset containing 662 TRUS volumes). In the fully-supervised setting, a vanilla U-Net equipped with our Shadow-AUG&Shadow-DROP outperforms the state-of-the-arts with statistical significance. In the semi-supervised setting, even with only 20% labeled training data, our SCO-SSL method still achieves highly competitive performance, suggesting great clinical value in relieving the labor of data annotation. Source code is released at https://github.com/DIAL-RPI/SCO-SSL.

Organ localization is an essential preprocessing step for many medical image analysis tasks such as image registration, organ segmentation and lesion detection. In this work, we propose an efficient method for multiple organ localization in CT image using 3D region proposal network. Compared with other convolutional neural network based methods that successively detect the target organs in all slices to assemble the final 3D bounding box, our method is fully implemented in 3D manner, thus can take full advantages of the spatial context information in CT image to perform efficient organ localization with only one prediction. We also propose a novel backbone network architecture that generates high-resolution feature maps to further improve the localization performance on small organs. We evaluate our method on two clinical datasets, where 11 body organs and 12 head organs (or anatomical structures) are included. As our results shown, the proposed method achieves higher detection precision and localization accuracy than the current state-of-theart methods with approximate 4 to 18 times faster processing speed. Additionally, we have established a public dataset dedicated for organ localization on http://dx. doi.org/10.21227/df8g-pq27. The full implementation of the proposed method have also been made publicly available on https://github.com/superxuang/caffe_3d_faster_rcnn.