Medical image analysis最新文献

{"title":"Image by co-reasoning: A collaborative reasoning-based implicit data augmentation method for dual-view CEUS classification","authors":"Peng Wan ,&nbsp;Haiyan Xue ,&nbsp;Shukang Zhang ,&nbsp;Wentao Kong ,&nbsp;Wei Shao ,&nbsp;Baojie Wen ,&nbsp;Daoqiang Zhang","doi":"10.1016/j.media.2025.103557","DOIUrl":"10.1016/j.media.2025.103557","url":null,"abstract":"<div><div>Dual-view contrast-enhanced ultrasound (CEUS) data are often insufficient to train reliable machine learning models in typical clinical scenarios. A key issue is that limited clinical CEUS data fail to cover the underlying texture variations for specific diseases. Implicit data augmentation offers a flexible way to enrich sample diversity, however, inter-view semantic consistency has not been considered in previous studies. To address this issue, we propose a novel implicit data augmentation method for dual-view CEUS classification, which performs a sample-adaptive data augmentation with collaborative semantic reasoning across views. Specifically, the method constructs a feature augmentation distribution for each ultrasound view of an individual sample, accounting for intra-class variance. To maintain semantic consistency between the augmented views, plausible semantic changes in one view are transferred from similar instances in the other view. In this retrospective study, we validate the proposed method on the dual-view CEUS datasets of breast cancer and liver cancer, obtaining the superior mean diagnostic accuracy of 89.25% and 95.57%, respectively. Experimental results demonstrate its effectiveness in improving model performance with limited clinical CEUS data. Code: <span><span>https://github.com/wanpeng16/CRIDA</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":18328,"journal":{"name":"Medical image analysis","volume":"102 ","pages":"Article 103557"},"PeriodicalIF":10.7,"publicationDate":"2025-03-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143748548","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"CLIP in medical imaging: A survey","authors":"Zihao Zhao ,&nbsp;Yuxiao Liu ,&nbsp;Han Wu ,&nbsp;Mei Wang ,&nbsp;Yonghao Li ,&nbsp;Sheng Wang ,&nbsp;Lin Teng ,&nbsp;Disheng Liu ,&nbsp;Zhiming Cui ,&nbsp;Qian Wang ,&nbsp;Dinggang Shen","doi":"10.1016/j.media.2025.103551","DOIUrl":"10.1016/j.media.2025.103551","url":null,"abstract":"<div><div>Contrastive Language-Image Pre-training (CLIP), a simple yet effective pre-training paradigm, successfully introduces text supervision to vision models. It has shown promising results across various tasks due to its generalizability and interpretability. The use of CLIP has recently gained increasing interest in the medical imaging domain, serving as a pre-training paradigm for image–text alignment, or a critical component in diverse clinical tasks. With the aim of facilitating a deeper understanding of this promising direction, this survey offers an in-depth exploration of the CLIP within the domain of medical imaging, regarding both refined CLIP pre-training and CLIP-driven applications. In this paper, we (1) first start with a brief introduction to the fundamentals of CLIP methodology; (2) then investigate the adaptation of CLIP pre-training in the medical imaging domain, focusing on how to optimize CLIP given characteristics of medical images and reports; (3) further explore practical utilization of CLIP pre-trained models in various tasks, including classification, dense prediction, and cross-modal tasks; and (4) finally discuss existing limitations of CLIP in the context of medical imaging, and propose forward-looking directions to address the demands of medical imaging domain. Studies featuring technical and practical value are both investigated. We expect this survey will provide researchers with a holistic understanding of the CLIP paradigm and its potential implications. The project page of this survey can also be found on <span><span>Github</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":18328,"journal":{"name":"Medical image analysis","volume":"102 ","pages":"Article 103551"},"PeriodicalIF":10.7,"publicationDate":"2025-03-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143675536","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Enhancing source-free domain adaptation in Medical Image Segmentation via regulated model self-training","authors":"Tianwei Zhang ,&nbsp;Kang Li ,&nbsp;Shi Gu ,&nbsp;Pheng-Ann Heng","doi":"10.1016/j.media.2025.103543","DOIUrl":"10.1016/j.media.2025.103543","url":null,"abstract":"<div><div>Source-free domain adaptation (SFDA) has drawn increasing attention lately in the medical field. It aims to adapt a model well trained on source domain to target domains without accessing source domain data nor requiring target domain labels, to enable privacy-protecting and annotation-efficient domain adaptation. Most SFDA approaches initialize the target model with source model weights, and guide model self-training with the pseudo-labels generated from the source model. However, when source and target domains have huge discrepancies (<em>e.g.</em>, different modalities), the obtained pseudo-labels would be of poor quality. Different from prior works that overcome it by refining pseudo-labels to better quality, in this work, we try to explore it from the perspective of knowledge transfer. We recycle the beneficial domain-invariant prior knowledge in the source model, and refresh its domain-specific knowledge from source-specific to target-specific, to help the model satisfyingly tackle target domains even when facing severe domain shifts. To achieve it, we proposed a regulated model self-training framework. For high-transferable domain-invariant parameters, we constrain their update magnitude from large changes, to secure the domain-shared priors from going stray and let it continuously facilitate target domain adaptation. For the low-transferable domain-specific parameters, we actively update them to let the domain-specific embedding become target-specific. Regulating them together, the model would develop better capability for target data even under severe domain shifts. Importantly, the proposed approach could seamlessly collaborate with existing pseudo-label refinement approaches to bring more performance gains. We have extensively validated our framework under significant domain shifts in 3D cross-modality cardiac segmentation, and under minor domain shifts in 2D cross-vendor fundus segmentation, respectively. Our approach consistently outperformed the competing methods and achieved superior performance.</div></div>","PeriodicalId":18328,"journal":{"name":"Medical image analysis","volume":"102 ","pages":"Article 103543"},"PeriodicalIF":10.7,"publicationDate":"2025-03-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143714209","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Revisiting medical image retrieval via knowledge consolidation","authors":"Yang Nan ,&nbsp;Huichi Zhou ,&nbsp;Xiaodan Xing ,&nbsp;Giorgos Papanastasiou ,&nbsp;Lei Zhu ,&nbsp;Zhifan Gao ,&nbsp;Alejandro F. Frangi ,&nbsp;Guang Yang","doi":"10.1016/j.media.2025.103553","DOIUrl":"10.1016/j.media.2025.103553","url":null,"abstract":"<div><div>As artificial intelligence and digital medicine increasingly permeate healthcare systems, robust governance frameworks are essential to ensure ethical, secure, and effective implementation. In this context, medical image retrieval becomes a critical component of clinical data management, playing a vital role in decision-making and safeguarding patient information. Existing methods usually learn hash functions using bottleneck features, which fail to produce representative hash codes from blended embeddings. Although contrastive hashing has shown superior performance, current approaches often treat image retrieval as a classification task, using category labels to create positive/negative pairs. Moreover, many methods fail to address the out-of-distribution (OOD) issue when models encounter external OOD queries or adversarial attacks. In this work, we propose a novel method to consolidate knowledge of hierarchical features and optimization functions. We formulate the knowledge consolidation by introducing Depth-aware Representation Fusion (DaRF) and Structure-aware Contrastive Hashing (SCH). DaRF adaptively integrates shallow and deep representations into blended features, and SCH incorporates image fingerprints to enhance the adaptability of positive/negative pairings. These blended features further facilitate OOD detection and content-based recommendation, contributing to a secure AI-driven healthcare environment. Moreover, we present a content-guided ranking to improve the robustness and reproducibility of retrieval results. Our comprehensive assessments demonstrate that the proposed method could effectively recognize OOD samples and significantly outperform existing approaches in medical image retrieval (p<span><math><mrow><mi>&lt;</mi><mn>0</mn><mo>.</mo><mn>05</mn></mrow></math></span>). In particular, our method achieves a 5.6–38.9% improvement in mean Average Precision on the anatomical radiology dataset.</div></div>","PeriodicalId":18328,"journal":{"name":"Medical image analysis","volume":"102 ","pages":"Article 103553"},"PeriodicalIF":10.7,"publicationDate":"2025-03-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143675549","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"MMR-Mamba: Multi-modal MRI reconstruction with Mamba and spatial-frequency information fusion","authors":"Jing Zou ,&nbsp;Lanqing Liu ,&nbsp;Qi Chen ,&nbsp;Shujun Wang ,&nbsp;Zhanli Hu ,&nbsp;Xiaohan Xing ,&nbsp;Jing Qin","doi":"10.1016/j.media.2025.103549","DOIUrl":"10.1016/j.media.2025.103549","url":null,"abstract":"<div><div>Multi-modal MRI offers valuable complementary information for diagnosis and treatment; however, its clinical utility is limited by prolonged scanning time. To accelerate the acquisition process, a practical approach is to reconstruct images of the target modality, which requires longer scanning time, from under-sampled k-space data using the fully-sampled reference modality with shorter scanning time as guidance. The primary challenge of this task lies in comprehensively and efficiently integrating complementary information from different modalities to achieve high-quality reconstruction. Existing methods struggle with this challenge: (1) convolution-based models fail to capture long-range dependencies; (2) transformer-based models, while excelling in global feature modeling, suffer from quadratic computational complexity. To address this dilemma, we propose MMR-Mamba, a novel framework that thoroughly and efficiently integrates multi-modal features for MRI reconstruction, leveraging Mamba’s capability to capture long-range dependencies with linear computational complexity while exploiting global properties of the Fourier domain. Specifically, we first design a <em>Target modality-guided Cross Mamba</em> (TCM) module in the spatial domain, which maximally restores the target modality information by selectively incorporating relevant information from the reference modality. Then, we introduce a <em>Selective Frequency Fusion</em> (SFF) module to efficiently integrate global information in the Fourier domain and recover high-frequency signals for the reconstruction of structural details. Furthermore, we devise an <em>Adaptive Spatial-Frequency Fusion</em> (ASFF) module, which mutually enhances the spatial and frequency domains by supplementing less informative channels from one domain with corresponding channels from the other. Extensive experiments on the BraTS and fastMRI knee datasets demonstrate the superiority of our MMR-Mamba over state-of-the-art reconstruction methods. The code is publicly available at <span><span>https://github.com/zoujing925/MMR-Mamba</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":18328,"journal":{"name":"Medical image analysis","volume":"102 ","pages":"Article 103549"},"PeriodicalIF":10.7,"publicationDate":"2025-03-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143675569","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Flip Learning: Weakly supervised erase to segment nodules in breast ultrasound","authors":"Yuhao Huang ,&nbsp;Ao Chang ,&nbsp;Haoran Dou ,&nbsp;Xing Tao ,&nbsp;Xinrui Zhou ,&nbsp;Yan Cao ,&nbsp;Ruobing Huang ,&nbsp;Alejandro F. Frangi ,&nbsp;Lingyun Bao ,&nbsp;Xin Yang ,&nbsp;Dong Ni","doi":"10.1016/j.media.2025.103552","DOIUrl":"10.1016/j.media.2025.103552","url":null,"abstract":"<div><div>Accurate segmentation of nodules in both 2D breast ultrasound (BUS) and 3D automated breast ultrasound (ABUS) is crucial for clinical diagnosis and treatment planning. Therefore, developing an automated system for nodule segmentation can enhance user independence and expedite clinical analysis. Unlike fully-supervised learning, weakly-supervised segmentation (WSS) can streamline the laborious and intricate annotation process. However, current WSS methods face challenges in achieving precise nodule segmentation, as many of them depend on inaccurate activation maps or inefficient pseudo-mask generation algorithms. In this study, we introduce a novel multi-agent reinforcement learning-based WSS framework called Flip Learning, which relies solely on 2D/3D boxes for accurate segmentation. Specifically, multiple agents are employed to erase the target from the box to facilitate classification tag flipping, with the erased region serving as the predicted segmentation mask. The key contributions of this research are as follows: (1) Adoption of a superpixel/supervoxel-based approach to encode the standardized environment, capturing boundary priors and expediting the learning process. (2) Introduction of three meticulously designed rewards, comprising a classification score reward and two intensity distribution rewards, to steer the agents’ erasing process precisely, thereby avoiding both under- and over-segmentation. (3) Implementation of a progressive curriculum learning strategy to enable agents to interact with the environment in a progressively challenging manner, thereby enhancing learning efficiency. Extensively validated on the large in-house BUS and ABUS datasets, our Flip Learning method outperforms state-of-the-art WSS methods and foundation models, and achieves comparable performance as fully-supervised learning algorithms.</div></div>","PeriodicalId":18328,"journal":{"name":"Medical image analysis","volume":"102 ","pages":"Article 103552"},"PeriodicalIF":10.7,"publicationDate":"2025-03-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143748644","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Automated planning of mandible reconstruction with fibula free flap based on shape completion and morphometric descriptors","authors":"Yan Guo ,&nbsp;Chenyao Li ,&nbsp;Rong Yang ,&nbsp;Puxun Tu ,&nbsp;Bolun Zeng ,&nbsp;Jiannan Liu ,&nbsp;Tong Ji ,&nbsp;Chenping Zhang ,&nbsp;Xiaojun Chen","doi":"10.1016/j.media.2025.103544","DOIUrl":"10.1016/j.media.2025.103544","url":null,"abstract":"<div><div>Vascularized fibula free flap (FFF) grafts are frequently used to reconstruct mandibular defects. However, the current planning methods for osteotomy, splicing, and fibula placement present challenges in achieving satisfactory facial aesthetics and restoring the original morphology of the mandible. In this study, we propose a novel two-step framework for automated preoperative planning in FFF mandibular reconstruction. The framework is based on mandibular shape completion and morphometric descriptors. Firstly, we utilize a 3D generative model to estimate the entire mandibular geometry by incorporating shape priors and accounting for partial defect mandibles. Accurately predicting the premorbid morphology of the mandible is crucial for determining the surgical plan. Secondly, we introduce new two-dimensional morphometric descriptors to assess the quantitative difference between the planning scheme and the full morphology of the mandible. We have designed intuitive and valid variables specifically designed to describe the planning scheme and constructed an objective function to measure the difference. By optimizing this function, we can achieve the best shape-matched 3D planning solution. Through a retrospective study involving 65 real tumor patients, our method has exhibited favorable results in both qualitative and quantitative analyses when compared to the planned results of experienced clinicians using existing methods. This demonstrates that our method can implement an automated preoperative planning technique, eliminating subjectivity and achieving user-independent results. Furthermore, we have presented the potential of our automated planning process in a clinical case, highlighting its applicability in clinical settings.</div></div>","PeriodicalId":18328,"journal":{"name":"Medical image analysis","volume":"102 ","pages":"Article 103544"},"PeriodicalIF":10.7,"publicationDate":"2025-03-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143685374","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"An extragradient and noise-tuning adaptive iterative network for diffusion MRI-based microstructural estimation","authors":"Tianshu Zheng ,&nbsp;Chuyang Ye ,&nbsp;Zhaopeng Cui ,&nbsp;Hui Zhang ,&nbsp;Daniel C. Alexander ,&nbsp;Dan Wu","doi":"10.1016/j.media.2025.103535","DOIUrl":"10.1016/j.media.2025.103535","url":null,"abstract":"<div><div>Diffusion MRI (dMRI) is a powerful technique for investigating tissue microstructure properties. However, advanced dMRI models are typically complex and nonlinear, requiring a large number of acquisitions in the <em>q</em>-space. Deep learning techniques, specifically optimization-based networks, have been proposed to improve the model fitting with limited <em>q</em>-space data. Previous optimization procedures relied on the empirical selection of iteration block numbers and the network structures were based on the <em>iterative hard thresholding</em> (IHT) algorithm, which may suffer from instability during sparse reconstruction. In this study, we introduced an <em>extragradient and noise-tuning adaptive iterative network</em>, a generic network for estimating dMRI model parameters. We proposed an adaptive mechanism that flexibly adjusts the sparse representation process, depending on specific dMRI models, datasets, and downsampling strategies, avoiding manual selection and accelerating inference. In addition, we proposed a noise-tuning module to assist the network in escaping from local minimum/saddle points. The network also included an additional projection of the extragradient to ensure its convergence. We evaluated the performance of the proposed network on the <em>neurite orientation dispersion and density imaging</em> (NODDI) model and <em>diffusion basis spectrum imaging</em> (DBSI) model on two 3T <em>Human Connectome Project</em> (HCP) datasets and a 7T HCP dataset with six different downsampling strategies. The proposed framework demonstrated superior accuracy and generalizability compared to other state-of-the-art microstructural estimation algorithms.</div></div>","PeriodicalId":18328,"journal":{"name":"Medical image analysis","volume":"102 ","pages":"Article 103535"},"PeriodicalIF":10.7,"publicationDate":"2025-03-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143714210","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"5D image reconstruction exploiting space-motion-echo sparsity for accelerated free-breathing quantitative liver MRI","authors":"MungSoo Kang ,&nbsp;Ricardo Otazo ,&nbsp;Gerald Behr ,&nbsp;Youngwook Kee","doi":"10.1016/j.media.2025.103532","DOIUrl":"10.1016/j.media.2025.103532","url":null,"abstract":"<div><div>Recent advances in 3D non-Cartesian multi-echo gradient-echo (mGRE) imaging and compressed sensing (CS)-based 4D (3D image space + 1D respiratory motion) motion-resolved image reconstruction, which applies temporal total variation to the respiratory motion dimension, have enabled free-breathing liver tissue MR parameter mapping. This technology now allows for robust reconstruction of high-resolution proton density fat fraction (PDFF), R<span><math><msubsup><mrow></mrow><mrow><mn>2</mn></mrow><mrow><mo>∗</mo></mrow></msubsup></math></span>, and quantitative susceptibility mapping (QSM), previously unattainable with conventional Cartesian mGRE imaging. However, long scan times remain a persistent challenge in free-breathing 3D non-Cartesian mGRE imaging. Recognizing that the underlying dimension of the imaging data is essentially 5D (4D + 1D echo signal evolution), we propose a CS-based 5D motion-resolved mGRE image reconstruction method to further accelerate the acquisition. Our approach integrates discrete wavelet transforms along the echo and spatial dimensions into a CS-based reconstruction model and devises a solution algorithm capable of handling such a 5D complex-valued array. Through phantom and in vivo human subject studies, we evaluated the effectiveness of leveraging unexplored correlations by comparing the proposed 5D reconstruction with the 4D reconstruction (i.e., motion-resolved reconstruction with temporal total variation) across a wide range of acceleration factors. The 5D reconstruction produced more reliable and consistent measurements of PDFF, R<span><math><msubsup><mrow></mrow><mrow><mn>2</mn></mrow><mrow><mo>∗</mo></mrow></msubsup></math></span>, and QSM compared to the 4D reconstruction. In conclusion, the proposed 5D motion-resolved image reconstruction demonstrates the feasibility of achieving accelerated, reliable, and free-breathing liver mGRE imaging for the measurement of PDFF, R<span><math><msubsup><mrow></mrow><mrow><mn>2</mn></mrow><mrow><mo>∗</mo></mrow></msubsup></math></span>, and QSM.</div></div>","PeriodicalId":18328,"journal":{"name":"Medical image analysis","volume":"102 ","pages":"Article 103532"},"PeriodicalIF":10.7,"publicationDate":"2025-03-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143675551","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Medical SAM adapter: Adapting segment anything model for medical image segmentation","authors":"Junde Wu ,&nbsp;Ziyue Wang ,&nbsp;Mingxuan Hong ,&nbsp;Wei Ji ,&nbsp;Huazhu Fu ,&nbsp;Yanwu Xu ,&nbsp;Min Xu ,&nbsp;Yueming Jin","doi":"10.1016/j.media.2025.103547","DOIUrl":"10.1016/j.media.2025.103547","url":null,"abstract":"<div><div>The Segment Anything Model (SAM) has recently gained popularity in the field of image segmentation due to its impressive capabilities in various segmentation tasks and its prompt-based interface. However, recent studies and individual experiments have shown that SAM underperforms in medical image segmentation due to the lack of medical-specific knowledge. This raises the question of how to enhance SAM’s segmentation capability for medical images. We propose the Medical SAM Adapter (Med-SA), which is one of the first methods to integrate SAM into medical image segmentation. Med-SA uses a light yet effective adaptation technique instead of fine-tuning the SAM model, incorporating domain-specific medical knowledge into the segmentation model. We also propose Space-Depth Transpose (SD-Trans) to adapt 2D SAM to 3D medical images and Hyper-Prompting Adapter (HyP-Adpt) to achieve prompt-conditioned adaptation. Comprehensive evaluation experiments on 17 medical image segmentation tasks across various modalities demonstrate the superior performance of Med-SA while updating only 2% of the SAM parameters (13M). Our code is released at <span><span>https://github.com/KidsWithTokens/Medical-SAM-Adapter</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":18328,"journal":{"name":"Medical image analysis","volume":"102 ","pages":"Article 103547"},"PeriodicalIF":10.7,"publicationDate":"2025-03-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143675535","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}