The Segment Anything Model and its variants, such as MedSAM, have demonstrated potential for medical image segmentation. However, they heavily rely on high-quality manual prompts, which are both time-consuming and require domain expertise. Even when using manual prompts, including sparse prompts like boxes, points, or text, and dense prompts such as masks, SAM and its variants like MedSAM (fine-tuned on medical images) fail to segment bones in ultrasound images due to significant domain shift. To address these limitations, we propose Sam2 for Radiology (Sam2Rad), a framework that extends SAM and its recent iteration SAM2 to segment bony regions in ultrasound images without requiring manual prompts. At the core of our approach is a Prompt Predictor Network (PPN) that uses a lightweight cross-attention mechanism to generate bounding box coordinates, mask prompts, and high-dimensional embeddings to be used as prompts. Specifically, PPN leverages hierarchical feature maps extracted from SAM's image encoder as keys and values, and learnable object embeddings as queries. The output of the cross-attention is then used to predict the bounding box coordinates, mask prompts, and high-dimensional prompts. These predicted prompts are subsequently fed to SAM's mask decoder to generate the final segmentation mask. By aligning the learned prompts with SAM's original training scheme, PPN enhances compatibility with SAM's architecture, requiring no additional standalone encoders. To preserve SAM's extensive world knowledge, we keep all SAM modules frozen and train PPN only. This approach enables efficient parameter utilization while retaining SAM's generalization capabilities. Additionally, Sam2Rad can operate in three modes: fully autonomous without human supervision, semi-autonomous with human-in-the-loop for iterative refinement, and fully manual for tasks like data labeling. We tested the proposed model - Sam2Rad on 3 musculoskeletal US datasets - wrist (3822 images), shoulder rotator cuff (1605 images), and hip (4849 images). Without Sam2Rad, all SAM2 variants failed to segment shoulder US in zero-shot generalization with bounding box prompts. Our model, Sam2Rad, improved the performance of all SAM base networks in all datasets, without requiring manual prompts. The improvement in dice score ranged from a 2.2%-5.8% for hip, 19.6%-32.8% for wrist wrist, and up to 51.3% improvement in Dice score (from 25.2% to 76.5% Sam2 large) on shoulder data. Notably, Sam2Rad could be trained with as few as 10 labeled images and it is compatible with any SAM architecture. The code is available at https://github.com/aswahd/SamRadiology.

Lung ultrasound (LUS) is a non-invasive bedside imaging technique for diagnosing pulmonary conditions, especially in critical care settings. A-lines and B-lines are important features in LUS images that help to assess lung health and identify changes in lung tissue. However, accurately detecting and segmenting these lines remains challenging, due to their subtle blurred boundaries. To address this, we propose TransBound-UNet, a novel segmentation model that integrates a transformer-based encoder with boundary-aware Dice loss to enhance medical image segmentation. This loss function incorporates boundary-specific penalties into a hybrid Dice-BCE formulation, allowing for more accurate segmentation of critical structures. The proposed framework was tested on a dataset of 4599 LUS images. The model achieved a Dice Score of 0.80, outperforming state-of-the-art segmentation networks. Additionally, it demonstrated superior performance in Specificity (0.97) and Precision (0.85), with a significantly reduced Hausdorff Distance of 15.13, indicating improved boundary delineation and overall segmentation quality. Post-processing techniques were applied to automatically detect and count A-lines and B-lines, demonstrating the potential of the segmented outputs in diagnostic workflows. This framework provides an efficient solution for automated LUS interpretation, with improved boundary precision.

Machine learning using supervised approaches has been widely applied to ultrasound image analysis. In contrast, reinforcement learning (RL), which is well suited for sequential decision-making, is underexplored. Ultrasound workflows involve sequential subtasks like image acquisition, quality assessment, summarization, and interpretation that can be integrated into RL frameworks. This scoping review examined RL applications in ultrasound. A comprehensive search was conducted in Scopus, PubMed, Embase, and MEDLINE for studies published between 2015 and 2026. Eligible studies used RL with clinical ultrasound data, and the data were summarized by application area, methods, and anatomical targets. The review also provides a brief overview of RL concepts as foundational knowledge for understanding various RL formulations used. From 326 records retrieved, 39 studies were included. Most studies used model-free deep RL, with value-based methods being the most common, particularly Deep Q-Network (DQN) and its variants, on retrospective data across diverse anatomical targets, with breast, fetal, and uterine ultrasound being the most frequently represented categories. RL was used to automate tasks including navigation, plane localization, landmark detection, and video summarization. RL in ultrasound imaging is an emerging field of research and has advantages for sequential workflow optimization tasks. Most approaches are at an early stage and have been tested on small datasets, lack consistent evaluation protocols, and report limited clinical translation.

Objective: Rotator cuff tears involving the supraspinatus tendon are a common cause of shoulder pain. Ultrasound (US) is accessible, dynamic, and lower cost than MRI, but diagnostic performance may vary by tear severity, operator expertise. This systematic review and meta-analysis compared the US with MRI for supraspinatus tear detection. Methods: MEDLINE, Embase, Scopus, and CINAHL Plus were searched for studies published from January 2015 to June 2026. Study quality was assessed using QUADAS-2. Diagnostic outcomes included sensitivity, specificity, accuracy, PPV, NPV. Meta-analyses were performed separately for any, partial-thickness, and full-thickness tears when complete 2 x 2 data were available. Results: 23 studies comprising 1,452 participants were included. US demonstrated stronger and more consistent performance for full-thickness than partial-thickness tears. Mean sensitivity, specificity, and accuracy were 88.5%, 95.2%, and 93.4%, respectively, for full-thickness tears, compared with 71.5%, 91.5%, and 79.2% for partial-thickness tears. HSROC analysis showed pooled sensitivity and specificity of 96% and 76% for any tear, 94% and 80% for partial-thickness tears, and 96% and 98% for full-thickness tears. Agreement measures also generally favored full-thickness tears. Operator expertise, study quality varied, and only one study evaluated a handheld ultrasound system. Conclusion: Ultrasound demonstrates strong diagnostic performance for supraspinatus tears, particularly full-thickness tears, while partial-thickness tears remain more challenging. Given the high observed sensitivity, well-performed US may serve as a practical first-line modality to rule out full-thickness tears, with MRI remaining complementary for equivocal partial-thickness tears, associated pathology, and preoperative assessment.