Fang Liu

Purpose: To quantify the amount of error in commonly used radiographic measurements of patella alta with progressive malposition of the knee. Methods: Three‐dimensional models were derived from computed tomography scans of patients with unilateral patellar instability. Models with knee flexion angles between 20° and 30° were projected onto 2‐dimensional radiographs to create a perfect lateral radiograph. Progressive internal/external rotation and ad/abduction of the knee were performed to simulate malpositioning of the knee. For each 5° increment of aberrant positioning, corresponding measurements of posterior and distal condylar overlap were measured, and measurements of patella alta were compared, including Insall‐Salvati (IS), modified IS, Blackburne‐Peel Index, Caton‐Deschamps Index, and patellotrochlear index, with a 0.05 change considered to be clinically significant. Results: A total of 40 knees from 20 patients were included in this study. On radiographic views, for every 5° of aberrant rotation or ab/adduction, the overlap between the posterior condyles increased by approximately 4 mm ( P < .001). IS measurements were the least sensitive to aberrant motion (range −0.023 to 0.012). Caton‐Deschamps Index and Blackburne‐Peel Index showed clinically and statistically relevant deviations at larger degrees of external rotation, whereas modified IS showed statistically relevant deviations in both abduction and adduction (0.054 at 10°). Conclusions: Malposition of the knee on 2‐dimensional radiographs produced clinically relevant errors in the measurements of patellar height that varied by the method of measurement. The extent of condylar overlap can be used as a surrogate to estimate the extent of malpositioning and the potential for error in the measurement of patella alta on radiographic images. Level of Evidence: Level III, case‐control study.

Purpose: To investigate whether a vision‐language foundation model can enhance undersampled MRI reconstruction by providing high‐level contextual information beyond conventional priors. Theory and Methods: We proposed a semantic distribution‐guided reconstruction framework that uses a pre‐trained vision‐language foundation model to encode both the reconstructed image and auxiliary information into high‐level semantic features. A contrastive objective aligns the reconstructed representation with the target semantic distribution, ensuring consistency with high‐level perceptual cues. The proposed objective works with various deep learning‐based reconstruction methods and can flexibly incorporate semantic priors from multimodal sources. To test the effectiveness of these semantic priors, we evaluated reconstruction results guided by priors derived from either image‐only or image‐language auxiliary information. Results: Experiments on knee and brain datasets demonstrate that semantic priors from images preserve fine anatomical structures and achieve superior perceptual quality, as reflected in lower LPIPS values, higher Tenengrad scores, and improved scores in the reader study, compared with conventional regularization. The image‐language information further expands the semantic distribution and enables high‐level control over reconstruction attributes. Across all evaluations, the contrastive objective consistently guided the reconstructed features toward the desired semantic distributions while maintaining data fidelity, demonstrating the effectiveness of the proposed optimization framework. Conclusion: The study highlights that vision‐language foundation models can improve undersampled MRI reconstruction through semantic‐space optimization.

This study introduces a domain‐conditioned and temporally guided diffusion framework for accelerated dynamic MRI reconstruction, in which the reverse diffusion process is explicitly guided to model spatiotemporal structure in time‐resolved data. The framework integrates temporal information from time‐resolved dimensions, allowing for the concurrent capture of intraframe spatial features and interframe temporal dynamics in diffusion modeling. Meanwhile, it employs additional spatiotemporal and self‐consistent frequency‐temporal priors to guide the diffusion process, ensuring precise temporal alignment and enhancing fine image detail recovery. To facilitate a smooth diffusion process, the nonlinear conjugate gradient algorithm is utilized during the reverse diffusion steps. The proposed model was tested on two types of MRI data: Cartesian‐acquired multicoil cardiac MRI and golden‐angle‐radial‐acquired multicoil free‐breathing lung MRI, across various undersampling rates. It achieved high‐quality reconstructions, demonstrating improved temporal alignment and structural recovery compared with other competitive reconstruction methods, both qualitatively and quantitatively. This diffusion framework exhibited robust performance in handling both Cartesian and non‐Cartesian acquisitions, effectively reconstructing dynamic datasets in cardiac and lung MRI under different imaging conditions.

To predict cartilage tumor malignancy from radiographic images combined with readily available non-imaging information based on a vision-language foundation model. This single-institution study assembled a dataset of 3336 radiographs from 455 patients with enchondroma or chondrosarcoma that was assembled from two sources: (1) patients with histopathology-confirmed diagnoses of enchondroma or chondrosarcoma, and (2) patients with imaging-stable enchondroma without biopsy, confirmed through long-term imaging follow-up. An adapted vision-language foundation model based on the pre-trained CLIP (Contrastive Language-Image Pretraining) architecture was fine-tuned with our proposed Medical Knowledge Adapters and evaluated using 10-fold patient-level cross-validation to predict cartilage tumor malignancy from plain radiographs and demographic information. Using radiographs alone, the model achieved an Areas Under the receiver operating characteristic Curve (AUC) of 0.91 ± 0.04. Incorporating demographics improved the AUC to 0.94 ± 0.02. Subgroup analysis demonstrated robust generalizability across tumor grades with an AUC of 0.91 ± 0.07 in distinguishing atypical cartilaginous tumors (ACT) previously known as low grade chondrosarcomas, and 0.95 ± 0.02 in differentiating high-grade chondrosarcomas from enchondromas. Within the clinically challenging extremity subgroup (enchondroma vs ACT/LGCS), the model achieved an AUC of 0.79 ± 0.14, reflecting diagnostic difficulty observed in clinical practice. This foundation model-based approach demonstrates strong performance using accessible data sources, offering a non-invasive, cost-efficient, and scalable solution for cartilage tumor assessment in musculoskeletal oncology.

Purpose: To develop a self‐supervised scan‐specific deep learning framework for reconstructing accelerated multiparametric quantitative MRI (qMRI). Methods: We propose REFINE‐MORE (REference‐Free Implicit NEural representation with MOdel REinforcement), combining an implicit neural representation (INR) architecture with a model reinforcement module that incorporates MR physics constraints. The INR component enables informative learning of spatiotemporal correlations to initialize multiparametric quantitative maps, which are then further refined through an unrolled optimization scheme enforcing data consistency. To improve computational efficiency, REFINE‐MORE integrates a low‐rank adaptation strategy that promotes rapid model convergence. We evaluated REFINE‐MORE on accelerated multiparametric quantitative magnetization transfer imaging for simultaneous estimation of free water spin–lattice relaxation, tissue macromolecular proton fraction, and magnetization exchange rate, using both phantom and in vivo brain data. Results: Under 4× and 5× accelerations on in vivo data, REFINE‐MORE achieved superior reconstruction quality, demonstrating the lowest normalized root‐mean‐square error and highest structural similarity index compared to baseline methods and other state‐of‐the‐art model‐based and deep learning approaches. Phantom experiments further showed strong agreement with reference values, underscoring the robustness and generalizability of the proposed framework. Additionally, the model adaptation strategy improved reconstruction efficiency by approximately fivefold. Conclusion: REFINE‐MORE enables accurate and efficient scan‐specific multiparametric qMRI reconstruction, providing a flexible solution for high‐dimensional, accelerated qMRI applications.

Purpose: To propose a signal acquisition and modeling framework for multi‐component tissue quantification that encompasses transmit field inhomogeneity, multi‐component relaxation and magnetization transfer (MT) effects. Theory and Methods: By applying off‐resonance irradiation between excitation and acquisition within an RF‐spoiled gradient‐echo scheme, in combination with multiple echo‐time acquisitions, both Bloch‐Siegert shift and magnetization transfer effects are simultaneously induced while relaxation and spin exchange processes occur concurrently. The spin dynamics are modeled using a three‐pool framework, from which an analytical signal equation is derived and validated through numerical Bloch simulations. Monte Carlo simulations were further performed to analyze and compare the model's performance. Finally, the feasibility of this novel approach was investigated in vivo in human brain and knee tissues. Results: Simulation results showed excellent agreement with the derived analytical signal equation across a wide range of flip angles and echo times. Monte Carlo analyses further validated that the three‐pool parameter estimation pipeline performed robustly over various signal‐to‐noise ratio conditions. Multi‐parameter fitting results from in vivo brain and knee studies yielded values consistent with previously reported literature. Collectively, these findings confirm that the proposed method can reliably characterize multi‐component tissue parameters in macromolecule‐rich environments while effectively compensating for inhomogeneity. Conclusion: A signal acquisition and modeling framework for multi‐component tissue quantification that accounts for magnetization transfer effects and inhomogeneity has been developed and validated. Both simulation and experimental results confirmed the robustness of this method and its applicability to various tissue types in the brain and knee.

Purpose: To investigate whether a vision-language foundation model can enhance undersampled MRI reconstruction by providing high-level contextual information beyond conventional priors. Methods: We proposed a semantic distribution-guided reconstruction framework that uses a pre-trained vision-language foundation model to encode both the reconstructed image and auxiliary information into high-level semantic features. A contrastive objective aligns the reconstructed representation with the target semantic distribution, ensuring consistency with high-level perceptual cues. The proposed objective works with various deep learning-based reconstruction methods and can flexibly incorporate semantic priors from multimodal sources. To test the effectiveness of these semantic priors, we evaluated reconstruction results guided by priors derived from either image-only or image-language auxiliary information. Results: Experiments on knee and brain datasets demonstrate that semantic priors from images preserve fine anatomical structures and achieve superior perceptual quality, as reflected in lower LPIPS values, higher Tenengrad scores, and improved scores in the reader study, compared with conventional regularization. The image-language information further expands the semantic distribution and enables high-level control over reconstruction attributes. Across all evaluations, the contrastive objective consistently guided the reconstructed features toward the desired semantic distributions while maintaining data fidelity, demonstrating the effectiveness of the proposed optimization framework. Conclusion: The study highlights that vision-language foundation models can improve undersampled MRI reconstruction through semantic-space optimization.

Quantitative susceptibility mapping (QSM) is a powerful magnetic resonance imaging (MRI) technique for assessing tissue composition in the human body. For imaging short-T2 tissues in the musculoskeletal (MSK) system, ultrashort echo time (UTE) imaging plays a key role. However, UTE-based QSM (UTE-QSM) often involves repeated acquisitions, making it vulnerable to inter-scan motion. In this study, we investigate the effects of motion on UTE-QSM and introduce strategies to reduce motion-induced artifacts. Eight healthy male volunteers underwent UTE-QSM imaging of the knee joint, while an additional seven participated in imaging of the ankle joint. UTE-QSM was conducted using multiple echo acquisitions, including both UTE and gradient-recalled echoes, and processed using the iterative decomposition of water and fat with echo asymmetry and least-squares estimation (IDEAL) and morphology-enabled dipole inversion (MEDI) algorithms. To assess the impact of motion, datasets were reconstructed both with and without motion correction. Furthermore, we evaluated a two-step UTE-QSM approach that incorporates tissue boundary information. This method applies edge detection, excludes pixels near detected edges, and performs a two-step QSM reconstruction to reduce motion-induced streaking artifacts. In participants exhibiting substantial inter-scan motion, prominent streaking artifacts were evident. Applying motion registration markedly reduced these artifacts in both knee and ankle UTE-QSM. Additionally, the two-step UTE-QSM approach, which integrates tissue boundary information, further enhanced image quality by mitigating residual streaking artifacts. These results indicate that motion-induced errors near tissue boundaries play a key role in generating streaking artifacts in UTE-QSM. Inter-scan motion poses a fundamental challenge in UTE-QSM due to the need for multiple acquisitions. However, applying motion registration along with a two-step QSM approach that excludes tissue boundaries can effectively suppress motion-induced streaking artifacts, thereby improving the accuracy of musculoskeletal tissue characterization.

Magnetic Resonance Imaging (MRI) is indispensable in clinical practice but remains constrained by fragmented, multi-stage workflows encompassing acquisition, reconstruction, segmentation, detection, diagnosis, and reporting. While deep learning has achieved progress in individual tasks, existing approaches are often anatomy- or application-specific and lack generalizability across diverse clinical settings. Moreover, current pipelines rarely integrate imaging data with complementary language information that radiologists rely on in routine practice. Here, we introduce OmniMRI, a unified vision-language foundation model designed to generalize across the entire MRI workflow. OmniMRI is trained on a large-scale, heterogeneous corpus curated from 60 public datasets, over 220,000 MRI volumes and 19 million MRI slices, incorporating image-only data, paired vision-text data, and instruction-response data. Its multi-stage training paradigm, comprising self-supervised vision pretraining, vision-language alignment, multimodal pretraining, and multi-task instruction tuning, progressively equips the model with transferable visual representations, cross-modal reasoning, and robust instruction-following capabilities. Qualitative results demonstrate OmniMRI's ability to perform diverse tasks within a single architecture, including MRI reconstruction, anatomical and pathological segmentation, abnormality detection, diagnostic suggestion, and radiology report generation. These findings highlight OmniMRI's potential to consolidate fragmented pipelines into a scalable, generalist framework, paving the way toward foundation models that unify imaging and clinical language for comprehensive, end-to-end MRI interpretation.