United Imaging Healthcare (China)

Magnetic resonance imaging (MRI) is an indispensable tool for clinical knee examination, which often scans 2D stacked slices from multiple views. Radiologists typically locate lesion regions in one view, and then refer to other views to formulate a comprehensive diagnosis. However, existing computer-aided diagnosis methods fall short of identifying and fusing local regions in multi-view scans, leading to a decline in diagnostic performance and a heavy reliance on extensively annotated data. This paper introduces a novel framework that represents multi-view MRI scans as a knee graph, and conducts diagnosis using the proposed Knee Graph Network (KGNet). Moreover, KGNet is greatly enhanced by multi-task pre-training, which requires KGNet to reconstruct masked knee local patches and segment unmasked ones working alongside corresponding decoders. Experimental evaluations on public and in-house clinical datasets confirm that our framework outperforms existing approaches in diagnosing cartilage defects, anterior cruciate ligament tears, and knee abnormalities. In conclusion, our framework demonstrates the potential of enhancing knee disease diagnosis by representing multi-view MRI scans as a graph and employing multi-task pre-training in the graph network. The code is publicly available at https://github.com/zixuzhuang/KGNet.

AG10 is a molecule that can bind to the thyroxine (T4) binding pocket of transthyretin (TTR). The reversible association between the hydrophilic peptides and TTR using AG10 could increase the circulation half-life (t1/2). We hypothesized that modifying PSMA-617 with AG10 could extend the t1/2 of PSMA-617 and mitigate the limitations caused by insufficient uptake. The aim of this study was to develop an optimal AG10-modified radiotherapeutic agent that could maximize both tumor uptake and absorbed dose, thereby enhancing therapeutic efficacy for the treatment of tumors, even with moderate PSMA expression. Lysine was used as a linker for the conjugation of PSMA-617 and AG10 to synthesize AG10-PSMA-617 (AP-01). AP-01 was labeled with 68Ga to image prostate carcinoma xenografts with moderate PSMA expression, serving as a reference for PSMA radio-ligand therapy (PRLT) based treatments. AP-01 was labeled with 177Lu for the treatment of prostate carcinoma xenografts with moderate PSMA expression. The binding properties and biodistribution of 68Ga-AP-01 and 177Lu-AP-01 were studied both in vitro and in vivo. The cytotoxic effects of the 177Lu-AP-01 were evaluated in prostate carcinoma xenografts. PSMA-617 was successfully conjugated with AG10 using lysine as the linker and labeled with 177Lu to produce PRLT radiopharmaceuticals-177Lu-AP-01. AP-01 was labeled with 68Ga for imaging prostate carcinoma with moderate PSMA expression. AP-01 demonstrated a high binding affinity (IC50 = 3.42 nM) for PSMA in vitro, comparable to that of PSMA-617 (IC50 = 3.38 nM) (P = 0.324). PET/CT imaging of 68Ga-AP-01 demonstrated significantly improved tumor uptake and retention as compared with 68Ga-PSMA-617. The maximum uptake of 68Ga-AP-01 and 68Ga-PSMA-617 by 22Rv1 tumors was 18.07 ± 1.19%ID/g and 9.27 ± 1.20%ID/g at 4h pi, respectively (P < 0.05). The in vivo targeting abilities of 177Lu-AP-01 were evaluated through SPECT/CT imaging. The tumor (maximum)-to-muscle (maximum) (T/Mmax) ratio of 177Lu-AP-01 was 49.28 ± 7.18 at 72h pi. However, the T/Mmax ratio of 177Lu-PSMA-617 was 8.39 ± 8.57 at 48h pi. Biodistribution studies further confirmed the significantly higher tumor uptake of 177Lu-AP-01 (4.87 ± 0.91%ID/g) compared to 177Lu-PSMA-617 (1.12 ± 0.01%ID/g) at 96h post-injection (P < 0.05). The inhibitory efficiency of 177Lu-AP-01 on 22Rv1 tumor xenografts was 88.36%, which was superior to that of 177Lu-PSMA-617 (73.36%) (P < 0.05). 177Lu-AP-01 demonstrated no toxic effects on hepatorenal function, routine blood tests, or major organs in mice when compared to the control group. In this study, AP-01 was successfully synthesized. 68Ga-AP-01 and 177Lu-AP-01 were obtained with high radiochemical purity and stability. High binding affinity and PSMA targeting specificity were identified in vitro and in vivo. With significantly enhanced tumor uptake and retention, 177Lu-AP-01 has the potential to improve therapeutic efficacy for the treatment of prostate cancer with moderate PSMA expression.

Although 16-cm wide-detector CT scanners with prospective ECG-gating enable coronary artery imaging within a single cardiac cycle at a low radiation dose, many institutions still rely on scanners with detector widths <16cm. These scanners typically use retrospective scanning, resulting in higher radiation exposure. This study tests the feasibility of lowering the radiation dose of ECG-gated retrospective CCTA on 8-cm detector scanners to the level of prospective acquisition on 16-cm ones, using deep learning reconstruction (DLR). This study involved 83 low-dose (group A) and 62 routine-dose (group B) retrospective CCTA cases on an 8-cm detector scanner, with the low-dose protocol targeting a radiation level comparable to prospective acquisitions on 16-cm scanners. Groups Ref1 and Ref2 were used as the dose and image quality benchmarks. They consisted of prospective CCTA cases acquired on 16-cm scanners from a shared vendor, and were extracted from published studies. Specific subsets (A1/B1 and A2/B2) were selected from group A/B based on weight (55 to 75kg) and HR (<75bpm) to match Ref1 and Ref2. DLR was used for group A. Image quality was evaluated using signal-to-noise ratio (SNR) and subjective scoring. No significant differences in demographics were noted among Ref1/A1/B1 or Ref2/A2/B2 (all P>0.05). CTDIvol for A1 was comparable to Ref1 (median: 13.6mGy vs. 13.2mGy, P>0.999), whereas A2 had lower CTDIvol than Ref2 (median: 13.3mGy vs. 16.1mGy, P=0.018). DLR improved SNRs in group A, with higher values than Ref and B (all P<0.05). With DLR, image quality scores were comparable between groups A and B (all P>0.05). DLR can lower the radiation dose in retrospective CCTA on 8-cm detector scanners to that of prospective CCTA on 16-cm ones, with non-inferior image quality.

Lengthy acquisition time remains a key bottleneck for the widespread use of MRI in clinics. While accelerated MRI can reduce scan duration, it often introduces increased noise, compromising image quality and diagnostic reliability. In this study, we present a unified deep learning-based denoising model for multi-organ accelerated MRI, designed to operate directly on reconstructed images from commercial MRI systems. Our model was trained on a prospectively collected, large-scale real-world dataset comprising 148,930 noisy-clean image pairs from six clinical centers and four major MRI vendors, spanning six organs and 96 MRI protocols. On a test set of 20,143 real-world image pairs, our model consistently outperforms state-of-the-art denoising methods. Importantly, downstream evaluation using tissue segmentation demonstrates a 7.05% improvement in Dice score across multiple organs compared to noisy images. The model further generalizes effectively to 46,870 external clinical images from four independent cohorts, highlighting its robustness across various scanners and acquisition protocols. To assess clinical utility, two experienced radiologists conducted blinded evaluations across multiple organs, focusing on overall image quality, diagnostic confidence, and disease diagnosis. The denoised images retained high visual fidelity and yielded diagnostic performance equivalent to clean images even with acceleration factor of 3× compared to clinical scanning setup, such that many acquisitions can be completed within one minute. This unified MRI denoising model holds great potential for various clinical applications.

This study aimed to develop and validate a CT-based nomogram incorporating three-dimensional fractal dimension (FD 3D) to noninvasively predict tumor spread through air spaces (STAS) in stage IA lung adenocarcinoma. A retrospective analysis was performed on 110 patients with stage IA lung adenocarcinoma who underwent surgical resection. CT morphological features and fractal-dimension metrics were collected. Patients were categorized into STAS-positive (n = 48) and STAS-negative (n = 62) groups based on pathology. Univariate and multivariate logistic regression analyses were conducted to identify independent predictors of STAS. Receiver operating characteristic (ROC) curve analysis evaluated predictive performance, and a nomogram model was constructed and internally validated. Based on the nomogram score, patients were further stratified into low- and high-risk STAS groups using the optimal cutoff value determined by the maximum Youden index. Univariate analysis showed significant differences in consolidation-to-tumor ratio (CTR) (p < 0.001), morphological irregularity (p = 0.006), lobulation (p = 0.039), pleural indentation (p = 0.004), vascular convergence (p = 0.010), and FD 3D (p < 0.001) between groups. Multivariate analysis identified CTR, morphological irregularity, lobulation, and FD 3D as independent predictors of STAS in stage IA lung adenocarcinoma. The nomogram model achieved an area under the curve (AUC) of 0.894 (95%CI: 0.821-0.944; p < 0.001), with a sensitivity of 75.00% and a specificity of 90.32%. At the optimal cutoff value of 0.56, the model demonstrated a positive predictive value (PPV) of 85.71% in the high-risk group (n = 42, 38.18%) and a negative predictive value (NPV) of 82.35% in the low-risk group (n = 68, 61.82%), with significant differences in STAS prevalence between groups (85.71% vs. 17.65%, χ 2 = 46.18, p < 0.001). The CT-based nomogram integrating FD 3D and key imaging features can noninvasively predict STAS status in stage IA lung adenocarcinoma. This model shows promise for assisting surgical decision-making, though prospective studies are needed to validate its clinical utility.

To investigate alterations in glymphatic system function and cerebral blood flow (CBF) in individuals with acute mountain sickness (AMS), and to determine the relationship between these imaging findings and clinical symptom severity using diffusion tensor imaging along the perivascular space (DTI-ALPS) and three-dimensional arterial spin labeling (3D-ASL).Forty-one participants who rapidly ascended to high altitude were prospectively enrolled and divided into an AMS group (n = 21) and a non-AMS group (n = 20). All underwent DTI and 3D-ASL MRI sequences. The ALPS index and CBF values were measured, compared between groups, and correlated with Lake Louise AMS Scores (LLS). No significant difference was found in the ALPS index between the AMS and non-AMS groups (P > 0.05). However, the mean ALPS index showed a significant negative correlation with LLS (r = − 0.523, P < 0.001). In contrast, CBF was significantly higher in the cerebral cortex and white matter of the AMS group compared to the non-AMS group (P < 0.05). A weak but significant positive correlation was found between CBF in the corpus callosum and LLS (r = 0.322, P = 0.046). While glymphatic function, as measured by the ALPS index, correlates with AMS severity, it does not significantly differ between AMS and non-AMS groups. Elevated CBF in white matter and cerebral cortex, particularly in the corpus callosum, may serve as a potential imaging biomarker of AMS, underscoring the value of 3D-ASL for non-invasive assessment of cerebral perfusion changes in high-altitude conditions.

Pathological complete response (pCR) following conversion therapy for initially unresectable hepatocellular carcinoma (uHCC) remains challenging to predict preoperatively. This study developed and validated a model integrating clinicopathological and radiomic features of the tumor to predict pCR. In this multicenter retrospective study, temporal radiomics features were extracted from baseline, post-treatment, and delta (change) MRIs. Serum AFP response was calculated as log₁₀(preoperative AFP)/log₁₀(baseline AFP). Univariate analysis, collinearity assessment, LASSO, and random forest were employed to perform feature selection. Fourteen machine learning models were benchmarked, with performance evaluated by using comprehensive metrics AUC, NPV, PPV, sensitivity, specificity, calibration, and decision curve analysis. The model was developed and validated in a training (n=78), an internal test (n=32), and an independent validation cohort (n=44). The delta radiomic model significantly outperformed both baseline (test AUC: 0.835 vs. 0.483, p <0.05; validation AUC: 0.783 vs. 0.434, p <0.05) and preoperative models (test AUC: 0.685, p <0.05; validation AUC: 0.506, p <0.05), demonstrating superior predictive performance and generalization capability in predicting lesion-level pCR. Notably, when predicting patient-level pCR, the radiomic model also showed robust discrimination, with AUCs of 0.819 in the test set and 0.781 in the validation set. The combined radiomics-AFP model achieved even higher AUCs of 0.920 (test) and 0.857 (validation) in predicting lesion-level pCR. Dynamic radiomic changes effectively predict pCR in uHCC after conversion therapy. Combining delta radiomics with AFP response significantly improves predictive performance, offering a non-invasive method for assessing pCR and potentially guiding personalized treatment decisions.