Pancreatic ductal adenocarcinoma (PDAC), the most deadly solid malignancy, is typically detected late and at an inoperable stage. Early or incidental detection is associated with prolonged survival, but screening asymptomatic individuals for PDAC using a single test remains unfeasible due to the low prevalence and potential harms of false positives. Non-contrast computed tomography (CT), routinely performed for clinical indications, offers the potential for large-scale screening, however, identification of PDAC using non-contrast CT has long been considered impossible. Here, we develop a deep learning approach, pancreatic cancer detection with artificial intelligence (PANDA), that can detect and classify pancreatic lesions with high accuracy via non-contrast CT. PANDA is trained on a dataset of 3,208 patients from a single center. PANDA achieves an area under the receiver operating characteristic curve (AUC) of 0.986–0.996 for lesion detection in a multicenter validation involving 6,239 patients across 10 centers, outperforms the mean radiologist performance by 34.1% in sensitivity and 6.3% in specificity for PDAC identification, and achieves a sensitivity of 92.9% and specificity of 99.9% for lesion detection in a real-world multi-scenario validation consisting of 20,530 consecutive patients. Notably, PANDA utilized with non-contrast CT shows non-inferiority to radiology reports (using contrast-enhanced CT) in the differentiation of common pancreatic lesion subtypes. PANDA could potentially serve as a new tool for large-scale pancreatic cancer screening.

Introduction: Pancreatic cancer (PC) has a poor prognosis with high mortality, due to the lack of effective early diagnostic and prognostic tools. Materials and methods: In order to target and diagnose PC, we developed a dual-modal imaging probe using Glypican-1 (GPC-1) antibody conjugated with Gd–Au nanoclusters (NCs; Gd-Au-NC-GPC-1). GPC-1 is a type of cell surface heparan sulfate proteoglycan, which is often highly expressed in PC. The probe was successfully prepared with a hydrodynamic diameter ranging from 13.5 to 24.4 nm. Results: Spectral characteristics showed absorption at 280 nm and prominent emission at 650 nm. Confocal microscopic imaging showed effective detection of GPC-1 highly expressed PC cells by Gd-Au-NC-GPC-1, which was consistent with flow cytometry results. In vitro relaxivity characterization demonstrated that the r1 value of the probe was 17.722 s⁻¹ mM⁻¹ Gd, which was almost 4 times higher compared with that of Gd-diethylenetriaminepentacetate (DTPA; r1 value =4.6 s⁻¹ mM⁻¹ Gd). Gd-Au-NC-GPC-1 exhibited similar magnetic resonance (MR) signals when compared to Gd-DTPA even at lower Gd concentrations. Much higher MR signals were registered in PC cells (COLO-357) compared with normal cells (293T). Furthermore, Gd-Au-NC-GPC-1 could effectively detect PC cells in vivo by dual-modal fluorescence imaging/magnetic resonance imaging (FI/MRI) at 30 minutes postinjection. In addition, Gd-Au-NC-GPC-1 did not show significant biotoxicity to normal cells at tested concentrations both in vitro and in vivo. Conclusion: Gd-Au-NC-GPC-1 has demonstrated to be a promising dual-modal FI/MRI contrast agent for targeted diagnosis of PC.

Background: Imaging factors, specifically baseline plaque features on high-resolution magnetic resonance vessel wall imaging (HR-VWI) that could be associated with in-stent restenosis (ISR), are still unknown. We aimed to investigate the presenting clinical and plaque features on HR-VWI associated with ISR.Methods: Sixty-four patients with intracranial stent placement for intracranial atherosclerotic stenosis who had pre- and post-contrast T1-weighted HR-VWI on 3.0T prior to stenting were included in this analysis. Student's t-test, Mann–Whitney U test, χ2 test, or the Cochran–Mantel–Haenszel (CMH) test were used to compare clinical and baseline HR-VWI characteristics of the patients between the ISR and non-ISR groups. Univariable and multivariable logistic analysis were used to test the clinical and imaging factors associated with ISR.Results: Among the 64 patients, 9 patients (14.06%) developed ISR during the 2-year follow-up period. Plaque burden (median 0.89 vs 0.92, P=0.04), minimum lumen area (0.009 cm²vs 0.006 cm², P=0.04), plaque eccentricity (55.6% vs 89.1%, P＜0.01), enhancement ratio (1.36 vs 0.84, P＜0.01), and enhancement involvement (type 2 represents ≥50% cross-sectional wall involvement; 100% vs 63.6%, P=0.03) all significantly differed between patients with and without ISR. Multivariable analysis revealed that lower frequency of plaque eccentricity (OR 0.18, 95% CI 0.04 to 0.96, P=0.04) and higher enhancement ratio (OR 3.57, 95% CI 1.02 to 12.48, P=0.04) were independently associated with ISR.Conclusions: Preliminary findings showed that ISR was independently associated with plaque concentricity and higher enhancement ratios on pre-stenting HR-VWI for patients with symptomatic intracranial atherosclerotic stenosis.