OBJECTIVE:To further investigate the prognostic value of advanced cervical lymph node (CLN) extranodal extension (ENE) involving different structures and its potential synergistic effects with other nodal features in nasopharyngeal carcinoma (NPC). METHODS:A total of 1,373 non-metastatic NPC patients from three centers between 2011 and 2021 were enrolled. For advanced CLN ENE, involved structures, maximum axial diameter (MAD), necrosis, and bilaterality were recorded. Adjusted hazard ratios (AHRs) and random survival forests were used to identify the high-risk structure group. Kaplan-Meier analysis and multivariable Cox models were applied to evaluate overall survival (OS), progression-free survival (PFS), locoregional relapse-free survival (LRRFS), and distant metastasis-free survival (DMFS). RESULTS:CLN metastasis was identified in 1135/1373 patients (82.7 %), including 169 (14.9 %) with advanced CLN ENE. The high-risk structure group comprised the hyoid muscles, scalene muscles, longissimus cervicis muscle, interval muscles, internal/external/common carotid artery, lower cranial nerves region, and skin; all other involved structures were classified as intermediate risk. Advanced CLN ENE involving high-risk structures was the only independent adverse prognostic factor after adjustment. Survival differed significantly between high-risk and intermediate-risk CLN ENE. Among N1/N2 patients, high-risk CLN ENE showed survival similar to N3 disease (AHRs > 1), whereas intermediate-risk CLN ENE showed better OS than N3, with comparable PFS and DMFS (AHRs < 1). CONCLUSIONS:Compared with necrosis, MAD, or bilaterality, structure-based stratification more effectively distinguished prognostic differences in advanced CLN ENE. N1/N2 patients with high-risk or intermediate-risk CLN ENE had outcomes close to N3 disease but with marked heterogeneity.

PURPOSE:Patients with connective tissue diseases (CTDs) and pulmonary arterial hypertension (PAH) have a poor prognosis, and there is a lack of effective noninvasive prognostic tools. This study aimed to retrospectively analyze clinical data and multislice computed tomography (MSCT) chest CT parameters in CTD-PAH patients, and to develop a noninvasive prognostic model incorporating indicators. MATERIALS AND METHODS:A total of 170 patients with CTD-PAH admitted to the First Affiliated Hospital of Nanjing Medical University between May 2010 and April 2022 were enrolled in this study. Data on chest computed tomography-derived pulmonary artery diameters, esophageal dilatation, and interstitial lung disease (ILD) scores were collected. Patients were followed for 5 years to assess all-cause mortality. A nomogram incorporating MSCT parameters was developed and validated to predict the long-term prognosis. RESULTS:Independent risk factors for 5-year all-cause mortality in CTD-PAH patients included main pulmonary artery diameter (MPAd) (HR: 1.109, 95% CI: 1.010-1.218, P =0.030*), esophageal dilatation (HR: 2.757, 95% CI: 1.220-6.230, P =0.015*), and ILD score (HR: 1.066, 95% CI: 1.019-1.114, P =0.005*). A threshold MPAd of >35.70 mm was associated with a worse prognosis. The nomogram model, with a score >125, predicted a significantly lower 5-year survival rate in CTD-PAH patients. CONCLUSIONS:MPAd, esophageal dilatation, and ILD score are independent risk factors for 5-year all-cause mortality in CTD-PAH patients. The nomogram, which integrates these MSCT parameters, provides a reliable noninvasive tool for predicting reduced 5-year survival, offering valuable prognostic insight for personalized management of CTD-PAH.

To investigate the potential of apparent diffusion coefficient (ADC) map-based deep learning and dose distribution-based dosiomics in predicting radiation-induced temporal lobe injury (RTLI) in nasopharyngeal carcinoma (NPC). This retrospective study included 3578 NPC patients from Jiangsu Cancer Hospital receiving intensity-modulated radiation therapy (IMRT). Ninety-four RTLI patients were recruited based on inclusion criteria and matched 1:1 with 97 control subjects using propensity scores. Patients were randomly assigned to the training cohort (n = 135) and the validation cohort (n = 59). Deep transfer learning (DTL) features and dosiomics features were extracted from ADC map and three-dimensional dose distribution, respectively. Pearson’s correlation coefficient and the least absolute shrinkage and selection operator (LASSO) regression were employed to identify predictive features. Subsequently, eight machine learning classification models were trained to establish a prediction framework, encompassing Support Vector Machine, K-Nearest Neighbor, Random Forest, Extremely Randomized Trees, eXtreme Gradient Boosting, Light Gradient Boosting Machine, Adaptive Boosting and Multilayer Perceptron. The performance of clinical, DTL, dosiomics and feature fusion model was compared by the area under the curve (AUC). We constructed six pre-trained transfer learning networks and extracted DTL features, respectively. The results showed that pre-trained WideResNet 101 exhibited superior performance with an AUC of 0.786 in the validation cohort. The clinical model based on D1cc and induction chemotherapy demonstrated an AUC of 0.794 and the dosiomics model demonstrated an AUC of 0.903. Features fusion model demonstrated the highest AUC values in both the training (0.988) and validation (0.940) cohorts. The fusion model based on pretreatment ADC map and dose distribution provided a promising way to predict RTLI in NPC patients receiving IMRT, which can support clinicians in making decisions to develop individualized treatment plans and implement preventive measures.

OBJECTIVES:This study aimed to develop and construct a predictive model based on the quantitative parameters of full-volume dual-energy computed tomography (DECT) to forecast the International Association for the Study of Lung Cancer classification of non-mucinous invasive pulmonary adenocarcinoma (IPA). METHODS:The preoperative clinical and imaging data of 161 patients with pure solid type non-mucinous IPA from September 2021 to May 2024 were retrospectively analysed. The semiautomated software was used to perform full-volume segmentation of the lesions and the associated DECT quantitative parameters were recorded. Through univariate and multivariate logistic regression analyses, we identified independent characteristic variables that distinguished high-grade from low-grade non-mucinous IPA. We subsequently used these characteristic variables to construct a multiparameter model. RESULTS:Volume, slope of the spectral curve (λ40keV-100keV) and normalized iodine concentration (NIC) were identified as independent feature variables to distinguish low-grade and high-grade non-mucinous IPA. By utilizing these 3 variables, we constructed a quantitative visualization nomogram to distinguish the new IASLC grade of non-mucinous IPA. The model exhibited excellent performance in both the training and testing groups, with area under the curve (AUC) values of 0.884 (95% CI: 0.826-0.943) and 0.848 (95% CI: 0.738-0.958), respectively. CONCLUSION:This study successfully established and validated a nomogram based on DECT quantitative parameters, which can effectively differentiate high-grade and low-grade non-mucinous IPA and provides potential value for clinical decision-making. ADVANCES IN KNOWLEDGE:This study is the first attempt to apply a nomogram based on DECT to assess the invasiveness of non-mucous IPA.

This study aimed to refine risk stratification of advanced extranodal extension (ENE) in retropharyngeal lymph nodes (RLNs) and evaluate its prognostic significance. A total of 1,373 non-metastatic NPC patients at three centers between 2011 and 2021 were retrospectively enrolled. For advanced RLN ENE, we documented all involved structures, maximum axial diameter (MAD), presence of lymph node necrosis (LNN), and bilaterality. High-risk structures were identified using adjusted hazard ratios and LASSO-Cox models. Multivariable Cox models and Kaplan-Meier analyses were used to evaluate associations with survival outcomes. Of the 1,373 patients, 1061 had RLN metastasis, including 168 with advanced RLN ENE. Involvement of interval structures, the lower cranial nerves region, internal jugular vein, and longus capitis muscle was classified as high-risk, whereas isolated internal carotid artery involvement was low-risk. Advanced RLN ENE with high-risk structure involvement or LNN was defined as high-risk RLN ENE. High-risk RLN ENE was an independent adverse prognostic factor for NPC. N1/N2 patients with high-risk RLN ENE showed survival comparable to N3 disease. High-risk RLN ENE is an independent adverse prognostic factor in NPC. N1/N2 patients with high-risk RLN ENE exhibit N3-like survival, supporting its potential incorporation into future refinements of N staging.

ABSTRACTBackgroundRight ventricular (RV) failure is a well‐recognized pivotal prognostic factor of adverse outcomes in pulmonary artery hypertension (PAH), while RV dilation provides significant implications for adaptive or maladaptive changes. PAH is a predominant cause of mortality among patients with connective tissue disease (CTD). This study aims to elucidate the prognostic significance of RV morphology, as assessed by echocardiography (ECHO), in with CTD associated with PAH (CTD‐PAH).MethodsIn this ambispective cohort study, 143 CTD‐PAH patients diagnosed by right‐sided heart catheterization (RHC) from 2013 to 2023 were enrolled. Clinical characteristics, laboratory data, echocardiographic parameters (right ventricular end‐diastolic basal diameter index (RVDDI), tricuspid annular plane systolic excursion (TAPSE) and pulmonary arterial systolic pressure (PASP)) and therapy were recorded. The primary endpoint was defined as clinical worsening within a five‐year timeframe. Analytical methods included Kaplan–Meier survival analyses, the log‐rank test, and multivariable Cox proportional hazards regression to evaluate prognostic factors.ResultsThe study enrolled a total of 143 patients with CTD‐PAH, with a notable female predominance (95.1%) and a median age of 41.67 years; SLE‐PAH (49%) and pSS‐PAH (34%) were the most common subtypes, and 94% of the participants were in WHO‐FC II‐III. Among the participants, 34 (23.8%) patients experienced clinical worsening during a median follow‐up period of 21 months. After adjusting for confounders such as age and sex, RVDDI, as determined by ECHO was correlated with clinical worsening (HR 1.090; 95% CI: 1.019–1.166; p = 0.012). RVDDI > 25.81 mm/m2 predicts higher incidence of clinical worsening in CTD‐PAH. In the subgroup of TAPSE/PASP > 0.19 mm/mmHg, patients with RVDDI > 25.81 mm/m2 had a higher incidence of clinical worsening. The estimated event‐free survival rates at 1 and 3 years were 93.5% and 53.7%, respectively.ConclusionThe study demonstrates that RVDDI, as evaluated by ECHO, is a significant prognostic indicator for clinical worsening in CTD‐PAH. Its inclusion in the assessment of RV function and risk stratification may provide valuable incremental prognostic information for this CTD‐PAH population.

OBJECTIVES:The primary objective of this study is to investigate the potential of cardiac magnetic resonance (CMR) parameters to augment prognostic evaluation in patients with CTD-associated pulmonary arterial hypertension (CTD-PAH). METHODS:A retrospective, single-centre cohort study was conducted on 110 patients with CTD-PAH who were diagnosed via right heart catheterization between 2017 and 2023. These patients underwent CMR examinations based on clinical indications. RESULTS:After a mean follow-up period of 27 months, 27 patients experienced clinical deterioration events. After adjusting for age, sex and COMPERA 2.0 risk assessment model parameters, five CMR metrics were identified as independent risk factors for clinical deterioration in CTD-PAH patients. Receiver operating characteristic curve analysis showed that combining COMPERA 2.0 risk assessment model with CMR metrics improved predictive performance, with interventricular septum extracellular volume (IVS ECV) providing the greatest benefit among tissue metrics and right ventricular ejection fraction (RVEF) showing the most improvement among right heart function metrics. Kaplan-Meier（KM） survival curves revealed that patients with RVEF <39.2% and IVS ECV >31.4% had the poorest prognosis. Calibration curves indicated that integrating RVEF and IVS ECV significantly enhanced the accuracy and reliability of the COMPERA 2.0 risk assessment model in predicting 1-, 2- and 3-year event-free survival rates in CTD-PAH patients, with the C-index improving from 0.626 to 0.805. CONCLUSION:Combining RVEF and IVS ECV with COMPERA 2.0 risk assessment model significantly enhances the model's predictive accuracy for clinical deterioration in CTD-PAH patients.

In the field of medical image segmentation, the scarcity of labeled data poses a major challenge for existing models to accurately perceive target regions. Compared with manual annotation, gaze data is easier and cheaper to obtain. As a classical semi-supervised learning framework, mean-teacher can effectively use a large number of unlabeled medical images for stable training through self-teaching and collaborative optimization. Our study is based on the mean-teacher framework. By combining gaze data, it aims to address two crucial issues in semi-supervised medical image segmentation: 1) expand the scale and diversity of the dataset with limited labeled data; 2) enhance the network's perception ability. We propose the Human Gaze-based Dual Teacher Guidance Learning model (HG-DTGL). In this model, human gaze serves as an additional hidden 'teacher' in the mean-teacher architecture. We introduce the GazeMix to generate reliable mixed data to expand the diversity and scale of the dataset, and the Multi-scale Gaze Perception (MGP) module is used to extract the multi-scale perception of the network. A Gaze Loss is designed to align the model's perception with human gaze. We have verified HG-DTGL on multiple datasets of different modalities and achieved superior performance on a total of ten different organs/tissues, with extensive experiments. This demonstrates that our method has strong generalization ability for medical images of different modalities, and shows the great application potential of gaze data in semi-supervised medical image segmentation.

Objectives:To evaluate the prognostic value of cardiovascular magnetic resonance (CMR) parameters in patients with connective tissue disease-associated pulmonary arterial hypertension (CTD-PAH). Materials and methods:This retrospective cohort study involved 135 patients with documented CTD-PAH. Cardiac functional parameters including right ventricle end-diastolic volume index (RVEDVI), end-systolic volume index (RVESVI), stroke volume index (RVSVI), ejection fraction (RVEF), and SV/ESV; the volumetric parameter; T1 mapping and late gadolinium enhancement (LGE) parameters; strain parameters like RV global longitudinal strain(GLS); and hemodynamics parameters like mean velocity at the pulmonary artery (mvPA) and PA relative area change (PA RAC) were calculated from CMR images. Survival analysis was conducted using Kaplan-Meier and Cox regression. The endpoint was the occurrence of clinical deterioration events. Results:The median follow-up time was 22.27 months (interquartile range: 11.5-31.9 months).RVIP ECV, RVGLS, RVEDVI, RVESVI, RVMI, and RAA, as well as mvPA, SV/ESV, RVEF, and PA RAC, were associated with adverse outcomes. SV/ESV ≤ 0.55, RV GLS > -12.1 %, and PA RAC≤ 16.1 % were significant independent predictors of prognosis. The combined parameters provided incremental prognostic value over COMPERA 2.0 (area under the curve (AUC) from 0.771 to 0.899; P = 0.001).And time-dependent ROC curve confirmed the predictive efficiency of the combined parameters in year 3, with the AUC reaching 0.821(95 % CI: 0.698-0.945). Conclusion:Multiparametric CMR provides a non-invasive, efficient method to assess prognosis in CTD-PAH patients. Key parameters including SV/ESV, RV GLS, and PA RAC significantly predicted survival and confer incremental prognostic utility over COMPERA 2.0, offering reliable prognostic markers for clinical interventions.

Background and Objectives:Pulmonary arterial hypertension (PAH) is a life-threatening condition that requires optimized medical therapy to maintain a low-risk profile. This study assessed the effects of initial PAH-specific combination therapy with tadalafil/sildenafil on clinical and functional outcomes in a real-world setting. Methods:We conducted a multicenter retrospective study of 85 patients diagnosed with connective tissue disease-associated PAH (CTD-PAH) via right heart catheterization from 2009 to 2023. Data on treatment regimens and efficacy measures, including 6-min walk distance (6MWD), N-terminal pro-B-type natriuretic peptide (NT-pro BNP), soluble suppression of tumorigenicity 2 (sST2), World Health Organization (WHO) functional class, risk stratification, treat-to-target status and survival, were collected. Results:Patients receiving initial combination therapy with endothelin receptor antagonists (ERAs) and phosphodiesterase-5 inhibitors showed varied improvements. The tadalafil plus ERAs combination significantly reduced NT-pro BNP levels and improved risk status (P < 0.05). Notable enhancements in 6MWD, soluble ST2, and WHO functional class were observed in the tadalafil plus ERA group (P < 0.001), but not in the sildenafil group (P > 0.05). Additionally, 1-year treat-to-target rates were higher in the tadalafil plus ERA group (73.5%) than in the sildenafil group (45.6%, P = 0.005). Conclusion:These findings suggest that tadalafil combined with ERAs leads to better improvements in exercise capacity, functional class, and treatment goals compared to sildenafil-based regimens, offering valuable insights for optimizing CTD-PAH treatment.

BackgroundBrain metastases (BM), originating from extracranial malignancies, significantly threaten patient health. Accurate BM identification is crucial but labor-intensive manually. This study developed and validated a system for BM diagnosis, assessing its performance and stability.Methods470 patients diagnosed with BM were divided into an 80% training set (n=379) and a 20% internal test set (n=91) using systematic sampling. An additional 172 patients were retrospectively enrolled for external validation. A comprehensive preprocessing pipeline was implemented. We developed a 3D U-Net model with a ResNet-34 backbone for BM prediction. MRI scans were resampled to 0.833 mm³ isotropic voxels, underwent skull stripping using SynthStrip, and were intensity-normalized via Z-score normalization. The model was trained on MRI scans paired with segmentation masks, utilizing ImageNet-pretrained encoder weights and a patch-based strategy (128×128×128 voxels).ResultsThe model maintained perfect specificity and AUCs across gender and age groups, with no significant differences in other metrics, confirming false positive exclusion unaffected by demographics. By cancer type: Internal testing showed significant difference of AUC (p<0.001) between lung cancer (n=74) and other cancers (n=17). The differences of other performance metrics were not statistically significant (p>0.13), though other cancers showed higher median F1/IoU/MCC. External validation showed other cancers (n=79) had significantly higher precision than lung cancer (n=93) (p<0.05). Lung cancer AUC (0.82) was significantly lower than other cancers (0.89) (p<0.001), suggesting need for sensitivity optimization; both maintained specificity=1.0000. Model time was significantly shorter than manual annotation (internal: 69s vs 113s; external: 66s vs 96s; both p<0.001), with high agreement.ConclusionThe model demonstrated strong robustness and perfect specificity across demographics. While showing cancer type dependency (requiring improved lung cancer sensitivity), its high efficiency (40%-50% time reduction) and generalization provide a solid foundation for clinical translation.

OBJECTIVES:To identify risk factors for myocardial involvement in idiopathic inflammatory myopathy (IIM) and evaluate their prognostic value. METHODS:We analysed 92 IIM patients with abnormal cardiac troponin T (cTnT). Myocardial involvement was diagnosed by late gadolinium enhancement on cardiovascular magnetic resonance. All-cause mortality was recorded during follow-up. RESULTS:Myocardial involvement occurred in 68.5% and was associated with higher cTnT/creatine kinase (CK) ratios and anti-Ro52 positivity. Anti-Ro52-positive patients exhibited higher rates of late gadolinium enhancement and increased E/e'. Both cTnT/CK [odds ratio (OR) = 1.030, P = .024] and anti-Ro52 (OR = 5.629, P = .003) independently predicted myocardial involvement. A cTnT/CK cutoff > 19.3% predicted myocardial involvement [area under the curve (AUC) = 0.660], rising to 0.780 when combined with anti-Ro52. Subgroup analysis showed cTnT/CK was discriminative only in anti-Ro52-negative individuals. During a 36-month follow-up, 18 deaths occurred. Adjusted Cox regression identified cTnI positivity [hazard ratio (HR) = 7.395, P = .001] and cTnT/CK (HR = 1.012, P = .037) as independent mortality predictors. Time-dependent receiver operating characteristic at 3 years showed AUCs of 0.68 (cTnI) and 0.64 (cTnT/CK). Kaplan-Meier analysis confirmed worse survival with positive cTnI or a high cTnT/CK. CONCLUSIONS:The cTnT/CK ratio identifies myocardial involvement and predicts mortality in IIM patients with abnormal cTnT. Combining it with anti-Ro52 antibodies improves the detection of myocardial involvement.

Purpose: The aim of this study was to explore the association of cardiac CT-based left atrium (LA) structural and functional parameters and left atrial epicardial adipose tissue (LA-EAT) parameters with postablation atrial fibrillation (AF) recurrence within 2 years. Materials and Methods: Contrast-enhanced cardiac CT images of 286 consecutive AF patients (median age: 65 y; 97 females) who underwent initial ablation between June 2018 and June 2020 were retrospectively analyzed. Structural and functional parameters of LA, including maximum and minimum volume and ejection fraction of LA and left atrial appendage (LAA), and LA-EAT volume, were measured. The body surface area indexed maximum and minimum volume of LA (LAVImax, LAVImin) and LAA (LAAVImax, LAAVImin), and LA-EAT volume index (LA-EATVI) were calculated. Independent predictors of AF recurrence were determined using Cox regression analysis. The clinical predictors were added to the imaging predictors to build a combined model (clinical+imaging). The predictive performance of the clinical, imaging, and combined models was assessed using the area under the receiver operating characteristics curve (AUC). Results: A total of 108 (37.8%) patients recurred AF within 2 years after ablation at a median follow-up of 24 months (IQR=11, 32). LA and LAA size and LA-EAT volume were significantly increased in patients with AF recurrence (P<0.05). After the multivariable regression analysis, LA-EATVI, LAAVImax, female sex, AF duration, and stroke history were independent predictors for AF recurrence. The combined model exhibited superior predictive performance compare to the clinical model (AUC=0.712 vs. 0.641, P=0.023) and the imaging model (AUC=0.712 vs. 0.663, P=0.018). Conclusion: Cardiac CT-based LA-EATVI and LAAVImax are independent predictors for postablation AF recurrence within 2 years and may provide a complementary value for AF recurrence risk assessment.

To explore the association of ventricle epicardial fat volume (EFV) calculated by cardiac magnetic resonance (CMR) and the insulin resistance indicator of triglyceride–glucose (TyG) index in patients with chronic HF (CHF), this retrospective cohort study included adult CHF patients with confirmed diagnosis of heart failure from January 2018 to December 2020. All patients underwent 3.0T CMR, and EFV were measured under short-axis cine. Spearman correlation, multivariate linear regression, and restricted cubic spline (RCS) regression were used to analyze their association. There were 516 patients with CHF, of whom 69.8

The extracellular volume (ECV) fraction derived from cardiac magnetic resonance (CMR) can reflect various pathologies. The application of ECVs was limited by the strict requirement that hematocrit (Hct0) should be obtained within 24 hours of CMR scan. The aim of this study was to obtain accurate and convenient ECV calculated from the venous Hct and synthetic Hct in CMR. A total of 839 subjects were retrospectively enrolled. The subjects were divided into derivation cohort for local sex-specific models and validation cohort for assessing the accuracy of different ECVs. In the validation cohort, venous Hcts from 7 days before the scan (Hct1 − 7), outside 7 days (Hct> 7), the closest day (Hctclosest), and Hctsyn were compared with Hct0. The agreement and correlation of the conventional ECV (ECV0) with the corresponding ECVs were analyzed. The factors affecting the accuracy of ECVsyn were assessed. ECV1–7 and ECVclosest had the best correlation and smallest bias with ECV0 (R = 0.959 and 0.951, bias = 0.02

Heart failure (HF) is associated with high rates of morbidity and mortality. The value of deep learning survival prediction models using chest radiographs in patients with heart failure is currently unclear. The aim of our study is to develop and validate a deep learning survival prediction model using chest X-ray (DLSPCXR) in patients with HF. The study retrospectively enrolled a cohort of 353 patients with HF who underwent chest X-ray (CXR) at our institution between March 2012 and March 2017. The dataset was randomly divided into training (n = 247) and validation (n = 106) datasets. Univariate and multivariate Cox analysis were conducted on the training dataset to develop clinical and imaging survival prediction models. The DLSPCXR was trained and the selected clinical parameters were incorporated into DLSPCXR to establish a new model called DLSPinteg. Discrimination performance was evaluated using the time-dependent area under the receiver operating characteristic curves (TD AUC) at 1, 3, and 5-years survival. Delong’s test was employed for the comparison of differences between two AUCs of different models. The risk-discrimination capability of the optimal model was evaluated by the Kaplan–Meier curve. In multivariable Cox analysis, older age, higher N-terminal pro-B-type natriuretic peptide (NT-ProBNP), systolic pulmonary artery pressure (sPAP) > 50 mmHg, New York Heart Association (NYHA) functional class III–IV and cardiothoracic ratio (CTR) ≥ 0.62 in CXR were independent predictors of poor prognosis in patients with HF. Based on the receiver operating characteristic (ROC) curve analysis, DLSPCXR had better performance at predicting 5-year survival than the imaging Cox model in the validation cohort (AUC: 0.757 vs. 0.561, P = 0.01). DLSPinteg as the optimal model outperforms the clinical Cox model (AUC: 0.826 vs. 0.633, P = 0.03), imaging Cox model (AUC: 0.826 vs. 0.555, P < 0.001), and DLSPCXR (AUC: 0.826 vs. 0.767, P = 0.06). Deep learning models using chest radiographs can predict survival in patients with heart failure with acceptable accuracy.

PURPOSE:The relationship between plaque progression and pericoronary adipose tissue (PCAT) radiomics has not been comprehensively evaluated. We aim to predict plaque progression with PCAT radiomics features and evaluate their incremental value over quantitative plaque characteristics. PATIENTS AND METHODS:Between January 2009 and December 2020, 500 patients with suspected or known coronary artery disease who underwent serial coronary computed tomography angiography (CCTA) ≥2 years apart were retrospectively analyzed and randomly stratified into a training and testing data set with a ratio of 7:3. Plaque progression was defined with annual change in plaque burden exceeding the median value in the entire cohort. Quantitative plaque characteristics and PCAT radiomics features were extracted from baseline CCTA. Then we built 3 models including quantitative plaque characteristics (model 1), PCAT radiomics features (model 2), and the combined model (model 3) to compare the prediction performance evaluated by area under the curve. RESULTS:The quantitative plaque characteristics of the training set showed the values of noncalcified plaque volume (NCPV), fibrous plaque volume, lesion length, and PCAT attenuation were larger in the plaque progression group than in the nonprogression group ( P < 0.05 for all). In multivariable logistic analysis, NCPV and PCAT attenuation were independent predictors of coronary plaque progression. PCAT radiomics exhibited significantly superior prediction over quantitative plaque characteristics both in the training (area under the curve: 0.814 vs 0.615, P < 0.001) and testing (0.736 vs 0.594, P = 0.007) data sets. CONCLUSIONS:NCPV and PCAT attenuation were independent predictors of coronary plaque progression. PCAT radiomics derived from baseline CCTA achieved significantly better prediction than quantitative plaque characteristics.