Background and purpose:Diffuse large B-cell lymphoma(DLBCL)is subclassified into germinal center B-cell-like(GCB)and non-GCB subtypes,which differ in prognosis and treatment response.However,current distinction still relies on invasive pathological assays.This study developed radiomics and deep-learning models based on multiparametric magnetic resonance imaging(MRI)to non-invasively differentiate the two subtypes preoperatively,thereby reducing dependence on histopathological examination.Methods:This study retrospectively included patients with pathologically confirmed DLBCL diagnosed at Huashan Hospital,Fudan University,and other institutions between March 2013 and December 2024.Using multiparametric MRI data,we developed DLBCL-subtype classification models that combined 4 radiomics-based machine-learning algorithms:support vector machine(SVM),logistic regression(LR),Gaussian process(GP)and Naive Bayes(NB),with 3 deep-learning architectures[densely-connected convolutional networks 121(DenseNet121),residual network 101(ResNet101)and EfficientNet-b5].Additionally,two radiologists with different experience levels independently classified DLBCL on MRI in a blinded fashion.Model and radiologist performance were quantified using the area under the receiver operating characteristic curve(AUC),accuracy(ACC),and F1-score to evaluate their ability to distinguish GCB from non-GCB subtypes.This study was approved by the Ethics Committee of Huashan Hospital of Fudan University(No.KY2024-663),and all patients signed informed consents.Results:A total of 173 patients were enrolled(55 with GCB subtype and 118 with non-GCB subtype).Radiomics and deep learning methods effectively distinguished DLBCL subtypes.Among these,the GP radiomics model(based on T1-CE+T2-FLAIR+ADC sequences)and DenseNet121 deep learning model(based on T1-CE+T2-FLAIR+ADC sequences)demonstrated optimal performance.Both achieved excellent results on the internal validation set(GP:AUC=0.900,ACC=0.896,F1=0.840;DenseNet121:AUC=0.846,ACC=0.854,F1=0.774)and maintained robustness on the external validation set.Furthermore,the classification efficacy of the optimal AI model surpassed that of experienced radiologists(highest physician AUC=0.678).Conclusion:Radiomics and deep-learning models based on multiparametric MRI features can effectively differentiate GCB from non-GCB subtypes of DLBCL.Among them,GP and DenseNet121 exhibit outstanding performance,especially when integrating multi-sequence feature sets for classifying DLBCL subtypes on complex imaging data.

Background and purpose:Diffuse large B-cell lymphoma(DLBCL)is subclassified into germinal center B-cell-like(GCB)and non-GCB subtypes,which differ in prognosis and treatment response.However,current distinction still relies on invasive pathological assays.This study developed radiomics and deep-learning models based on multiparametric magnetic resonance imaging(MRI)to non-invasively differentiate the two subtypes preoperatively,thereby reducing dependence on histopathological examination.Methods:This study retrospectively included patients with pathologically confirmed DLBCL diagnosed at Huashan Hospital,Fudan University,and other institutions between March 2013 and December 2024.Using multiparametric MRI data,we developed DLBCL-subtype classification models that combined 4 radiomics-based machine-learning algorithms:support vector machine(SVM),logistic regression(LR),Gaussian process(GP)and Naive Bayes(NB),with 3 deep-learning architectures[densely-connected convolutional networks 121(DenseNet121),residual network 101(ResNet101)and EfficientNet-b5].Additionally,two radiologists with different experience levels independently classified DLBCL on MRI in a blinded fashion.Model and radiologist performance were quantified using the area under the receiver operating characteristic curve(AUC),accuracy(ACC),and F1-score to evaluate their ability to distinguish GCB from non-GCB subtypes.This study was approved by the Ethics Committee of Huashan Hospital of Fudan University(No.KY2024-663),and all patients signed informed consents.Results:A total of 173 patients were enrolled(55 with GCB subtype and 118 with non-GCB subtype).Radiomics and deep learning methods effectively distinguished DLBCL subtypes.Among these,the GP radiomics model(based on T1-CE+T2-FLAIR+ADC sequences)and DenseNet121 deep learning model(based on T1-CE+T2-FLAIR+ADC sequences)demonstrated optimal performance.Both achieved excellent results on the internal validation set(GP:AUC=0.900,ACC=0.896,F1=0.840;DenseNet121:AUC=0.846,ACC=0.854,F1=0.774)and maintained robustness on the external validation set.Furthermore,the classification efficacy of the optimal AI model surpassed that of experienced radiologists(highest physician AUC=0.678).Conclusion:Radiomics and deep-learning models based on multiparametric MRI features can effectively differentiate GCB from non-GCB subtypes of DLBCL.Among them,GP and DenseNet121 exhibit outstanding performance,especially when integrating multi-sequence feature sets for classifying DLBCL subtypes on complex imaging data.

Tumor-targeted immunotherapy is a remarkable breakthrough, offering the inimitable advantage of specific tumoricidal effects with reduced immune-associated cytotoxicity. However, existing platforms suffer from low efficacy, inability to induce strong immunogenic cell death (ICD), and restrained capacity of transforming immune-deserted tumors into immune-cultivated ones. Here, an innovative platform, perfluorooctyl bromide (PFOB) nanoemulsions holding MnO₂ nanoparticles (MBP), was developed to orchestrate cancer immunotherapy, serving as a theranostic nanoagent for MRI/CT dual-modality imaging and advanced ICD. By simultaneously depleting the GSH and eliciting the ICD effect via high-intensity focused ultrasound (HIFU) therapy, the MBP nanomedicine can regulate the tumor immune microenvironment by inducing maturation of dendritic cells (DCs) and facilitating the activation of CD8⁺ and CD4⁺ T cells. The synergistic GSH depletion and HIFU ablation also amplify the inhibition of tumor growth and lung metastasis. Together, these findings inaugurate a new strategy of tumor-targeted immunotherapy, realizing a novel therapeutics paradigm with great clinical significance.

Objective: : This study aims to investigate the relationship between aneurysm wall enhancement and clinical rupture risks based on the magnetic resonance vessel wall imaging (MR-VWI) quantitative methods. Methods: : One hundred and eight patients with 127 unruptured aneurysms were prospectively enrolled from Feburary 2016 to October 2017. Aneurysms were divided into high risk (≥10) and intermediate-low risk group (<10) according to the PHASES (Population, Hypertension, Age, Size of aneurysm, Earlier SAH history from another aneurysm, Site of aneurysm) scores. Clinical risk factors, aneurysm morphology, and wall enhancement index (WEI) calculated using 3D MR-VWI were analyzed and compared. Results: : In comparison of high-risk and intermediated-low risk groups, univariate analysis showed that neck width (4.5±3.3 mm vs. 3.4±1.7 mm, p=0.002), the presence of wall enhancement (100.0% vs. 62.9%, p<0.001), and WEI (1.6±0.6 vs. 0.8±0.8, p<0.001) were significantly associated with high rupture risk. Multivariate regression analysis revealed that WEI was the most important factor in predicting high rupture risk (odds ratio, 2.6; 95% confidence interval, 1.4–4.9; p=0.002). The receiver operating characteristic (ROC) curve analysis can efficiently differentiate higher risk aneurysms (area under the curve, 0.780; p<0.001) which have a reliable WEI cutoff value (1.04; sensitivity, 0.833; specificity, 0.67) predictive of high rupture risk. Conclusion : Aneurysms with higher rupture risk based on PHASES score demonstrate increased neck width, wall enhancement, and the enhancement intensity. Higher WEI in unruptured aneurysms has a predictive value for increased rupture risk.

Objective: The purpose of our study was to investigate the predictive abilities of clinical and computed tomography (CT)features for outcome prediction in patients with coronavirus disease (COVID-19). Materials and Methods: The clinical and CT data of 238 patients with laboratory-confirmed COVID-19 in our two hospitalswere retrospectively analyzed. One hundred sixty-six patients (103 males; age 43.8 ± 12.3 years) were allocated in thetraining cohort and 72 patients (38 males; age 45.1 ± 15.8 years) from another independent hospital were assigned in thevalidation cohort. The primary composite endpoint was admission to an intensive care unit, use of mechanical ventilation, ordeath. Univariate and multivariate Cox proportional hazard analyses were performed to identify independent predictors. Anomogram was constructed based on the combination of clinical and CT features, and its prognostic performance wasexternally tested in the validation group. The predictive value of the combined model was compared with models built on theclinical and radiological attributes alone. Results: Overall, 35 infected patients (21.1%) in the training cohort and 10 patients (13.9%) in the validation cohortexperienced adverse outcomes. Underlying comorbidity (hazard ratio [HR], 3.35; 95% confidence interval [CI], 1.67–6.71;p < 0.001), lymphocyte count (HR, 0.12; 95% CI, 0.04–0.38; p < 0.001) and crazy-paving sign (HR, 2.15; 95% CI, 1.03–4.48;p = 0.042) were the independent factors. The nomogram displayed a concordance index (C-index) of 0.82 (95% CI, 0.76–0.88),and its prognostic value was confirmed in the validation cohort with a C-index of 0.89 (95% CI, 0.82–0.96). The combinedmodel provided the best performance over the clinical or radiological model (p < 0.050). Conclusion Underlying comorbidity, lymphocyte count and crazy-paving sign were independent predictors of adverseoutcomes. The prognostic nomogram based on the combination of clinical and CT features could be a useful tool for predictingadverse outcomes of patients with COVID-19.

Objective To investigate the dynamic changes of blood-brain barrier(BBB) permeability after acute cerebral ischemia in rats with middle cerebral artery occlusion (MCAO) by dynamic contrast-enhanced(DCE)-MRI. Methods Sixty MCAO rat models were established by suture-occlusion method. All rats were divided randomly into twelve groups with different ischemia duration (3 hours, 6 hours, permanent) and reperfusion times (2, 6, 12 and 24 hours after reperfusion). Each group was examined by MRI at the time points. The BBB permeability parameters(Ktrans, Ve, Kep, rKtrans, rVe, rKep) were calculated by Siemens workstation and compared with Evans blue(EB) extravasation results. Multivariate analysis of variance (M-ANOVA), one-way analysis of variance (one-way ANOVA), Pearson analysis were respectively used to verify the influences of ischemia duration and reperfusion time on BBB permeability parameters, EB extravasation and relationships between parameters. Results In 3 hours and 6 hours ischemia duration groups, change of BBB permeability after reperfusion appeared biphasic. At 2 hours and 6 hours after reperfusion, BBB permeability increased, while rKtrans values and rVe values rose and rKep values dropped. BBB permeability decreased at 12 hours and increased again at 24 hours after reperfusion. The highest BBB permeability was observed at 6 hours after reperfusion. However, BBB permeability in permanent ischemia groups had uniphasic change, as its increase was rather mild as ischemia time went on. rKtrans values(1.99± 0.79)were positively correlated with rVe values(2.88 ± 1.78) (r=0.93, P<0.01) and negatively correlated with rKep values(0.66 ± 0.21) (r=-0.84, P<0.01). The negative correlation between rVe values and rKep valueswas also significant(r=- 0.80, P<0.01). EB extravasation results were consistent with MRI findings. Conclusions BBB permeability change was biphasic in reperfusion groups, while it was uniphasic in permanent ischemia groups. DCE-MRI may accurately reflect the changes of BBB permeability after acute cerebral ischemia. Both ischemic duration and reperfusion time had influences on BBB permeability. With prolongation of ischemic time, the duration of BBB permeability increase became shorter, BBB damage appeared earlier, with increased degree of ischemic damage.

Objective To evaluate the MRI application in the common non-traumatic brachial plexopathies.Methods Twenty-seven patients with non-traumatic brachial plexopathies were retrospectively analyzed,which consisted of 10 males and 17 females with age range from 22 to 70 years old.Among the patients,2 were neurofibromatosis,8 were schwannoma,3 were metastases,2 were radiation plexopathy,1 was multifocal motor neuropathy (MMN),and 11 were typical thoracic outlet syndrome (TOS).Before treatment,all patients underwent MRI of brachial plexus,in which 8 patients underwent Gd-DTPA enhanced MRI.Results In 2 cases of neurofibromatosis,the tumors appeared as fusiform,bead-like masses,growing along the course of brachial plexus with involvement of rootlets in vertebral canal,and iso-to little hypointensity on T1WI,inhomogeneous hyperintensity on T2WI and inhomogeneously enhanced.Schwannoma of 8 cases displayed as fusiform mass,eccentric to the original nerve with sharply defined edge,and hypointensity or inhomogeneous hyperintensity on T1WI,inhomogeneous hyperintensity on T2WI with cystic necrosis in 3 cases which was inhomogeneously enhanced.Two of the 3 cases of metastases manifested as multiple masses besides brachial plexus,the other one appeared as widespread lesion infiltrating brachial plexus and surrounding structures.Two cases of radiation plexopathy displayed as diffused thickened nerves,with increased signal intensity on short time inversion recovery (STIR),clear structure and no mass surrounding the plexus.One case of MMN showed thickened nerves with increased signal intensity on STIR.Eleven cases of typical TOS manifested as arched elevation of C8,T1 and inferior trunk,with thickened nerves and increased signal intensity on STIR.Elevated subclavian artery could be seen in 8 cases.And in one case of TOS,a mass in the end of cervical rib compressed the local brachial plexus.Conclusions MRI can clearly show the location,involved scope and the relationship with adjacent structures.So MRI can provide reliable information for clinical diagnosis and treatment.

Objective To assess the clinical features and associated factors of interstitial lung disease (ILD) in patients with dermatomyositis (DM).Methods Clinical data were retrospectively analyzed on 206 patients with DM collected at the Department of Dermatology,Huashan Hospital,Fudan University in the past 6 years.Chi-square test and t test were performed for statistical analysis.Results The prevalence of ILD was 49.03％ in the 206 patients with DM.Heliotrope rash on the upper eyelids,Gottron's sign (papules),arthralgia,and cough were correlated with the incidence of ILD in DM patients (all P ＜ 0.05),and of these factors,the prevalence of artharalgia and cough were positively correlated with the incidence of ILD,while the presence of Gottron's papules was negatively correlated.The patients with DM and ILD showed a higher prevalence of abnormal serum levels of lactate dehydrogenase (LDH),hydroxybutyrate dehydrogenase (HBDH) and anti-Jo-1 antibodies,as well as with a poorer pulmonary function,compared with those suffering from DM only (all P ＜0.05).Characteristic imaging findings on computed tomography (CT) scan in patients with DM and ILD included linear opacity (57.4％),high-density patchy opacity (31.7％),reticular opacity (16.8％) and even ground glass-like opacity (13.8％),usually at the bottom or apex of the lungs.Conclusions In patients with DM,the prevalence of artharalgia and cough is positively correlated,whereas the presence of Gottron's papules is negatively correlated,with the incidence of ILD.Characteristic imaging findings on CT scan in patients with DM and ILD are linear opacity,high-density patchy opacity,reticular opacity and ground glass-like opacity at the bottom and apex of lungs.

Objective To explore the feasibility of 256-slice whole-brain CT perfusion (CTP) in evaluate graft reperfusion after surgical revascularization and hemodynamic alterations before and after surgery in Moyamoya disease. Methods Twenty-five cases with Moyamoya disease were scanned on a 256-slice CT.CTP was performed pre- and post- surgical revascularization. The wolumetric CT angiographic ( CTA ) images were generated from volumetric data acquired at the arterial phase of CTP. CBF, CBV, TTP and MTT were measured in functional maps at the operated side within middle cerebral artery perfusion areas and contralateral mirroring areas. Relative CBF( rCBF), relative CBV (rCBV), relative TTP (rTTP), relative MTT (rMTT) were also obtained. Differences in perfusion CT values pre- and post operation were assessed with the paired t test or matched-pairs signed-ranks test. Data with normal distribution was present as : (x-)± s,while those with the non-normal distribution were present as M ( P25-P75 ). Results All the direct graft patencies were displayed on volumetric CTA. No significant differences were found between volumetric CTA and conventional CTA. Postoperative CBF, rCBF and rCBV values of the operated side [ 72. 86 (55.54-112. 19) ml · 100 g-1 · min-1 , 1. 31 ( 1.05-1.73), 1.45 ±0. 62] were significantly higher than those before operation [46.72(28.57-57.67) ml · 100 g-1 · min-1, 0.53(0.33-0.82), 1.01 ±0.36](Z=- 2.72, - 2. 98, t = - 2. 85, P ＜ 0. 05 ). Postoperative MTT, TTP and rTTP values of the operated side [ (3.98 ± 2. 36 ) s, ( 17.56 ± 4. 38 ) s, 1.01 ± 0. 09 ] were significantly lower than those before operation [(5.43±2.07) s,(19.40±3.87) s,1.14±0.28] (t=2.41,2.17,2.17, respectively, P＜0.05).However, no significant differences were detected for changes of CBV and rMTT after revascularization ( P ＞0. 05). Conclusion 256-slice CT has the potential value for the non-invasive assessment of both the graft patency and cerebral hemodynamics changes in moyamoya disease after surgery with administration of one contrast medium bolus in a single examination.

Objective To verify the accuracy of blood oxygenation level dependent (BOLD)-based activation using electrocortical stimulation mapping (ESM) and explore the value of language fMRI in the navigating operation of neurosurgery. Methods In 8 cases with brain tumors,BOLD-fMRI examinations were done before the operations. Under the state of awake anesthesia,the patients were aroused and ESM was conducted. Point-to-point comparison between the BOLD signal activations and the ESM was carried out under the surveillance of the neuro-navigation technology. In order to observe the sensibility and specificity of BOLD activations, the location of BOLD activations and the point of ESM was compared to calculate the stimulating positive points inside the regions of BOLD signals(real positive), outside BOLD regions(pseudo-negative), the stimulating negative points inside the regions of BOLD signals(pseudo-positive), and outside BOLD region(real negative). Two kinds of criteria for assessment were used. One was that the positive stimulating points were located in BOLD regions, and the other was that the positive stimulating points were located within 1 cm around the range of BOLD regions. Removal of the lesions were conducted with the tissue 1 cm around the language region preserved, and the cortex inside 0.5-1.0 cm distance from the positive points were retained. Results Of the 8 cases, only 6 finished the tasks. Among them, 3 cases were with astrocytoma of grade 2,2 were with astrocytoma of grade 3, and one with glioblastoma. The total number of stimulating points was 48, among which the positive points were 11. When the first criteria was applied, the sensitivity was 72.7% (8/11), and the specificity was 81.8% (30/37). When the second criteria was applied, the sensitivity was 82.0% (9/11),and the specificity was 75.6% (28/37). Follow-up after operation showed no aphasia occurred. Conclusions BOLD-fMRI had a high sensitivity and specificity in displaying the language regions. But due to the great variation of brain language area among the people, we need more studies of large sample to obtain enough experience before it can be used clinically.