METHODS: We retrospectively collected CT scan data from 276 patients with pathologically confirmed primary bone tumors from 4 medical centers in Guangdong Province between January, 2010 and August, 2021. A convolutional neural network (CNN) was employed as the deep learning architecture. The optimal baseline deep learning model (R-Net) was determined through transfer learning, and an optimized model (S-Net) was obtained through algorithmic improvements. Multivariate logistic regression analysis was used to screen the clinical features such as sex, age, mineralization location, and pathological fractures, which were then connected with the imaging features to construct the deep learning fusion model (SC-Net). The diagnostic performance of the SC-Net model and machine learning models were compared with radiologists' diagnoses, and their classification performance was evaluated using the area under the receiver operating characteristic curve (AUC) and F1 score. RESULTS: In the external test set, the fusion model (SC-Net) achieved the best performance with an AUC of 0.901 (95% CI: 0.803-1.00), an accuracy of 83.7% (95% CI: 69.3%-93.2%) and an F1 score of 0.857, and outperformed the S-Net model with an AUC of 0.818 (95% CI: 0.694-0.942), an accuracy of 76.7% (95% CI: 61.4%-88.2%), and an F1 score of 0.828. The overall classification performance of the fusion model (SC-Net) exceeded that of radiologists' diagnoses. CONCLUSIONS The deep learning fusion model based on multi-center CT images and clinical features is capable of accurate classification of osseous and chondroid matrix mineralization and may potentially improve the accuracy of clinical diagnoses of osteogenic versus chondrogenic primary bone tumors.
Humans ; Deep Learning ; Bone Neoplasms/diagnostic imaging* ; Retrospective Studies ; Tomography, X-Ray Computed/methods* ; Neural Networks, Computer ; Male ; Female ; ROC Curve ; Algorithms

Epithelial and stromal tissues are components of the tumor microenvironment and play a major role in tumor initiation and progression. Distinguishing stroma from epithelial tissues is critically important for spatial characterization of the tumor microenvironment. Here, we propose BrcaSeg, an image analysis pipeline based on a convolutional neural network (CNN) model to classify epithelial and stromal regions in whole-slide hematoxylin and eosin (H&E) stained histopathological images. The CNN model is trained using well-annotated breast cancer tissue microarrays and validated with images from The Cancer Genome Atlas (TCGA) Program. BrcaSeg achieves a classification accuracy of 91.02％, which outperforms other state-of-the-art methods. Using this model, we generate pixel-level epithelial/stromal tissue maps for 1000 TCGA breast cancer slide images that are paired with gene expression data. We subsequently estimate the epithelial and stromal ratios and perform correlation analysis to model the relationship between gene expression and tissue ratios. Gene Ontology (GO) enrichment analyses of genes that are highly correlated with tissue ratios suggest that the same tissue is associated with similar biological processes in different breast cancer subtypes, whereas each subtype also has its own idiosyncratic biological processes governing the development of these tissues. Taken all together, our approach can lead to new insights in exploring relationships between image-based phenotypes and their underlying genomic events and biological processes for all types of solid tumors. BrcaSeg can be accessed at https://github.com/Serian1992/ImgBio.

Background & Objective: Endovascular treatment is the widely accepted treatment for patients with anterior circulation stroke within 6 hours of onset of stroke. We aimed to evaluate the advantages of endovascular treatment compared to standard medical treatment in treating patients with anterior circulation stroke beyond the 6-hour therapeutic window. Methods: We reviewed the literature concerning endovascular treatment versus medical treatment beyond the 6-hour therapeutic window. Using random-effects meta-analysis, we evaluated the following outcomes: modified Rankin scale in the three-month follow-up [excellent outcome (mRS≤1), functional independence (mRS≤2), moderate outcome(mRS≤3)], recanalization rate at 24 hours, mortality at 90 days or in-hospital, symptomatic intracranial hemorrhage, parenchymal hematoma type 2 and hemorrhagic infarction 1. Results: Four studies including 642 patients were evaluated. Endovascular treatment was associated with higher odds of excellent outcome (OR 2.55; 95% CI 1.48 to 4.41,), functional independence (OR 3.64; 95% CI 2.43 to 5.45), moderate outcome (OR 2.70; 95% CI 1.95-3.74) and recanalization rate at 24 hours (OR 8.81; 95%CI 2.81 to 27.69) compared to MT. No difference in the rates of mortality, symptomatic intracranial hemorrhage, parenchymal hematoma type 2 or hemorrhagic infarction 1 was found between the 2 groups. Studies using strict perfusion imaging inclusion selection showed better moderate outcome in comparison to the studies without perfusion imaging inclusion selection (P <0.012). Conclusion: Our study highlights the superiority of endovascular treatment over standard medical treatment alone for treating patients with anterior circulation stroke beyond 6 hours since stroke onset, although more studies are required for further investigation. Standard of strict selection for eligible patients before endovascular treatment should be based on DAWN or DEFFUSE 3 inclusion criteria.

Objective To compare the effect of repeated transcranial magnetic stimulation (rTMS) at different frequencies acting on left dorsolateral prefrontal cortex (LPFC) on arousal for coma patients after brain stem injury.Methods Ninety-nine patients with coma resulted from brain stem injury,admitted to our hospital from February 1,2015 to April 10,2018,were chosen in our study.Among them,30 patients weren't treated with rTMS (control group),33 patients were treated with 10 Hz rTMS (10 Hz rTMS treatment group),and 36 patients were treated with 20 Hz rTMS (20 Hz rTMS treatment group);treatments lasted for 20 d.All patients received routine coma arousal treatment.Glasgow coma scale (GCS) scores,electroencephalogram (EEG) grading,brainstem auditory evoked potential (BAEP) grading and incidence of adverse reactions were compared among the three groups before and after treatment.Results Before treatment,there were no significant differences in GCS total scores,language response scores,motor response scores,eye opening reaction scores,EEG grading and BAEP grading among the three groups (P＞0.05).The total GCS scores,and scores of language response,motor response and open eye response of patients in the 20 Hz rTMS treatment group and 10 Hz rTMS treatment group after treatment were significantly higher than those in the control group (P＜0.05);the total GCS scores and motor response scores of patients in the 20 Hz rTMS treatment group were significantly higher than those in 10 Hz rTMS treatment group after treatment (P＜0.05).After treatment,patients in the control group,10 Hz rTMS treatment group,and 20 Hz rTMS treatment group showed statistically significant differences in EEG grading and BAEP grading (P＜0.05).The incidence of adverse reactions in the 20 Hz rTMS group (19.4％) was statistically higher than that in the 10 Hz rTMS group (3.0％,P＜0.05).Conclusion High-frequency rTMS has an awakening effect on patients with coma resulted from brain stem injury,and the coma awakening effect of 20 Hz rTMS is partially better than that of 10 Hz rTMS,but it has the risk of increasing side effects such as epilepsy and scalp bum.

Objective To analyze the influence of general anesthesia (GA) on bilateral subthalamic nucleus deep brain stimulation (STN-DBS) in treating Parkinson's disease (PD) through microelectrode recording (MER),and discuss the differences between different modes of anesthesia.Methods A retrospective analysis was performed on clinical data of 31 PD patients accepted bilateral STN-DBS in our hospital from June 2015 to June 2017.Nine patients accepted surgery under GA (A group):4 patients were treated with intravenous anesthesia (A1 group),and 5 patients were treated with inhalation anesthesia (A2 group);22 patients accepted surgery under local anesthesia LA group.MER indexes,including STN discharge frequency,STN recorded length,and maximum target error,and short-term (6 months) efficacy were recorded.A linear regression analysis was performed to find possible influence factors on discharge frequency and improving rate of UPDRS scores.Results The discharge frequencies of B group,A1 group and A2 group were 51.42 Hz±6.28 Hz,35.79 Hz±7.02 Hz and 43.18 Hz±5.87 Hz,respectively,with significant differences (F=12.181,P=0.000);as compared with that in the B group,the discharge frequencies of A1 group and A2 group were significantly lower (P＜0.05).The STN recorded lengths of B group,A1 group and A2 group were 5.48 mm±0.33 mm,5.06 mm±0.15 mm and 5.22 mam±0.16 mm,respectively,with significant differences (F=4.115,P=0.027);as compared with that in the B group,the recorded lengths of A1 group and A2 group were significantly shorter (P＜0.05).A1 group had the maximum target error,but no significant differences were noted among the 3 groups (P＞ 0.05).Six months after the surgery,the UPDRS-Ⅲ scores and Schwab-England scores of A group and B group were decreased and daily levodopa equivalent (LEDD) was decreased.As compared with B group,A group had significantly better improvement in Hoehn & Yahr grading (P＜0.05).Disease durations were positively correlated with discharge frequency (r=0.539,P=0.002);age and improving rate of UPDRS scores were negatively correlated (r=-0.572,P=-0.001);preoperative LEDD and improving rate of UPDRS scores were positively correlated (r=0.725,P=-0.000).Conclusions Bilateral STN-DBS performed under GA in PD enjoys good efficacy,which shows no obvious difference as compared with that under LA.Inhalation anesthesia had less influence on electrophysiology than intravenous anesthesia.