Retrospective Cohort Study On Perioperative Integrated Chinese And Western Medicine Treatment Of Hemorrhagic Stroke Based On U-Net Convolutional Neural Network Mode

Purpose:Based on Cite Space software,this article conducts a visualization analysis of the research hotspots of Western and traditional Chinese medicine treatments for perihematomal edema in hemorrhagic stroke,explores new research directions,and establishes an automatic segmentation model for hematoma based on U-Net convolutional neural network to verify the accuracy of the model.It provides a reference for the application of artificial intelligence technology in the field of traditional Chinese medicine,reviews the clinical efficacy of traditional Chinese medicine in promoting neurological function recovery in the perioperative period of hemorrhagic stroke,and evaluates the safety of traditional Chinese medicine.At the same time,the hematoma measurement model in cerebral hemorrhage is applied in clinical practice to evaluate its clinical value in traditional Chinese medicine.Material and method:1.Paper I:Data were retrieved from the Web of Science Core collection database,China National Knowledge Infrastructure(CNKI),Wanfang Database,VIP database and China Biology Medicine Disc(CBM),and the retrieval time range was from the establishment of the database to December 31,2022.Literature on topics related to cerebral hemorrhage and perihematoma edema was retrieved.Citespace software is used to map and analyze the information of countries,institutions,magazines,authors and keywords.2.Paper II:From January 2017 to December 2021,a total of 7659 CT image slices of perioperative patients with intracerebral hemorrhage were included.Python language was used to build a model through U-Net convolutional neural network,train and test the model.The Accuracy of the model was evaluated by observation accuracy,Dice,Precision,Recall,m Io U,and m PA.3.Paper III:This study was a retrospective cohort study.The clinical data of 146 patients with acute hemorrhagic stroke who underwent craniotomy or minimally invasive surgery in the Affiliated Hospital of Liaoning University of Traditional Chinese Medicine from January 1,2017 to December 31,2021 were collected to compare the clinical efficacy.The patients treated with TCM acupuncture and TCM combined with conventional western medicine during the perioperative period were included in the experimental group,and the patients treated with conventional western medicine were included in the control group.The volume of intracranial hematoma,Glasgow coma Scale(GCS)score,National Institutes of Health Stroke Scale(NIHSS)score,activity of daily living scale(Barthel)score,Hamilton depression score(HAMD),CT hematoma volume,intracranial pressure and blood pressure were collected before and after treatment in the two groups.The total effective rate was calculated.Use U-Net convolutional neural network model for clinical evaluation of hematoma and peripheral edema volume.Statistical methods were used to analyze the data of the two groups and compare the changes of scores and efficacy of the two groups before and after treatment.Results:1.Paper I:A total of 1946 papers were included,and the number of published papers increased rapidly in the past 20 years,with outstanding research popularity.160 journals were involved,2230 authors were involved,5 core authors were involved,and 82 papers were published by core authors.The first five authors are Schwab,Stefan,Hanley,Daniel F,Huttner,Hagen B,Staykov,Dimitre,Doerfler,Arnd.The number of publications is the largest in the United States,China,Germany,Australia and Canada.The United States plays a leading role in this field of research,followed by China in this field of research,but it still needs to strengthen the cooperation between countries.Among the top 10 institutions,the University of Nuremberg has the most total citations,with 11.3 average citations,19 total publications,and 0.11 centrality;Johns Hopkins University has 18 total publications,0.23centrality,and 20.14 average citations.The key words cited with high frequency were 259 cases of cerebral hemorrhage,208 cases of perihematoma edema,64 cases of cerebral edema,63 cases of stroke,55 cases of injury,and 50 cases of volume.Early intensive hypotension,blood sugar control,reduce the growth of edema around hematoma,CT auxiliary software semi-automatic segmentation,image segmentation,artificial intelligence are promising techniques for CT diagnosis of edema around cerebral hemorrhage.2.Paper II:U-Net model test results:ROI region Dice 88.9%;intersection ratio was 80.39%;The recall rate was 86.64%.The accuracy was 91.76%,and the average intersection ratio was90.06%.The average accuracy was 93.27%.After clinical prediction,the accuracy of manual segmentation was 97.7%in the U-Net group and 90.8%in the control group,x~2=35.246,P<0.05,which was statistically significant.Compared with Tada formula volume measurement,there was no significant difference between U-Net group and Tada formula group(Z=-0.189,P>0.05).It shows that the model has high accuracy and can be used for hematoma volume labeling in clinical hemorrhagic stroke.3.Paper III:The changes of scores in the two groups before and after treatment:the NIHSS score,the Barthel score and the volume of cerebral hematoma in the two groups after treatment were significantly improved compared with those before treatment;The improvement of NIHSS score,GCS score and HAMD score in the experimental group were significantly better than those in the control group.There was no significant difference in the volume of intracerebral hematoma and Barthel score between the experimental group and the control group after treatment(P>0.05).There was no significant difference between the manual and automatic hematoma volume assessment methods(P>0.05).Clinical efficacy evaluation:in the experimental group,20 cases were markedly effective,39 cases were effective,14 cases were ineffective,and the total effective rate was 80.8%.In the control group,3 cases were markedly effective,15 cases were effective,and 11 cases were ineffective,with a total effective rate of 62.1%.After rank sum test analysis,the efficacy of the two groups was statistically significant(P<0.05).Adverse reactions:during the treatment,there were no significant differences in the incidence of stroked pneumonia, tracheotomy rate,cerebrospinal fluid leakage rate,wound healing grade,heart failure,lower extremity deep vein thrombosis,bedsore and other complications between the experimental group and the control group(P>0.05).There was no significant difference in the mortality rate and the incidence of cerebrospinal fluid leakage complications between the craniotomy group and the minimally invasive surgery group(P>0.05).The overall mortality of the included patients was significantly correlated with admission GCS score,NIHSS score,and preoperative hematoma volume(P<0.05).This study found that the top three TCM syndrome differentiation types in the perioperative period of hemorrhagic stroke were phum-heat blocking orif syndrome(33 cases),liver-yang hyperactivity syndrome(21 cases),Yin exhaustion and Yang removal syndrome(18 cases).The analysis within the traditional Chinese medicine treatment group found that electroacupuncture treatment was more effective than conventional acupuncture in improving the NIHSS score and Barthel score of cerebral hemorrhage during perioperative period(P<0.05),and electroacupuncture treatment did not increase the risk of depression,and there was no significant difference in the effect of increased intracranial pressure compared with the acupuncture group(P>0.05).Conclusion:1.Hemorrhagic stroke-induced perihematomal edema is currently a focal point of research,focusing on its mechanisms and the development of corresponding therapeutic drugs.Early intensified blood pressure control,blood glucose management,and minimally invasive techniques remain effective measures to reduce edema.Semi-automatic CT-assisted software and computer artificial intelligence technologies will play a positive role in monitoring perihematomal edema in severe hemorrhagic stroke.2.Clinical applications of AI technology,such as the automatic segmentation model for hematoma images based on the U-Net convolutional neural network,have demonstrated the feasibility of AI in traditional Chinese medicine clinical assessments.3.Traditional Chinese medicine treatment during the perioperative period of hemorrhagic stroke can promote patients’neurological recovery,improve depressive symptoms,and is safe and effective.The main TCM pattern in hemorrhagic stroke is"Phlegm-Heat Obstructing the Orifices."Risk factors for mortality include concomitant heart failure,renal dysfunction, preoperative hematoma volume,and admission GCS and NIHSS scores.