Automatic Quantification Of Aneurysmal Subarachnoid Hemorrhage Based On The DeepLabv3+ Model And The Prognostic Value Of Bleeding Volume In Predicting Prognosis

Objective:Aneurysmal subarachnoid hemorrhage(a SAH)is a cerebrovascular disease with high morbidity and mortality.The amount of bleeding after aneurysm rupture has been identified as an important factor in judging the severity of the patient ’s condition.Therefore,this paper intends to use the DeepLabv3+ network model to automatically quantitatively analyze the correlation between a SAH bleeding volume and short-term prognosis of patients,and to preliminarily explore the value of bleeding volume in the evaluation of a SAH patients.Methods:A total of 177 consecutive patients with a SAH admitted to the Affiliated Hospital of Yangzhou University from August 2015 to December 2021 were retrospectively collected.The patient ’s head computed tomography(CT)image data(DICOM format)were processed by two neuroimaging physicians who were skilled in using ITK-SNAP 3.8software,and the bleeding area was manually marked and divided.Finally,another expert physician verified and corrected the results of the specialist ’s treatment.The clinical data of the patients were collected,including name,gender,age,blood pressure at admission and past medical history.Two experienced specialists independently used the Hunt-Hess grading scale to assess the clinical status of patients at admission,the modified Fisher scale to assess the amount of subarachnoid hemorrhage at admission,and the glasgow outcome scale(GOS)to assess the prognosis of patients 3 months after discharge.When two physicians ’ evaluation results are inconsistent,another expert physician will make a final judgment on the controversial results.According to the GOS score,the patients were divided into two groups.The GOS score ≤ 3 was defined as the poor prognosis group,and the GOS score > 3 was defined as the good prognosis group.The dice similarity coefficient(DSC)was used to evaluate the accuracy of the model in segmenting lesions.The Kolmogorov-Smirnov test was used to analyze the normality of the data,and the independent sample t test or Mann-Whitney U test was used to analyze the differences in variables between the two groups.Spearman was used to analyze the correlation between the amount of bleeding,modified Fisher grade and Hunt-Hess grade and the prognosis outcome GOS scale.The receiver operating characteristic(ROC)curve was drawn to analyze the prognostic predictive ability of bleeding volume,modified Fisher grade and Hunt-Hess grade,and to provide the cut-off value of bleeding volume closely related to poor prognosis.P < 0.05 was considered statistically significant.Results:The Dice value of DeepLabv3 + network model segmentation method is 0.91,and the time of segmenting a single image is 0.037 s.The admission blood loss,modified Fisher grade and Hunt-Hess grade in the poor prognosis group were higher than those in the good prognosis group,and the differences were statistically significant(P < 0.05).The amount of bleeding was negatively correlated with GOS score at 3 months after discharge(r =-0.332,P < 0.05).Modified Fisher grade was negatively correlated with GOS score 3 months after discharge(r =-0.189,P < 0.05).At the same time,Hunt-Hess grade was also negatively correlated with GOS score(r =-0.398,P < 0.05).The area under the curve of blood loss to predict adverse outcomes in patients with a SAH was0.741(95 % CI,0.670,0.804).The area under the curve of the modified Fisher grade for predicting adverse outcomes was 0.644(95 % CI,0.568,0.714).The area under the curve of Hunt-Hess grade for predicting adverse outcomes was 0.727(95 % CI,0.655,0.791).The optimal cut-off value of bleeding volume was 25 ml,the sensitivity was 66.0 %,and the specificity was 78.6 %.Conclusions:It is feasible to use the DeepLabv3+ network model to quickly and accurately quantify the bleeding volume of a SAH on CT.The amount of bleeding,modified Fisher grade and admission Hunt-Hess grade were associated with the prognosis of a SAH.Compared with the qualitative assessment scale,the quantitative assessment of the prognostic value of the amount of bleeding is higher,and it is more instructive for the judgment of the condition and prognosis of patients with a SAH.When the amount of bleeding in patients with a SAH is found to be ≥ 25 ml,early intervention should be carried out to reduce the risk of poor prognosis.