| Objective:Left atrial appendage closure(LAAC)has been increasingly used for patients with non-valvular atrial fibrillation to prevent thromboembolism.The structure of the left atrial appendage is complex and changeable.The accurate measurement of the ostium and depth of the left atrial appendage is the key to the selection of the model of occlusive device,which is related to the success or failure of LAAC.In this study,manual measurement,intelligent measurement based on Cardiac computed tomography(CCTA)images and Digital subtraction angiography(DSA)were compared.Three methods were used to measure the ostium and depth of the LAA,and their correlation,consistency and guiding significance for the selection of occlusive device models were analyzed.Meanwhile,the evaluation results of PDL,DRT and re-endothelialization after LAAC surgery were compared with the manual evaluation and intelligent evaluation of postoperative CCTA imaging,to explore the application value of Artificial intelligence(AI)based left atrial appendage CT image analysis in LAAC.Methods:A total of 69 patients who underwent pre-operation LAAC assessment in the Department of Cardiology,Jinan Central Hospital from January 1,2020 to February 28,2022 were selected as the research objects,and their clinical and imaging data were collected for retrospective analysis.Conventional CT image analysis software was used to manually measure the left atrial appendage for preoperative CCTA images of all patients,while ANYTHINKβ left atrial appendage intelligent analysis system was used to measure the left atrial appendage intelligtively,and the occluder model was predicted according to the measurement results.Intraoperative DSA measurements were collected and the final occluder model was implanted.The ostium and depth results of CT manual measurement,CT intelligent measurement and DSA measurement methods were compared to analyze their correlation and consistency and guiding significance for the selection of sealer type,P<0.05 showed statistically significant difference,the correlation coefficient |r|≥0.8 is highly relevant,|r|≤0.3 is irrelevant.All patients were followed up with CCTA.Two kinds of CCTA image analysis software were used for manual evaluation and intelligent evaluation to evaluate the postoperative efficacy of LAAC(whether there was periinstrumental leakage,whether there was instrument-related thrombus,whether reendothelialization was complete,and whether there was occluder displacement or loss).Results:(1)General information of patients:A total of 69 patients underwent preoperative evaluation,and 67 patients were successfully implanted with WATHMAN occluder,with a success rate of 97.10%.Among the 69 patients with three-dimensional reconstruction of CCTA,the most morphological classification was chicken wing type(44 cases,63.77%).Cauliflower type was the least,with 6 cases accounting for 8.70%.There were 9 cases of weathervane type and 10 cases of cactus type,accounting for 13.04%and 14.49%,respectively.(2)Preoperative imaging measurement results:the maximum diameter of LAA ostium was measured by three methods,and the manual CT measurement value was the largest,with an average of 26.38±4.19mm.Followed by CT intelligent measurement,average 24.66±4.06mm;DSA was the smallest,with an average of 24.61±4.21mm.The results of the three methods were significantly correlated with each other(P<0.05).The results of CT manual measurement and DSA measurement were different,and the results of CT manual measurement and CT intelligent measurement were different,and the difference was statistically significant(P<0.05).The results of CT intelligent measurement and DSA measurement had no significant difference,P>0.05.The LAA depth was measured by three methods,and the DSA value was the largest,with an average value of 24.26±4.45mm.The second was intelligent CT measurement with an average of 23.43±4.36mm.CT manual measurement results were between 15.4mm-23.7mm,with an average value of 23.35±4.30mm.The correlation coefficients of the three methods were all<0.05,indicating a significant correlation.There was no significant difference in LAA depth between CT manual measurement and CT intelligent measurement(P=0.50),but there were differences between CT intelligent measurement and DSA measurement,CT manual measurement and DSA measurement,and the difference was statistically significant(P<0.05).(3)Comparison between predicted occluder model and actual implantation:67 patients were successfully implanted with the WATCHMAN sealer.The size of the sealer was predicted according to the ostium size and depth measured by manual CT and intelligent CT.The average size predicted by manual CT was 29.46±3.06mm,and the average size predicted by intelligent CT was 28.97±3.30mm.The average size of implanted occluder was 29.15±3.39mm.There was a good correlation between artificial prediction and actual implantation(r=0.86),and there was a good correlation between intelligent prediction and actual implantation(r=0.96).The difference between artificial prediction and actual implantation was statistically significant(P=0.17),while there was no significant difference between intelligent prediction and actual implantation(P=0.58).(4)PDL evaluation:After 45 days of follow-up,22 cases of PDL were found by manual evaluation and 22 cases by intelligent evaluation.There was no significant difference in the detection rate of PDL between the two detection methods(P<0.05).The 65 patients were followed up 1 year after surgery,and 19 cases of PDL were found by manual evaluation and 19 cases by intelligent evaluation.There was no significant difference in the detection rate of PDL between the two detection methods(P<0.05).(5)NE evaluation:Of the 45 patients without PDL followed up 45 days after surgery,25 were completely reendothelialized,accounting for 55.55%;of the 46 patients without PDL followed up 1 year after surgery,39 were completely reendothelialized,accounting for 84.78%.Conclusions:(1)The AI-based left atrial appendage CT image analysis technology can accurately measure the ostium and depth of the left atrial appendage before surgery,which has a good correlation with contrast measurement,and can better guide the selection of occlusive device model.(2)The AI-based left atrial appendage CT image analysis technology can better evaluate the PDL and NE status after LAAC operation.(3)The left atrial appendage CT image analysis technology based on AI has high application value in the preoperative and postoperative evaluation of LAAC. |
PDF Full Text Request