Photoshop-assisted Endoscopic Ultrasonography For The Identification Of Gastrointestinal Stromal Tumors And Leiomyomas

Objective:Gastrointestinal stromal tumors（GIST）and leiomyomas（LM）under endoscopic ultrasound（EUS）are susceptible to confusion,and EUS alone cannot accurately distinguish them.Therefore,in this study,we introduced Photoshop technology,and used Photoshop software to convert the subjective judgment indicators of echogenic endoscopic images of gastrointestinal stromal tumors and leiomyomas—echo intensity and homogeneity of echoes into objective data for comparison.A new way to distinguish between the two.Method and materials:Gastrointestinal submucosal tumors（SMT）were collected from the Department of Gastroenterology of the Second Hospital of Tianjin Medical University from April2013 to June 2019,and endoscopic or surgically resected.Immunohistochemical diagnosis of GIST or LM,and exclusion of incomplete data records eventually included 71 cases（26 cases of GIST,45 cases of LM,23 cases of esophagus,2 cases of cardia,43 cases of stomach,2 cases of duodenum,1 colorectum example）.Retrospective study of lesions with complete data records,including indicators of gender,age,and EUS（diameter,location,level of origin,echo,shape,defoliation,surface condition,growth pattern,smooth edges,and calcification）;The differences between EUS and multi-slice spiral CT（MSCT）in the diagnosis of two types of SMT were compared.Photoshop CS6 software was used to measure the average gray value of the EUS image（representing the echo intensity,the stronger the echo intensity,the higher the gray average value）and the standard deviation（representing the degree of echo uniformity,the worse the degree of echo uniformity,the higher the standard deviation of gray level high）.The EUS image of each lesion was measured 5 times to reduce the error,and the average value was taken and expressed as the mean gray value（Tmean）and the standard deviation(T_SD).The characteristics of gray Tmean and T_SDof GIST/LM in all parts and GIST/LM only in the stomach were analyzed.The receiver operating characteristic curve（ROC）was used to evaluate the gray Tmean and T_SDindicators to identify GIST and LM values.This study was a single-center retrospective study.SPSS 22.0 was used to process the data and Graph Pad Prism 7 was used to draw the chart.Measurement data are expressed as mean±standard deviation（（?）±s）,and count data are expressed as percentage（%）.The%comparison between the groups was tested byχ2 test or Fisher exact test,and the mean comparison was performed by t test（in line with normal distribution）or non-parametric rank sum test.P<0.05 was considered statistically significant.Result:1.Distribution of GIST and LM lesions:45 cases of LM were distributed in 21cases of esophagus,2 cases of cardia,20 cases of stomach,1 case of duodenum and colorectum,18 cases were male and 27 cases,and the average age was（54.67）.±9.71)years old.GIST was distributed in 2 cases of esophagus,23 cases of stomach,and 1case of duodenum.There were 10 and 16 cases of male and female,with an average age of（62.62±11.56）years.2.The diagnostic value of EUS for GIST and LM:According to the comparison with postoperative pathology,the sensitivity,specificity,positive predictive value,negative predictive value,and accuracy rate of EUS for LM diagnosis were found to be between 56.8%-96.3%.The sensitivity,specificity,positive predictive value,negative predictive value,and accuracy of EUS in the diagnosis of GIST range from55.8%to 92.9%.The accuracy rate of qualitative and localized diagnosis of both types of SMT by EUS is higher than that of MSCT（60.7%vs 17.9%,100%vs51.8%）.Under EUS,the differences between GIST and LM in terms of age of onset,location of lesions,level of origin,echo,and calcification were statistically significant（P<0.05）;in diameter,growth pattern,surface condition,shape,defoliation,and smooth edges There was no significant difference（P>0.05）.3.GIST and LM grayscale Tmean and T_SDfeatures:GIST grayscale Tmean is higher than LM（32.193±8.294 vs 20.867±11.449,P=0.000）,grayscale T_SDis also higher than LM（12.180±4.307 vs 7.326±3.187,P=0.000）.The same method was used to compare the grayscale Tmean and T_SDcharacteristics of GIST（23cases）and LM（20 cases）located only in the stomach,and found that the grayscale Tmean of gastric GIST was higher than that of gastric LM（31.300±8.370 vs16.170±8.930,P=0.000）,the grayscale T_SDis also higher than the stomach LM（11.700±4.220 vs 5.820±2.910,P=0.000）.4.ROC curve to evaluate the differential diagnosis value of gray Tmean and T_SDfor GIST and LM:（1）The area under the ROC curve（AUC）of gray Tmean and T_SDto distinguish GIST and LM are 0.821（95%CI 0.723-0.918）and 0.817（95%CI 0.718-0.913）,respectively.The sensitivity and The critical value corresponding to the highest specificity(Tmean=33.281,T_SD=10.103).When Tmean≥33.281 or T_SD≥10.103,the sensitivity,specificity,positive predictive value,negative predictive value,and accuracy rate of GIST from LM were between 69.0%-85.7%.（2）In the same method,the ROC curves of gray Tmean and T_SDfor gastric GIST（23 cases）and gastric LM（20 cases）were performed.The AUC was 0.893（95%CI0.796-0.991）and 0.891（95%CI 0.799-0.983）,select the best critical value(Tmean=22.298,T_SD=10.048).When Tmean≥22.298 or T_SD≥10.048,the sensitivity,specificity,positive predictive value,negative predictive value,and accuracy of identifying gastric GIST from gastric LM are between 80.0%-91.3%.Conclusion1.EUS alone has a certain misdiagnosis rate when it is used to diagnose GIST/LM.2.EUS has higher positioning and qualitative accuracy than MSCT for both SMTs.3.Photoshop software assisted EUS has certain value in distinguishing GIST and LM,especially for gastric GIST and gastric LM.GIST echo intensity is stronger than LM,and echo uniformity is worse than LM.This method is a non-invasive and economical inspection method,and has certain clinical promotion value.