The Research Of Individualized Liver Segmentation Assisted With Computer

PurposeAt present,there are many researches on liver segmentation,and there are many different segmentation methods and theories.Because the liver has double blood supply systems and the degree of vascular variation is complex,how to segment the liver according to the vascular distribution of patients has become the focus of research.Now,there are few studies on computer-assisted liver segmentation.Therefore,this research is to develop a new computer-assisted individualized liver segmentation method.Using living human CT data to reconstruct the liver and intrahepatic ducts and build the branched models of hepatic vein and portal vein.Based on the portal vein drainage area,the nearest neighbor region approximate segmentation(NNSA)was used to determine the three-dimensional distribution range and adjacent boundary of the blood flow in the tertiary branch of the portal vein.The main hepatic vein and its main branches were used to calibrate the range of hepatic segment in the portal vein drainage area so as to realize individualized hepatic segment.And through regular segmental/lobectomy to compare the consistency between the volume of the actual segmental/lobectomy and the volume of the simulated segmental liver resection.Meanwhile,intraoperative ICG real-time fluorescence imaging of the liver segment/lobe was used to observe the liver plane seen during the two liver segmental methods.Finally.a new computer-assisted individualized liver segmentation method was developed,which combined the blood flow of the third branch of the portal vein with the main hepatic vein.Method(1)data collection and preprocessing:(1)20 cases of upper abdominal CT examination of 64 or 256 layers from September 2016 to December 2018 in zhujiang hospital of southern medical university were collected.(2)clinical data of 27 patients undergoing regular hepatectomy in hepatobiliary department 1 of zhujiang hospital from March 2017 to January 2019 were collected,including 18 males and 9 females.All patients underwent enhanced CT scanning of the upper abdomen,and thin layer data with a thickness of 1.0mm were collected at the four stages of plain scan,arterial phase,portal vein phase and hepatic vein phase.Thin slice CT data were imported into MI-3DVS and Smart Vision 3DVWorks,respectively,for liver,artery,portal vein,hepatic vein and lesion segmentation and three dimensional reconstruction,and the segmentation based on portal venous drainage was performed.On this basis,the hepatic segment boundary of the portal venous drainage area was calibrated using the plane of the hepatic vein trunk and its main branches.Using the new segmental method for surgical planning,determining the excision plane,and the volume of pre-excised liver segment/lobe and the volume of retained liver segment/lobe were measured.Anatomic hepatectomy was performed by ICG fluorescence staining and following the hepatic vein direction.The volume of tumor bearing liver was determined by drainage.Statistical analysis:mean±standard deviation was used to indicate the degree of difference,which t-test was used to compare the differences in liver segment volumes between the two liver segmentation methods,and the correlation between the two liver segmentation methods was compared by Person correlation test.The Person test and Bland-Altman method were used to analyze the correlation and consistency between the simulated preoperative resection volume and the actual intraoperative resection volume of the tumor-bearing liver.Result(1)Three-dimensional reconstruction and individualized subsection:all patients successfully underwent three-dimensional reconstruction,which can clearly show the spatial relationship between liver,gallbladder,artery,portal vein,hepatic vein and lesion;The boundaries of hepatic segments base on portal vein drainage area were irregular and wavy.And the new liver segmenttion method show a wavy between the upper and lower hepatic segmental boundaries and the left and right hepatic segmental boundaries were irregular curved.(2)Surgical results:all the 27 patients underwent anatomic hepatectomy,including 7 cases of left lateral lobe resection,6 cases of right hemihepatectomy,4 cases of right posterior lobe resection,and 10 cases of left hemihepatectomy.The average operative time was 354.37±113.92min.The mean blood loss was 316.29±148.59mL,and the mean postoperative hospital stay was 8.6±4.12d.(3)Statistical analysisThe hepatic segment volumes base on the portal vein drainage area and the new liver segmentation method were statistically analyzed,P>0.05.There was no statistically significant difference in the two liver segmentation methods.The Person correlation coefficient was 0.85-0.98,and the two liver segmentation methods had good consistency.The mean tumor bearing segment volume of simulated resection before surgery was 426.66±217.65mL,and the actual mean tumor bearing segment volume was 355±223.48mL by drainage method after surgery.Through the correlation test,the Person correlation coefficient r=0.960.The bland-altman plot shows that 24 of the 27 points fall within the 95%confidence interval.ConclusionIn this study,a new computer-assisted individualized liver segmentation method was developed that took into account both the third branch of the portal vein drainage area and the hepatic vein trunk plane.It not only took into account the upper and lower hepatic segment boundaries of the hepatic segmentation method in the portal vein drainage area,but also took the hepatic vein(hepatic fissure)as the boundary between the left and right adjacent hepatic segments in the liver parenchyma.Moreover,it has the function of accurate volume measurement,which is more consistent with the intraoperative anatomy of clinical practice,and this new method provides a new evaluation technology for the implementation of clinical anatomic hepatectomy.