Methodfor accurate calculation of early residual tumors after surgery of low-grade gliomas

Ref: Scherer M,et alJNeurosurg2018 Jul 27:1-9doi: 10.3171/2018.2.JNS172951Thethe maximum safe removal of gliomas is an ideal goal for surgeryFor low-grade gliomas (LGGs) at the non-enhanced WHO class II, tumor boundaries are often depicted through the FLAIR high signal area of preoperative MRI (preOP MRI)Because surgical injury can cause ischemia in brain tissue, the FLAIR high-signal area of early MRI (epMRI) within 72h after surgery often overestimates residual tumors; The FLAIR high signal area range of follow-up magnetic resonance (fuMRI) 3 months after surgery was associated with the patient's prognosisMoritz Scherer of Neurosurgery at Heidelberg University Hospital in Germany, among others, assessed the survival prognosis of patients with residual tumors after LGG's surgery through an early postoperative MRI, published in The July 2018 issue of J Neurosurgthe retrospective study included 43 WHO II-grade LGG patients who were mrI scans before surgery, within 72h early after surgery, and during follow-up periods of 3-4 months after surgery, to obtain FLAIR sequences and ADC imagesFirst, the tumor tissue in the high signal area of flAIR sequence image of three points in time is artificially semi-automatically depicted, and the tumor volume is calculatedIn postoperative epMRI, the FLAIR sequence is compared with the ADC image, and the tumor range of the high signal area depicted by the FLAIR sequence is automatically corresponding to the ADC imageThe results of the study, which distinguish tumors from normal brain tissue according to ADC values, have been more accurate, so the authors used the maximum expectation (expectation, EM) calculation to make a probabilistic segment of the corresponding areas in the ADC image, into three parts of residual tumor, ischemic lesions and normal brain white matter, and calculate the volume and average ADC values of each partThe volume of residual tumor seisicity obtained by probability segmentation in epMRI image was analyzed in correlation with the volume of residual tumor seist in the flAIR sequence high signal region in the fuMRI image (Figure 1)Figure 1 A.preOP MRI's FLAIR sequence image depicts tumor area; B.epMRI FLAIR sequence and ADC image registration; C.epMRI FLAIR high signal region; D FLAIR sequence high epMRI Signal region projected into ADC image; E.fuMRI's FLAIR sequence shows residual tumors (red arrow) and cerebral ischemic softening stove (green arrow); F ADC value analysis of epMRI's FLAIR sequence high signal projection region and make probability segmentation, black line for the original ADC histogram, blue line for the original ADC histogram transformed from the cluster Gaussian curve, green Gauss ian curve for low ADC value ischemia, red Gaussian curve for high ADC value Residual tumors, the gray Gaussian curve is normal brain white matter, C.1 projects the results of probability segmentation on the FLAIR sequence of epMRI, and D.1 Projects the results of probability segmentation on the ADC image of epMRI results show that the tumor range is measured according to the FLAIR sequence high signal region, the average volume of LGG before surgery is 32.8 to 27.0 ml, the epMRI image is 19.4 to 16.5 ml and the fuMRI image is 8.4 to 10.2 ml The average tumor volume difference of 11.0 to 10.6 ml in the FLAIR serial high signal region of epMRI and fuMRI is statistically significant (p.0001) After the proticus segmentation of epMRI images according to ADC values, the tumor volume was 7.6 to 10.2 ml, accounting for 32%, the volume of ischemia was 8.1 to 5.9 ml, accounting for 48%, and the normal brain tissue volume was 3.7 to 4.9 ml, accounting for 20% (Figure 2) After the probability segmentation, the tumor volume of epMRI image and the high signal area of the FLAIR sequence of fuMRI were obtained, and the tumor volume difference was smaller, with an average of 0.8 to 3.7 ml, which was not statistically significant (p-0.16) The correlation between tumor volume after proprobability segmentation of epMRI image and tumor volume of flAIR high signal of fuMRI was analyzed, and the Correlation coefficient of Pearson was 0.96 (P 0.0001) and the consistency correlation coefficient was 0.89 The average ADC value in the three parts after the split was calculated in the epMRI image, and combined with the average ADC ratio of the normal frontal lobe region of the tumor to the side, it was worth the ADC ratio The three parts of the ADC ratio were 1.48 to 0.46 residual tumor, normal brain tissue 1.11 to 0.31, ischemia 0.76 to 0.23, and there was a significant difference between the three (p.0001) The results are consistent with the conclusion that the ADC value of tumor tissue in the previous reports is higher than the ADC value of the ischemia, which shows that the EM calculation method is feasible for probability segmentation Figure 2 Box diagram for quantitative analysis of tumor volume The tumor volume of preOP MRI, epMRI and fu MRI, as well as the tumor volume (Clustered epMRI) obtained by probabilistic segmentation of epMRI, are calculated by the FLAIR sequence high signal region the study found that the volume of residual tumor suspent on the high signal evaluation of residual tumor seislum through FLAIR sequence early after the removal of non-reinforced LGG, and the use of EM calculation to carry out probability segmentation of the FLAIR high signal region according to The ADC value, which can reduce the interference of the ischemia after surgery, obtain a more accurate residual tumor range, and provide a basis for patient survival prediction, postoperative radiotherapy, and evaluation of LGG progress (Professor Wang Zhiqiu , of Huashan Hospital, Fudan University, is reviewed by Dr Yang Jia
of the Department of Neurosurgery, Marburg University, Germany, and related links to Professor Wang Zhiqiu, editor-in-chief of The Sun's Outside Information, and Professor Chen Jicheng , affiliated with Fudan University.