Y Li1, M M Li, Y Zhang, J L Cheng, Z G Shang, C X Bu. 1. Department of Magnetic Resonance, First Affiliated Hospital of Zhengzhou University, Zhengzhou 450052, China.
Abstract
OBJECTIVE: To analyze magnetic resonance images of the posterior cranial fossa tumors, and evaluate the texture characteristics of three kinds of children's posterior cranial fossa tumors by support vector machine (SVM). METHODS: From May 2008 to August 2015, a total of 76 patients with children's posterior cranial fossa tumors in First Affiliated Hospital of Zhengzhou University were retrospectively analyzed, texture analysis was performed on the three types of tumors. RESULTS: The 5 texture parameters extracted from the gray level co-occurrence matrix had at least 3 parameters which had statistically significant difference between the two different classes of data sets (P<0.05). For all kinds of data sets, the variance parameters of gray value of ROI region had statistically significant differences.The accuracy rate of SVM test in the remarkable difference of the ependymomas and medulloblastomas, the ependymomas and astrocytomas and the medulloblastomas and astrocytomas was 86.15%±4.16%, 73.63%±5.82% and 74.32%±5.85%, respectively. CONCLUSION: The analysis of texture features can provide more quantitative information which could provide a new thought and method for the differential diagnosis of tumors in the posterior cranial fossa.
OBJECTIVE: To analyze magnetic resonance images of the posterior cranial fossa tumors, and evaluate the texture characteristics of three kinds of children's posterior cranial fossa tumors by support vector machine (SVM). METHODS: From May 2008 to August 2015, a total of 76 patients with children's posterior cranial fossa tumors in First Affiliated Hospital of Zhengzhou University were retrospectively analyzed, texture analysis was performed on the three types of tumors. RESULTS: The 5 texture parameters extracted from the gray level co-occurrence matrix had at least 3 parameters which had statistically significant difference between the two different classes of data sets (P<0.05). For all kinds of data sets, the variance parameters of gray value of ROI region had statistically significant differences.The accuracy rate of SVM test in the remarkable difference of the ependymomas and medulloblastomas, the ependymomas and astrocytomas and the medulloblastomas and astrocytomas was 86.15%±4.16%, 73.63%±5.82% and 74.32%±5.85%, respectively. CONCLUSION: The analysis of texture features can provide more quantitative information which could provide a new thought and method for the differential diagnosis of tumors in the posterior cranial fossa.