Texture Characteristics Of Articular Cartilage Of Patients With Articular Cartilage Osteoarthritis Based On MR T2 Mapping Images

Background and purposeKnee is one of the major joints of human body,and also one of the most important joints,So it is easy to degeneration.Articular cartilage plays a key role in the knee joint,such as cushion pressure.It's function is to reduce joint friction.Articular cartilage is very important in maintaining the function and activity of the knee.The degeneration of articular cartilage has been proved to be an important sign in the development of OA.Osteoarthritis(OA)has been defined as a heterogeneous group of conditions that leads to joint symptoms and signs which are associated with defective integrity of articular cartilage,in addition to related changes in the underlying bone at the joint margins.Therefore,The morphological changes and the changes of internal components were found in the clinical work,which is helpful to understand the degree of knee joint degeneration and the development process of the disease.T2 mapping was sensitive to the content changes of articular cartilage and can be used to evaluate the articular cartilage situation of OA patients.Texture has been used to analyze medical image in recent years.There is a relationship between the texture feature of a certain position and the gray variation around the position,It is mainly based on the change of the gray level to distinguish the change of tissue composition,is a kind of visual perception of the local image features.In this study,we furtherly evaluated the OA by combining the T2 mapping and the texture characteristics extracted from the T2 mapping images within 7 separated regions of articular cartilage.Besides,based on T2 value and texture parameters,multiple classifiers were selected to furtherly differentiate the OA patients from normal controls.Materials and methodsThe study was approved by the Zhengzhou University First Affiliated Hospital's institutional review board.Forty-four patients(aged 46.5±9.8 years old;14 males and 30 females)with OA and Thirty-nine healthy volunteers(aged32.5±9.3years old;13 males and 26 females)were recruited in this study.All subjects were evaluated by using the Western Ontario and McMaster Universities Osteoarthritis Index(WOMAC)for OA severity.Each subject underwent a MR scanning by using a 3T MRI scanner(Ingenia,Philips Healthcare,Best,the Netherlands)with an 8-chanel knee receiver coil.A multi-echo TSE-based sequence was performed for T2 mapping image calculation with the below protocol parameters: multi-slice sagittal acquisition,6 echos with first TE=10ms and delta TE=10ms,TR=2000ms,FA=90°,FOV=100mm×100mm,voxel size=0.5mm×0.5mm,slice thickness=2.5mm,no slice gap,sense factor=2,fold-over direction=AP,totally 40 slices were scanned within 12 mins.T2 values were extracted from 7 separated articular cartilage ROIs(Regions of Interest)(as shown in figure 1)and the texture parameters(based on histogram,co-occurrence matrix,run-length matrix,absolute gradient)were extracted within the 7 regions by using Mazda software.The ROIs are as follows: the anterior part of the femur,the anterior horn of the meniscus,the medial part of the meniscus,the posterior horn of the meniscus,the anterior horn of the meniscus,the medial part of the meniscus,and the posterior horn of the meniscus.The most complete level of cartilage was selected in the sagittal T2 mapping sequence,MazDa(version 4.7,The Technical University of Lodz,Institute of Electronics,http://www.eletel.p.lodz.pl/mazda/)software was used to delineate the cartilage in different regions along the edge of cartilage,and texture parameters were extracted for each of the 7 regions of interest,Finally,77×7 parameters were obtained for each subject.Two-sample T-test was carried out to compare the difference between OA patients and normal controls for each parameter and for each ROI.The parameter within a specific ROI which showed a significant difference according to the Student's T-test was selected to form a feature vector for each subject.Classification of the subjects based on feature vectors were performed by using Artificial Neuro Network(ANN),Random Forest(RF),Naive Bayes(NB)and Linear Discriminant Analysis(LDA)methods.ResultsFinally,83 patients were enrolled in this study,including the normal control group of 39 cases,OA patients in 44 cases.There was no significant difference in gender and BMI between the two groups(P > 0.05),There were significant differences in age,WOMAC and LEFS between the two groups(P < 0.05).Sum of 7 regions,the mean T2 was 52.99±4.61 ms in patients and was 43.74±3.82 ms incontrols(P<0.05).In OA group,the most vulnerable parts of the knee are ROI4,ROI3 and ROI1,and the T2 value are 67.89±10.57 ms,60.02±9.02 ms and 54.96±9.31 respectively.For all the 7 ROIs,the texture parameters showed a significant difference(p<0.05).Some texture features extracted from ROI1,ROI2,ROI4,ROI5 and ROI6 showed a significant difference(P<0.05).Texture parameters within ROI5 showed the most significant difference.The results of classification using different classifiers were ANN=79.25%,RF=73.58%,NB=69.80%,LDA=77.36%.The ANN showed the best performance with a high classification accuracy compared to other classifiers.The correlation analysis between T2 mapping and WOMAC and LEFS scale showed that there was a strong correlation between T2 values and WOMAC(P=0.00).T2 values and LEFS have strong correlation(P=0.00).There was no significant difference between the two groups of T2 values.Age was correlated with T2 values of ROI 5.ConclusionsThe T2 values of knee cartilage in patients with OA was higher than that in healthy adults.There were significant differences in texture parameters between OA group and healthy group.Compared to the normal controls,the increased T2 values and some other changed texture parameters may be related to the cartilage degeneration.Besides,some of the texture parameters extracted from ROI1,ROI2,ROI4,ROI5 and ROI6 showed a significant difference.In clinic,the diagnosis of OA mainly rely on the clinical symptom,related clinical evaluation and morphological imaging examination,which can't diagnose disease earlier.The classification results showed that the texture analysis would provide more OA related features to differentiate the OA patients which would help doctors through a more non-objectively and reliable way.This study aim to demonstrate the feasibility of using texture analysis to characterize the spatial distribution of T2 values within articular cartilage in OA patients.MRI T2 mapping combined with texture analysis in the OA patients can detect the change of cartilage tissue,can realize the computer aided diagnosis,improve the reliability and accuracy of the diagnosis of OA.