论文部分内容阅读
PURPOSE : To evaluate the feasibility of using texture analysis of apparent diffusion coefficient(ADC)maps of bone marrow for differential diagnosis of multiple myeloma(MM)and monoclonal gammopathy of undetermined significance(MGUS).METHODS: 17 consecutive patients with plasma disorders was recruited prospectively,including 8 clinically diagnosed MM and 9 patients with MGUS.All patients received whole-body DWI MRI.A hyperintense appearence on high-b-value DWI images was considered as myeloma infiltration.Texture Analysis of ADC maps on lumber vertebral bodies in each patient were performed using TexRAD commercially available research software(TexRAD Ltd,Cambridge,UK)by manually delineating a round region of interest covering the middle cross-sectional area of each vertebra.Vertebrae with focal lesions or compression fracture were excluded from the analysis.The technique selectively filters and extracts textures at different size scales(fine,medium and coarse)followed by quantification of the histogram using 6 parameters: mean,standard-deviation(SD),kurtosis,entropy,skewness,and mean value of positive pixels(MPP).Students t test was performed to compare the texture analysis parameters between myeloma-infiltrated vertebrae and non-infiltrated vertebrae.ROC analysis was performed to assess the diagnostic performance of these parameters to detect bone marrow infiltration of MM.RESULTS: A total of 22 myeloma-infiltrated vertebrae(from MM patients)and 27 non-infiltrated vertebrae(from MGUS patients)were evaluated.At fine texture scale,most of the parameters(ie.mean,,kurtosis,entropy,skewness,and MPP)except SD were significantly different between the two groups(P<0.05).ROC analysis identified three texture parameters(at fine texure scale)with highest AUCs,including mean(0.974),entropy(0.998)and MPP(0.974).The sensitivity of mean,entropy and MPP was 90.9%,100%and 90.9%respectively.The specificity of mean,entropy and MPP was equal(96.2%).CONCLUSION : Texture analysis of ADC Maps may complement conventional DWI MRI to differentiate between myelomainfiltrated vertebrae from MM patients and non-infiltrated vertebrae from MGUS patients.