| BackgroundOver the past few years, Diffusion-weighted Imaging (DWI) has become one of themost popular MR imaging technique in the world. Based on the change of water diffusionin Intracellular and extracellular environment, DWI could indirectly investigate themicroenvironmental structural changes in tumor tissue, and reflexes the cellular density.Malignant tumor usually has more compact density than normal tissue, and is with morecellular substance which could reduce the free diffusing space of water molecule. Highsignal can be observed as the reduction of diffusing coefficient, as well as a low ADCvalue from mono-exponential model calculation. As a classical measurable parameter,ADC has been well accepted and applied in clinic. However, traditional mono-exponentialmodel is built on the isotropy of human tissue. As the complexity of tissue structure,mono-exponential model cannot reflex the information of composition and structuralchanges in practice. Additionally, acquiration of ADC is based on a single b value, underwhich diffusing coefficient of water molecule cannot be well relaxed, as well as thephysical impact factors (eg. cellular density,length of spatial diffusion), which are thedirect elements for evaluation of malignant tumor's biological behavior. From the resultsof exploration research on multi b value imaging, higher b value is not associated withsingle exponential declining DWI signal, which demonstrates the limitation of ADC value.Therefore, as the great improvement of DWI Imaging mechanism research in the resentyears, several diffusion models was revealed for diffusion research, such as Bi-exponentialmodel, Statistical model, Stretched-exponential model and FRactional Order Calculusmodel, as well as new parameters like fast/slow-ADC value(Bi-exponential model),ADC and σ value(Statistical model), DDC and (Stretched-exponential model), D, β,μ(FRactional Order Calculus model). These novel models and parameters lead to differentnew ways to investigate the motion of water molecule in human tissue, and may providediagnostic criteria for diseases.Brain tumor is the most common solid tumor in children and the second commonpediatric malignant tumor after leukemia. With the highest mortality in pediatric malignanttumor, brain tumor seriously threatens children life. Regarding the shortage in someaspects, traditional MR technique dose not perform well in diagnosis and distinguish of pediatric brain tumors. Therefore, diagnostic accuracy remains the clinical key point forpediatric brain tumor.ObjectiveThis study initially uses the FROC model for pediatric brain tumor, to explore theclinical feasibility and prospect. Based on quantitative analysis, relationship betweenparameters, as well as association between parameters and brain tumor tissue structure willbe studied. Together with the tumor grade distinguish by different parameters, proper waysfor tumor diagnosis will be explored. Another goal of this study is to compare the results ofdifferent model for a certain brain tumor. New MR technique could potentially be appliedin clinic according to the results of this study.MethodFROC model was used in healthy volunteers to characterize diffusion imaging innormal tumor tissue and to investigate the clinic significance of three parameters in theFROC model. A total of32pediatric patients with brain tumor were included. All thepatients were divided into two groups with different tumor grades according to2007WHOclassification for central nervous system tumor. Comparasion between two groups wasbased on FROC model with three parameters, in view of which color maps with differentparameters was drawn. To analyze the association between model and parameters, differentdiffusion models (Stretched-exponential model, Statistical model, mono-exponential model)together with parameters were compared for certain brain tumor (PA). Additionally,histological characteristics were analyzed for tumor tissue from normal brain tissue.Results3healthy volunteers were included for confirmation the feasibility of FROC modelwith high b value in normal brain tissue. In FROC model, parameter D was diffusioncoefficient, β correlated to tissue homogeneity, and μ was associated to diffusion separationdistance of water molecule.A total of32pediatric patients with brain tumor were involved in this study.2patientswith PA and1patient with MB were excluded due to motion effects. Thus, a total of29patients were finally included, including18males and11females with an average age3.3year (range from4month to7year). According to WHO classification, all the29cases with pathological confirmation were divided in two groups (low grade group with grade Iand II tumors, high grade group with grade III and IV tumors).14cases were included inlow grade group while15for high grade one. Parameter D, β, μ values in high grade groupwere significant lower than the ones in low group (p<0.05). According to the calculation ofAUC in ROC curve, parameter D, β, μ were useful to differentiate the tumors with low orhigh grade. However, as the overlap of D and μ in two tumor grades, β was the bestparameter for tumor distinguishes.Based on the results of four diffusion models for single MB and PA, no statisticaldifferences were found between D values in different models(p>0.05), while σ, β, μ valuesin brain tumor tissues were statistical different from the ones in normal tissues(p<0.05).Quantitative analysis for water molecule diffusion coefficient, free length and homogeneityof different tissue structure was obtained by FROC, which provided parameters at mostwith most meticulous explanation in all the models.ConclusionFROC could be applied not only in healthy volunteers but also in pediatric patientswith brain tumor. D, β, μ are all useful to differentiate the tumor grades while β is the bestparameter in all. D values are similar in four models and D, β, μ in tumor tissue aredifferent from the ones in normal tissue. With the acquiration of quantitative analysis fortissue structure, FROC provides parameters at most with most meticulous explanation inall the models, which may be applied in clinic in the future.
|
PDF Full Text Request