3D Volumetric Measurement of Normal Pediatric Livers: Creating a Reference Database and Predictive Model
MetadataShow full item record
PublisherThe University of Arizona.
DescriptionA Thesis submitted to The University of Arizona College of Medicine - Phoenix in partial fulfillment of the requirements for the Degree of Doctor of Medicine.
AbstractBackground: Accurate and reproducible measurements of pediatric organs are necessary for defining normal organ volume, size, growth rates, and patterns of development, which aids in determining pathological variants. Currently, no modern reliable database exists for normal liver volume (LV) in children, and although predictive equations have been proposed, many are based on adult data, ethnically homogenous populations, or are derived from smaller samples and have not utilized advanced imaging technology in determining LV in vivo. Objective: To establish normal LV measurements in children, using a three-dimensional (3D) volumetric approach, with additional consideration for height, weight, body surface area (BSA), and body mass index (BMI), and to develop a predictive model using these parameters. Materials and methods: A retrospective review of normal contrast enhanced abdomen and pelvis CT images of 184 patients from 1 month to 18 years, identified within the Phoenix Children’s Hospital picture archive communications system (PACS) was performed. Gender, age, height and weight were recorded for each patient; BSA and BMI were calculated. LV measurements were obtained using segmentation images software (IntelliSpace, Phillips Healthcare, Haifa, Israel). Results: Univariate analysis of LV was most strongly correlated with and predicted by BSA (R2 = 0.90, p < 0.0001), which could be defined by: LV = -115.5 + 941.7*BSA. In multivariate analysis, BSA (p < 0.0001), gender (p = 0.01), and height (p = 0.001) were the covariates that best predicted LV with an adjusted R2 value of 0.90. 3 Stratifying the model by age did not modify the predictive capabilities of the covariates. Further stratifying by gender revealed inconsistent effect modification in some age groups. Conclusion: Univariate analysis of LV was most strongly correlated with and predicted by BSA, which can be defined by: LV = -115.5 + 941.7*BSA.