BME MS Thesis Defense: Kristina MacLeod “Regional 3D High-Resolution Peripheral Quantitative CT Registration & Analysis: Methods & Applications”

Master’s Thesis Defense  

“Regional 3D High-Resolution Peripheral Quantitative CT Registration & Analysis: Methods & Applications” 

Kristina MacLeod 

Abstract: High-resolution peripheral quantitative CT (HRpQCT) is widely used in research to measure changes in bone microstructure in response to interventions such as drugs and physical activity. Whole-bone analyses cannot detect localized changes to bone that may be important. We developed and validated a method to perform subregional analysis of HRpQCT scans of the distal tibia in native SCANCO software. 

Distal tibia HRpQCT scans (XtremeCT I, SCANCO, Brüttisellen, Switzerland) include 110 slices over a 9.01mm region of interest at a resolution of 82µm. Using SCANCO's internal Image Processing Language, we developed scripts to automatically divide scans into quadrant subregions for microstructural evaluation. This method was validated against whole-tibia measures and coefficient of variation (%CV) within a precision analysis dataset was calculated. Total area, bone mineral density (BMD), and bone volume to total volume ratio (BV/TV) were recovered from subregions in a test set of 8 scans and compared to whole-bone values. To investigate regional changes, we analyzed a set of 10 physically active non-runners (control) and 11 runners training for their first marathon (experimental). We compared subregional adaptation before (T1) and after (T2) the 8-month marathon training period using repeated measures ANOVA (time x region repeated).  

Subregion division takes ~ 2 minutes, using 2 scripts. Average %CV across all parameters was 0.64%, 1.09%, 1.14%, 0.73%, and 0.92% for the whole tibia and each region respectively. Average summed area and area-weighted total BMD and BV/TV of the four subregions were within 1% of the whole tibia values. In the experimental group, the runners exhibited significant regional increases over time in cortical density (P = 0.02) and decreases in BV/TV (P < 0.05) that were not captured by whole-tibia analysis. Additionally, significant differences between regions were observed for all measured parameters in both groups (P < 0.01).  

This novel workflow produces consistent subregional values relative to the total bone metrics. Applying this method, we observed that marathon training increased bone formation most in the medial/posterior quadrants. These results demonstrate an efficient and automated method for accurate subregional analysis. Our data further shows the benefit of subregion analysis by highlighting local bone adaptation that can be overlooked using traditional whole-bone methods. 

For a zoom link, please email kharrison@wpi.edu