Cancellous Bone Density Evaluation using Ultrasound Backscatter from an Imaging System: Exploring the Possibility for Fetal Bone Density Evaluation

Osteoporosis is a common chronic disease and a well known major source of morbility and mortality among the elderly. Low bone density also occurs in infants and small children during development and can be problematically excessive if the fetus experiences issues during pregnancy such as malnutrition, lack of vitamin D and smoking. Currently the only available methodologies for fetal bone density evaluation are Dual-energy X-ray Absorptiometry (DEXA) or Magnetic Resonance Imaging (MRI). Both are sensitive to movement artifacts. DEXA exposes the subjects to significant radiation so is not suggested during pregnancy. Quantitative MRI is noisy, expensive, slow (8-20 mins) and the effects of high field strengths on the developing fetus is unknown. Therefore, the goal of this study is to find a fast, accurate and non-ionizing method for the evaluation of fetal bone density. In this study, the quantitative ultrasound backscatter coefficient (BSC) was chosen to evaluate bone density using the B-mode ultrasound system. Compared with the speed of sound and ultrasound attenuation in the traditional ultrasound measurement for bone density, the backscatter method is more accessible to central sites such as the human spine and fetal femur bone. Additionally, it has a rapid path to commercialization with the potential to be added as a new feature in the current commercial ultrasound imaging systems for bone density evaluation. The contributions of this work are: 1. A simulation study was accomplished that compared backscatter coefficients from a single element transducer, a linear array transducer, and a curved array transducer with the change of trabecular thickness and trabecular spacing. An overall similar Pearson correlation (single: R = 0.94, SD = 10.84dB, linear: R = 0.92, SD = 6.6dB, curved: R=0.95, SD=6.89dB) between the BSC and porosity was found from three transducers, but the standard deviation (SD) was smaller from the two array probes. This improved standard deviation may result from the wider spatial range of the array transducers. 2. A simulation model using COMSOL for the fetal bone density evaluation was built based on the Biot’s poroelastic theory and the backscatter coefficient. The theoretical backscatter coefficient from the Biot model was calculated with the best available biomechanical parameters from the human femoral cancellous bone and the geometrical features of the fetal femur. This work also proposed a method for compensating the ultrasound signal attenuation from abdominal tissue, femur tissue, amniotic fluid between the probe and fetal femur. The result showed good correlation of BSC (R = -0.9970, P = 2.0058e^-04, SD = 10.21%) and apparent integrated backscatter (AIB) (R = -0.9469, P = 0.0146, SD = 10.62%) with the porosity. This suggests in vivo ultrasound bone evaluation could be implemented in the current commercial ultrasound B-mode systems. 3. An in vitro study was conducted that compared the backscatter coefficient (BSC), the apparent integrated backscatter (AIB) and the Spectrum Centroid Shift (SCS) from the fundamental backscatter signal and the second harmonics of the ultrasound imaging system. The result from the second harmonics (R : BSC = 0.7374, AIB = 0.6243, SCS =-0.6421) showed better correlation than the fundamental backscatter (R : BSC = 0.7055, AIB = 0.5393, SCS = -0.5858) with a gold standard bone mineral density obtained from DEXA scans of the same samples. An analysis from the Farran cylindrical model and the second harmonics of a rigid cylinder showed the second harmonics has less noise and showed better performance than the fundamental backscatter approach. In conclusion, the backscatter coefficient from ultrasound imaging showed good correlation in both the simulation studies and the in vitro study. It has the potential to be a convenient, fast, cheap methodology for adult and fetal bone density evaluation