Radonc/BME Medical Image and
Computational Analysis Laboratory
Randi Dias
BS Biology, pre-med
expected graduation: Spring 2022

email link
  1. My Resume (Jan. 2022)
  2. My page on LinkedIn
  3. Research Overview

    Validating lung vessel size and application to early pre-term gestation patients

    Background and Significance
    While a standardized scoring system exists to quantify the degree of underdevelopment in preterm babies (preemies), lung development in preemies has not been fully understood, thus preventing the advancement in care in improving their early survival. For example, preemies are often put on respirators until their lungs are adequately developed, yet no objective criteria exists to inform when it is safe to remove the respirator. This project’s goal is to identify quantifiable metrics of lung development in preemies relative to full-term children, eventually providing physicians clinical guidance that can further improve their patients’ well-being.

    My hypothesis is that children who are born prematurely will have a slower progression in vessel growth than children born at the standard gestational period and will be more likely to develop lung complications as they continue to grow.

    Chest CT scans from 11 adolescent patients aged 5-19 years that were mostly born prematurely were retrospectively gathered from UF Health Shands Hospital under an approved protocol. Pediatric patients up to 18 years were selected from those who received >2 follow-up chest CT scans from 2014-19. In-house software built upon the NIH ImageJ platform was used to extract the pulmonary vessel trees and characterize the radius and length of each vessel branch. This approach was applied to the right-hemi-lung of each patient over 3-7 follow-up scans to generate branch-radii histograms for branches > 1 mm in diameter.he lung growth rate for premature children and be able to determine the best time to remove a premature child off a respirator.

    Results to-date
    The number of branches in preterm patients increased between the ages of 5-7 (Figure A). However, between the ages 6-12, the number of vessels decreased and slowly began to increase and stabilize between the ages of 12-16. On the other hand, the number of branches for full-term patients steadily increased with age. A two-sample t-test was conducted with the null hypothesis being that the two data sets are equal, with the resulting p-value being 0.057.

    Because of the marginal p-value, additional data points are needed to identify significant differences in lung vessel development for preemies. An additional 49 age-matched data sets are available for analysis. We observed that CT slice thicknesses and overall image quality greatly affected the number branches detected by the program. Calibration curves are being created to account for these effects.

    Overall, the data showed the number of branches increased with age in the 2-19 year range. Once a child reaches his or her mid-teens, it is expected that the number of branches will level off. The initial limited dataset does not show a consistent difference in branch number versus age between preemie and full-term subjects. However, analysis of additional pediatric datasets is on-going to increase the statistical power for detecting trends and group differences. Future analyses will include subjects in the 1-week to 5-year age group where larger group differences are expected. With the larger dataset, vascular development will be correlated with gestational age, sex, body weight, and presence of secondary infection/disease (e.g. cystic fibrosis).

  4. National Presentations:
    Saini A, Dias R. Williams C, Siva Kumar S, O’Dell W.
          Distinguishing markers of lung vasculature development in pre-term vs. full-term patients.
          American Academy of Pediatrics Annual Meeting (virtual):
          Oct. 10, 2020.

  5. Related MIACALab projects
    1. Vessel Segmentation

  6. Close Research Collaborators
    1. W. O'Dell, PhD (Radiation Oncology Research)
    2. Sreekala Prabhakaran, MD (Pediatric Pulmonology)
    3. Jennifer Co-Vu, MD (Pediatric Cardiology)