Statistics Collaborative - Design and analysis for biomedical research

CapabilitiesCapabilities

Diagnostic Imaging

Statistics Collaborative, Inc. (SCI) has extensive experience with statistical methods for clinical studies of diagnostic imaging technologies. SCI has contributed to reader study trials, which have different analytic issues from other common statistical analyses in clinical trials. With access to the “truth”, SCI has examined and assessed various diagnostic tools’ abilities to reach the same conclusions using sensitivity, specificity, and ROC curves. SCI has established methods for determining accurate results from studies using both patients and physicians as samples.

Dr. Alicia Toledano, whose expertise in this area extends back to 1991, has served on the Food and Drug Administration (FDA) Advisory Committee for Radiological Devices and was Deputy Director of the American College of Radiology Imaging Network (ACRIN) Biostatistics and Data Management Center, where she worked on several trials in magnetic resonance imaging (MRI), PET scanning, CT, and digital mammography.

Examples of SCI’s staff work in Diagnostic Imaging Technologies:

  • CT colonography: Dr. Toledano provided study design and statistical planning and analysis for both a retrospective and large, multicenter, prospective study of a potential alternative to optical colonoscopy.
  • Digital Mammography in Screening Trial: Dr. Toledano designed multi-reader, multi-case studies comparing diagnostic accuracy of digital and film screen mammography with focus on a third variable. For example, one study examined the effect of the proportion of cases with cancer in the study sample; another examined the impact of breast density and how well irregular growths could be detected on film versus digital imagery.
  • Investigational MRI contrast agent: SCI analyzed whether an investigational contrast agent for MRIs aided in the distinction between metastatic and non-metastatic lymph nodes in the prostate, cervix, and uterus. SCI developed complex sensitivity and specificity estimates for regional nodal analyses that accounted for intra-patient correlations and compared enhanced versus unenhanced MRI diagnoses against the “true” results from lesion biopsies.