Quantitative Imaging: Protocol Development and Quality Assurance Issues for Medical Imaging in Clini

1
Quantitative Imaging Protocol Development and
Quality Assurance Issues for Medical Imaging in
Clinical Trials

• H. Cecil Charles, Ph.D.
• Director
• Duke Image Analysis Laboratory
• Duke University Medical Center

2
Overview

• Quantitative Imaging vs Clinical Imaging
• Protocol Development Issues
• Centrally Monitored QC/QA in multi-center trials
• Central Data Analysis/Archival Issues

3
What Is Quantitative Imaging (QI) and How Is It
Different From Clinical Imaging (CI)?
4
CI QI

• Visualization of lesions and/or disease
• Radiologic interpretation
• Rule-out or rule-in a diagnosis
• Diagnostic tree/1,2,3 diagnosis

• Determination of tissue characteristics from
  imaging parameters
• Algorithm/SOP/scaled interpretation
• Numeric output
• Incorporation in hypothesis testing or goal
  driven evaluation

Effect Monitoring
Diagnosis
5
Is there a use for CI in trials

• If imaging is part of the diagnostic inclusion or
  exclusion criteria, a screening scan may be
  required
• The screening scan may or may not be according to
  the QI protocol
• Subsequent imaging sessions (including a baseline
  scan) are based on the QI protocol

6
Examples of QI

• Organ volumes or Subvolumes
• Perfusion/Permeability/blood flow
• Atrophy indices
• Necrosis/Hypoxia Indices

• Metabolic Indices (e.g. pH, energetics)
• Ligand Binding
• Vascular Indices

7
ISSUES for QI

• Study Protocol Design
• Data Quality
• Data Format Issues
• Data Cleaning
• Data Registration (serial studies)
• Data Analysis
• Data Archival

8
Study Protocol General

• Driven by Study goals and QI Algorithm(s)
  (Analyses)
• Maximize Information Content per unit time
• Deterministic Figure of Merit (FOM)
• CNR/(unit resolution unit time)
• Patient Comfort/Compliance

9
Multiple Sites/Platforms

• Imaging protocol cross-validation
• Rationalize Nomenclature
• Uniform site training tailored to
  manufacturer/HW/SW status
• Retrain with upgrades if necessary
• Centrally monitored protocol compliance (QC)

10
Data Quality Assessment

• SNR/CNR
• Artifacts (e.g. Motion) quantitative criteria
  clutter/noise
• Protocol Adherence
• Scan Parameters
• Schedule
• Technical Parameters (e.g. contrast dose and
  rate)
• System Performance
• Spatial Fidelity! (esp. in serial studies)

11
Incoming Data Formats

• Native data from multiple manufacturers and
  multiple S/W releases
• Varying Media Formats
• MODs, DAT(s), CDROM(s)
• Varying File Formats
• DICOM(s)
• Proprietary Formats
• ACR/NEMA
• Local formats (non-commercial PACS)

12
Data Storage Formats

• Imaging Industry standards
• DICOM (a flexible standard)
• Alternate Standards
• (e.g. Analyze, TIFF)

13
Data Cleaning

• Prospective QC
• Rescan if possible
• Minimization of Lost Data
• Data Rejection
• Quantitative Basis!
• Site Notification

14
Data Registration

• Minimize positioning errors in protocol
• Use Immobilizers to alleviate motion
• REGISTER serial scans

15
Data Registration

• Even with on site training and quality
  technologists, some misalignment will occur in
  serial studies
• Alignment of the datasets minimizes the impact of
  this problem

16
Data Analysis

• Prospective Criteria based on needs of study
• Optimize FOM and QC criteria to match needs of
  algorithm
• SOP
• Replicate analysis to address drift

17
Data Archival

• Driven by needs of sponsor and regulatory
  agencies
• Central Consolidation and Storage
• Coordinating Center Archival

18
Summary

• Close Intellectual and technological relationship
  among the sponsor, imaging site(s) and imaging
  coordinating center
• Ongoing QC/QA
• Blinded quantitative data analysis
• Regulatory compliance