Difference between revisions of "VolCT - Group 1B"

From QIBA Wiki
Jump to navigation Jump to search
Line 1: Line 1:
 
==Current Goals & Status:==
 
==Current Goals & Status:==
  
−
* Question: What is measurement variation for different software/user methods for a reference set of image data?
+
* Question: What is the minimum detectable level of change in patient datasets under a “No Change” condition?
−
** Intra-rater reliability, i.e., test-retest precision for pre-specified tumors (a.k.a. "marked up") by single image analysis operators
+
** AIMS: For patient datasets acquired over a very short time interval (“no change” condition”) investigate variance in measuring change in volume, diameter and bi-directional diameters of lesions
−
** Ground Truth:  Pooled from all image analyses meeting quality criteria to estimate confidence intervals around "true" volume.
+
** Investigate inter-observer variability in each task
−
** Approach: Patient Change Data from RIDER Database:
+
** Investigate Intra-observer variability in each task
−
*** > 300 cases as of Sept 2008
+
 
−
*** Most with multiple time points
+
** Methods - Materials
−
*** Few with thin slice (<1.5 mm thick) image data
+
*** Image Data from MSK Coffee Break Experiment
−
*** No contours provided (user has to provide their own as part of measurement)
+
*** 32 NSCLC patients, imaged twice on same scanner over period < 15 minutes
 +
*** Thin slice (1.25 mm thick) image data
 +
*** Selected one lesion per patient
 
*** no outcome data
 
*** no outcome data
 
*** Publicly Available on NCIA
 
*** Publicly Available on NCIA
 +
 +
** Methods - Readings
 +
*** 5 readers
 +
*** Read CT at timepoint 1, CT at timepoint 2 and repeat either CT timepoint 1 or timeoint 2 to assess intra-reader variability
 +
** Methods - Measurements
 +
*** Measure Single Longest Diameter, manually (SLD)
 +
*** Measure Longest Diameter and Perpendicular Diameter; take product to obtain Product of Diameters (POD)
 +
*** Contour Full Boundary of lesion with semiautomated tool; calculate volume (Vol)
 +
*** Extract other measures as well (SLD from POD measureement; SLD from VOl and POD from Vol).
 +
** The Order of Readers, Lesions and Measurement Methods were Randomized (Measurements done in Blocks of 10)
 +
 +
** Analyses
 +
*** Pooled Analysis across readers and lesions; Percent Difference between Scan 1 and Scan 2 for each measurement method
 +
*** Subgroup Analyses: (a) Based on Measureable/Non-Measureable lesions identified by Mike O'Neal of CoreLabs; (b) Difficult, Moderate and Easy as Identified by Mike McNitt-Gray
 +
 +
 
* Related Profile: [[CT_Lung_Nodule_Volume_Quantification_Profile#Activity:_Measurement| Lung Vol Quantification - Measurement Activity]]
 
* Related Profile: [[CT_Lung_Nodule_Volume_Quantification_Profile#Activity:_Measurement| Lung Vol Quantification - Measurement Activity]]
  

Revision as of 04:34, 20 April 2011

Current Goals & Status:

  • Question: What is the minimum detectable level of change in patient datasets under a “No Change” condition?
    • AIMS: For patient datasets acquired over a very short time interval (“no change” condition”) investigate variance in measuring change in volume, diameter and bi-directional diameters of lesions
    • Investigate inter-observer variability in each task
    • Investigate Intra-observer variability in each task
    • Methods - Materials
      • Image Data from MSK Coffee Break Experiment
      • 32 NSCLC patients, imaged twice on same scanner over period < 15 minutes
      • Thin slice (1.25 mm thick) image data
      • Selected one lesion per patient
      • no outcome data
      • Publicly Available on NCIA
    • Methods - Readings
      • 5 readers
      • Read CT at timepoint 1, CT at timepoint 2 and repeat either CT timepoint 1 or timeoint 2 to assess intra-reader variability
    • Methods - Measurements
      • Measure Single Longest Diameter, manually (SLD)
      • Measure Longest Diameter and Perpendicular Diameter; take product to obtain Product of Diameters (POD)
      • Contour Full Boundary of lesion with semiautomated tool; calculate volume (Vol)
      • Extract other measures as well (SLD from POD measureement; SLD from VOl and POD from Vol).
    • The Order of Readers, Lesions and Measurement Methods were Randomized (Measurements done in Blocks of 10)
    • Analyses
      • Pooled Analysis across readers and lesions; Percent Difference between Scan 1 and Scan 2 for each measurement method
      • Subgroup Analyses: (a) Based on Measureable/Non-Measureable lesions identified by Mike O'Neal of CoreLabs; (b) Difficult, Moderate and Easy as Identified by Mike McNitt-Gray



  • Conference Paper on Reproducibility and Accuracy of Uni- and Bi-Dimensional Tumor Measurements from MRI and CT Volumes, which looks at RECIST, WHO and Volume measurements.
Retrieved from "https://qibawiki.rsna.org/index.php?title=VolCT_-_Group_1B&oldid=3064"