Receiver-operating-characteristic analysis of an automated program for analyzing striatal uptake of 123I-ioflupane SPECT images: Calibration using visual reads

Phillip Hsin Kuo, Ryan Avery, Elizabeth Krupinski, Hong Lei, Adam Bauer, Scott Sherman, Natalie McMillan, John Seibyl, George Zubal

Research output: Contribution to journalArticlepeer-review

12 Scopus citations

Abstract

A fully automated objective striatal analysis (OSA) program that quantitates dopamine transporter uptake in subjects with suspected Parkinson's disease was applied to images from clinical 123I-ioflupane studies. The striatal binding ratios or alternatively the specific binding ratio (SBR) of the lowest putamen uptake was computed, and receiver-operating-characteristic (ROC) analysis was applied to 94 subjects to determine the best discriminator using this quantitative method. Methods: Ninety-four 123I-ioflupane SPECT scans were analyzed from patients referred to our clinical imaging department and were reconstructed using the manufacturer-supplied reconstruction and filtering parameters for the radiotracer. Three trained readers conducted independent visual interpretations and reported each case as either normal or showing dopaminergic deficit (abnormal). The same images were analyzed using the OSA software, which locates the striatal and occipital structures and places regions of interest on the caudate and putamen. Additionally, the OSA places a region of interest on the occipital region that is used to calculate the background-subtracted SBR. The lower SBR of the 2 putamen regions was taken as the quantitative report. The 33 normal (bilateral comma-shaped striata) and 61 abnormal (unilateral or bilateral dopaminergic deficit) studies were analyzed to generate ROC curves. Results: Twenty-nine of the scans were interpreted as normal and 59 as abnormal by all 3 readers. For 12 scans, the 3 readers did not unanimously agree in their interpretations (discordant). The ROC analysis, which used the visual-majority-consensus interpretation from the readers as the gold standard, yielded an area under the curve of 0.958 when using 1.08 as the threshold SBR for the lowest putamen. The sensitivity and specificity of the automated quantitative analysis were 95% and 89%, respectively. Conclusion: The OSA program delivers SBR quantitative values that have a high sensitivity and specificity, compared with visual interpretations by trained nuclear medicine readers. Such a program could be a helpful aid for readers not yet experienced with 123I-ioflupane SPECT images and if further adapted and validated may be useful to assess disease progression during pharmaceutical testing of therapies.

Original languageEnglish (US)
Pages (from-to)26-31
Number of pages6
JournalJournal of nuclear medicine technology
Volume41
Issue number1
DOIs
StatePublished - Mar 1 2013

Keywords

  • Brain SPECT
  • Datscan
  • Ioflupane
  • Molecular neuroimaging
  • Software analysis

ASJC Scopus subject areas

  • Radiological and Ultrasound Technology
  • Radiology Nuclear Medicine and imaging

Fingerprint

Dive into the research topics of 'Receiver-operating-characteristic analysis of an automated program for analyzing striatal uptake of <sup>123</sup>I-ioflupane SPECT images: Calibration using visual reads'. Together they form a unique fingerprint.

Cite this