Whole-body dosimetry for targeted radionuclide therapy using spectral analysis

Antigoni Divoli, Antonello Spinelli, Sarah Chittenden, David Dearnaley, Glenn Flux

Research output: Contribution to journalArticlepeer-review


The whole-body dose (WBD) is routinely calculated for targeted radionuclide therapy (TRT). The aim of this work was to investigate the feasibility of using spectral analysis (SA) for the automatic delineation of decay phases, and consequently, the calculation of the WBD given a whole-body (WB) time-activity curve (TAC). SA characterizes the TAC as an arbitrary sum of exponential functions determined by fitting the data with a non-negative least-squares (NNLS) algorithm. The cumulated activity (CA) is calculated analytically as the integral of the fitted curve while the number of phases describing the kinetics of the radiopharmaceutical and the half-lives of the phases can be determined from the spectrum. The uncertainty associated with the estimation of the WBD can be obtained using bootstrap techniques. SA was applied to WB TACs from 186Re-HEDP and 131I-mIBG therapies. The results were compared to results obtained using a semiempirical method and showed good agreement in the calculated WBDs. Bootstrapping with resampling on a subset of patients from the two therapies showed much larger coefficient-of-variation (CV) for the 186Re-HEDP TACs than for the 131I-mIBG therapies. We concluded that SA provides a fast, accurate, and reproducible method to obtain WBDs and accurate estimates of the parameters describing the radiotracer kinetics. The method could be extended to other dosimetric applications.

Original languageEnglish
Pages (from-to)66-71
Number of pages6
JournalCancer Biotherapy and Radiopharmaceuticals
Issue number1
Publication statusPublished - 2005


  • Radionuclide therapy
  • Spectral analysis
  • Whole-body dose

ASJC Scopus subject areas

  • Cancer Research
  • Pharmacology
  • Oncology


Dive into the research topics of 'Whole-body dosimetry for targeted radionuclide therapy using spectral analysis'. Together they form a unique fingerprint.

Cite this