Toggle Main Menu Toggle Search

Open Access padlockePrints

A Visual-Search Model Observer for Multitarget Volumetric SPECT Images

Lookup NU author(s): Dr Anando SenORCiD


Full text for this publication is not currently held within this repository. Alternative links are provided below where available.


The main objective of this work was to test a visual search (VS) human-model observer for a multitarget tumor localization task featuring single photon emission computed tomography (SPECT) image volumes. The image volumes were simulated from a digital anthropomorphic phantom containing tumors of three different sizes in the prostate and lymph-node areas of a single phantom. Image volumes were simulated with six different lesion to background contrast ratios. This paper reports the comparison between the VS and the channelized non pre-whitening (CNPW) observers with jackknife free-response ROC (JAFROC) study. The performance of the VS was always lower than that of the CNPW and as the contrast increased, the performance was also increased for both the model-observers. The results were consistent with previous results obtained from 2D single-slice studies.

Publication metadata

Author(s): Banerjee K, Sen A, Chakrabarty D, Gifford HC

Editor(s): King M; Glick S; Mueller K

Publication type: Conference Proceedings (inc. Abstract)

Publication status: Published

Conference Name: 13th International Meeting on Fully Three-Dimensional Image Reconstruction in Radiology and Nuclear Medicine

Year of Conference: 2015

Pages: 114-117

Online publication date: 01/06/2015

Acceptance date: 19/02/2015

Publisher: Fully3D