Comparison of rectangular and dual-planar positron emission mammography scanners

Abstract

Comparison of Rectangular and Dual-Planar Positron Emission Mammography Scanners Jinyi Qi, Member, IEEE, Chaincy Kuo, Ronald H Huesman, Fellow, IEEE, Gregory J Klein, Member, IEEE, William W Moses, Senior Member, IEEE, and Bryan W Reutter, Member, IEEE Abstract— Breast imaging using dedicated positron emission to- mography (PEM) has gained much interest in the medical imaging field. In this paper, we compare the performance between a rectan- gular geometry and a parallel dual-planar geometry. Both geome- tries are studied with DOI detectors (detectors capable of measur- ing the depth of interaction) and non-DOI detectors. We compare the Fisher information matrix, lesion detection, and quantitation of the four systems. The lesion detectability is measured by the signal-to-noise ratio (SNR) of a prewhitening numerical observer for detecting a known hot spot on a uniform background. Results show that the rectangular system with DOI has the highest SNR for the detection task and the lowest bias at any given noise level for the quantitation task. They also show that for small simulated lesions the parallel dual-planar system with DOI detectors outperforms the rectangular system with non-DOI detectors, while the rectan- gular system with non-DOI detectors can outperform the parallel dual-planar system with DOI detectors for large simulated lesions. mm mm (a) I. I NTRODUCTION Breast imaging using positron emission tomography (PET) has gained much interest in the medical imaging field. Dedi- cated PET scanners, commonly referred to as positron emission mammography (PEM) cameras, are being developed to improve the cost effectiveness for detecting breast cancers [1-5]. These PEM scanners have special geometries that are different from the conventional circular design of whole body PET scanners. Compared to whole body PET, PEM has much greater sensitiv- ity for detecting breast cancers [3]. A PEM camera (Fig. 1a) is under development at our labora- tory that consists of four banks of detector modules (two banks of 3×3 modules left and right and two banks of 3×4 modules top and bottom). As its cross section is rectangular, we refer to this geometry as “rectangular” or “R-PEM.” Each detector module consists of an 8×8 array of 3×3×30 mm 3 lutetium oxy- orthosilicate (LSO) crystals. The LSO crystals are coupled to a photodiode (PD) array at the front and a photomultiplier tube (PMT) at the back. By measuring the signals from both the PD and the PMT, the depth of interaction (DOI) of the photon can be estimated [4]. For each crystal, the system digitizes DOI in- formation with three bits. Each detector is placed in coincidence with all detectors in the other three banks, giving rise to 172 mil- lion possible lines of response (LORs). The maximum field of view (FOV) of the system is 96×72×72 mm 3 . This work was supported by the U.S. Department of Health and Human Ser- vices under grant P01 HL25840, by the National Cancer Institute under grant R01 CA 67911, and by the Director, Office of Science, Office of Biological and Environmental Research, Medical Sciences Division of the US Department of Energy under contract DE-AC03-76SF00098. The authors are with the Center for Functional Imaging, Lawrence Berkeley National Laboratory, Berkeley, CA 94720 USA (telephone: 510-486-4695, e- mail: JQi@lbl.gov). S = 72 mm W = 168 mm D = 72 mm (b) Fig. 1. (a) Rectangular PEM Geometry. (b) Parallel dual-planar PEM. Another popular design that has been pursued by researchers is a parallel dual-planar geometry (Fig. 1b) [1], [5]. It is similar to using only the top and bottom detector banks in the rectan- gular PEM scanner. We refer to this geometry as “parallel” or “P-PEM.” This design is easier to construct than the rectangular PEM and has greater flexibility in positioning the object into the FOV. However, it suffers from a limited angle of view. In this paper we explore the imaging performance of these geometries, comparing the R-PEM with the P-PEM for lesion detection and quantitation. Both designs are studied with DOI detectors (detectors that are capable of measuring the depth of interaction) and non-DOI detectors. The P-PEM systems are assumed to be built with the same number of detectors as the R- PEM so that each parallel detector bank has 3×7 detector mod- ules. The width (W) and depth (D) of each detector bank are 72 mm