R. D. Badawi

Use of positron emission tomography in oncology and its potential role to assess response to imatinib mesylate therapy in gastrointestinal stromal tumors (GISTs)

The reliability of established anatomical imaging techniques, such as computed tomography (CT) and magnetic resonance imaging (MRI), is compromised in following response to certain types of treatment if metabolic improvement occurs before morphologic change is apparent. Thus, traditional imaging techniques cannot discriminate early tumor response because they are based on purely visual structural assessments. Recently, the use of positron emission tomography (PET), most commonly employing the radiotracer 18F-fluoro-2-deoxy-D-glucose (FDG), has been shown to improve the assessment of tumor behavior by highlighting early functional changes in tumor glucose metabolism that appear to correlate closely with metabolic tumor response to imatinib mesylate. Like CT and MRI, PET can identify an abnormal mass; its improvement over these techniques lies in its ability to differentiate active tumor from necrosing tissue, malignant from benign tissue, and recurrent tumor from scar tissue. Understanding and using this tool should improve our ability to accurately follow response in GIST patients treated with imatinib mesylate, and permit this new therapeutic approach to be used optimally with accurate follow-up assessments and informed therapeutic decision-making.

CT and PET: Early Prognostic Indicators of Response to Imatinib Mesylate in Patients with Gastrointestinal Stromal Tumor

OBJECTIVEnWe report results from a pilot study aimed at optimizing the use of CT bidimensional measurements and 18F-FDG PET maximum standardized uptake values (SUVs-(max)) for determining response to prolonged imatinib mesylate treatment in patients with advanced gastrointestinal stromal tumors (GISTs).nnnSUBJECTS AND METHODSnSixty-three patients enrolled in a multicenter trial evaluating imatinib mesylate therapy for advanced GIST underwent FDG PET at baseline and 1 month after initiation of treatment. Of these 63 patients, 58 underwent concomitant CT. Time-to-treatment failure (TTF) was used as the outcome measure. Patients were followed up over a range of 23.7 to 37 months (median, 31.7 months). The predictive power of change in CT bidimensional measurements, change in PET SUVmax, and PET SUVmax at 1 month after initiation of treatment were determined, optimized, and compared. The effectiveness of combining metrics was also evaluated.nnnRESULTSnBoth a threshold PET SUVmax value of 2.5 at 1 month (p = 0.04) and the European Organization for Research and Treatment of Cancer (EORTC) criteria for partial response on FDG PET (25% reduction in PET SUVmax) at 1 month (p = 0.004) were predictive of prolonged treatment success. The Southwest Oncology Group (SWOG) criteria for partial response ((3) 50% reduction in CT bidimensional measurements) at 1 month were not predictive (p = 0.55) of TTF. Optimizing metrics improved results performance. An optimized PET SUVmax threshold of 3.4 (p = 0.00002), a reduction in the SUVmax of 40% (p = 0.002), and an optimized CT bidimensional measurement threshold--that is, no growth from baseline to 1 month (p = 0.00005)--outperformed the existing standards (i.e., EORTC and SWOG criteria). Combinations of metrics did not improve performance.nnnCONCLUSIONnThe two best metrics were the optimized PET SUVmax threshold of 3.4 at 1 month (p = 0.00002) and the optimized CT bidimensional measurement threshold (no growth from baseline to 1 month, p = 0.00005) in this patient group.

A multiscanner evaluation of PET image quality using phantom studies

We present preliminary results from an investigation of the impact of phantom diameter on sinogram and image quality for two PET scanners, the Allegro and ECAT HR/sup +/, both operated in fully-3D acquisition mode. Here, we evaluate the global sinogram NEC measurement and in subsequent studies will be looking at a local image signal-to-noise ratio (SNR) estimate derived from a nonprewhitening matched filter observer (SNR/sub NPW/). We evaluate two different metrics, the global sinogram NEC measurement and a local image signal-to-noise (SNR) estimate derived from nonprewhitening matched filter observer (SNR/sub NPW/). There is a significant decrease in the NEC rates when going from small to large diameter phantoms. As the diameter of the tissue-equivalent attenuating cylinder increases from 20 cm to 35 cm, the peak NEC rate drops by a factor of six for the Allegro scanner. If /spl radic/NEC is an accurate measure of image quality, we expect to see a corresponding degradation in detection specific tasks, such as the SNR/sub NPW/ for a nonprewhitening matched filter.

Count-rate dependent event mispositioning and NEC in PET

Most current PET detector designs suffer from event mispositioning at high count rates, as scintillation light from nearby and nearly simultaneous gamma ray conversions becomes mixed. We have used the NEMA NU 2-2001 70 cm test phantom and a Na-22 point source to quantify this effect as a function of activity on two block-detector tomographs (the Siemens/CTI HR+ and the General Electric Discovery LS), and two Anger-type PET tomographs (the Siemens ECAM DUET and the Philips CPET+). After accounting for event losses due to dead time, we find that the number of counts on LORs passing through a cylinder of diameter and height equal to the point-source full width at tenth-maximum measured at low rate surrounding the point source decreases by between 9% (HR+) and 35% (CPET+) at the activity giving rise to peak noise equivalent count (NEC) rate. Mispositioned events act to reduce signal-to-noise ratio, both by reducing apparent activity at the originating location and by increasing the signal background. We have reformulated the conventional expression for NEC rate to account for this phenomenon. The new formulation of NEC, which we call NEC/sup */, results in a lower peak value which in turn occurs at a lower activity concentration than for the conventional formulation.

A simulation-based assessment of the revised NEMA NU-2 70 cm long test phantom for PET

A 70 cm long polyethylene cylinder of radius 10 cm containing an off-center line source has been suggested for the characterization of PET tomograph whole-body performance in terms of scatter fraction, sensitivity to randoms and count-rate capability. In this work we perform a series of Monte Carlo simulations of a full-ring BGO system operating in 3D mode and a full-ring sodium iodide system operating in 3D mode to determine how closely scatter fraction and NEC results from the 70 cm cylinder follow those found in anthropomorphic activity and attenuation distributions. We find that the 70 cm cylinder measurements correctly rank-order the tomographs in terms of scatter fraction and NEC performance, but do not always accurately predict the actual scatter fractions or the peak NEC values encountered in the anthropomorphic models. Peak NEC for the phantom occurs at around double the activity concentrations which give rise to peak NEC in the anthropomorphic objects.

Relative lesion detectability in 3D vs. 2D dedicated multi-ring PET

The authors estimated the detectability of spheres of different sizes but equal activity contrast, embedded in a clinically realistic phantom in order to compare two-dimensional (2D) and three-dimensional (3D) whole-body (WB) PET imaging for a relevant diagnostic task. Five plastic spheres with inside diameters of 0.8 to 3.4 cm, containing 74 kBq/ml of Ge-68, were placed in an anthropomorphic torso phantom. The background organs contained F-18 activity concentrations in appropriate physiologic proportions, as did a head phantom positioned outside the field of view (FOV) of the authors ECAT-HR+ system. The phantom was scanned for 9 hours at 1 bed position as the F-18 decayed from 97 to 3.2 kBq/ml. The authors obtained 10, 1-minute scans for each activity contrast level, alternating among 3 acquisitions: 2D mode with standard maximum ring difference (MRD=7), standard 3D mode (MRD=22), and 3D mode with MRD=13 (3D*). Images from 2D and 3D acquisitions were reconstructed by filtered backprojection and 3D reprojection (3DRP); 3D data were also reconstructed by FBP after Fourier rebinning (FORE+FBP). Sphere detectability was estimated using non-prewhitening (NPW) matched filtering to compute the detection signal-to-noise ratio, NPW SNR. In almost all cases, NPW-SNR was greater for 3D or 3D* than for 2D, although 2D outperformed 3D with 3DRP reconstruction at the earliest time points for 2 spheres located near opposite ends of the axial FOV; FORE+FBP reconstruction significantly improved the detectability of these spheres, compared to 3DRP, and demonstrated the expected near equivalence of 3D and 3D* data from spheres near the ends of the FOV. The authors results were not predictable from global NEC considerations alone.

Impact of acquisition geometry and patient habitus on lesion detectability in whole body FDG-PET: a channelized hotelling observer study

Although 3D septaless imaging has been shown to have advantages over 2D imaging under many conditions, it is not routinely used in the clinic in whole-body (WB) FDG-PET studies partly due to contradictory reports comparing 3D to 2D. The aim of this work was to evaluate the impact of the acquisition mode (2D vs 3D) on lesion detectability in WB FDG-PET studies for different lesion and patient sizes. Forty eight bed positions were acquired in thirty six patients, each in both 2D and 3D mode, 1-4 hours post-injection (740 MBq). Three spheres (1 cm, 1.3 cm and 1.6 cm diameter) containing FDG were also, imaged separately in air, at ten different locations corresponding to possible lesion sites in the 48 bed positions, also each in both 2D and 3D (480 tat-gets per condition). Each bed position was acquired for 7 min in 2D and 6 min in 3D and corrected for randoms. Sphere sinograms,were attenuated using the exact 2D or 3D attenuation map of the patient, after scaling 2D and 3D sinograms with identical factors to ensure marginal detectability. The resulting volumes were corrected for scatter and reconstructed using ordered subsets expectation maximization (OSEM) in 2D and Fourier rebinning (FORE)+OSEM in 3D. Next, 2D and 3D acquisition modes were compared on the basis of performance of a three-channel Hotelling observer (CHO) which incorporated internal noise in detecting the presence of a sphere of unknown size on an anatomic background. 3D imaging yielded better lesion detectability than 2D (p<0.025, two-tailed paired t-test) in patients of typical size (Body Mass Index <33). However 2D imaging yielded better lesion delectability than 3D in large patients (BMI>34) as 3D performance significantly deteriorated in large patients (P<0.05). Finally, 2D and 3D yielded similar results for different lesion sizes.

Evaluation of a Monte Carlo scatter correction in clinical 3D PET

Phantom and patient data were used to compare performance of a one-iteration Monte Carlo scatter correction (MC-SC-1i) for 3D PET, a vendor-supplied one-iteration single scatter model-based correction (SSS-1i) for 3D PET, unscatter-corrected 3D PET (No-SC), a SSS-1i followed by Monte Carlo scatter correction as a second iteration (MC-SSS) for 3D PET, and a convolution-subtraction scatter correction for 2D PET in terms of quantitative accuracy and lesion detectability. ROI analysis showed 2D PET images were more accurate than 3D, particularly for large phantoms, and MC-SSS corrected 3D PET images were more accurate than SSS-1i corrected 3D PET images for this data set. 2D and 3D PET images were reconstructed from 59 patient data sets. Bias of 3D PET images with respect to 2D images was determined using Corresponding Intensity Variance. 3D PET uncorrected images overestimated activity by /spl sim/50% (smallest patients) to /spl sim/150% (largest patients). The average absolute bias of SSS-1i corrected images (16%) was twice that of MC-SSS (8%) and more dependent on patient size. Lesion detection sensitivity in these patient images was evaluated using a Channelized Hotelling Observer. Scatter corrected 3D PET images performed /spl sim/10% better than uncorrected 3D PET images for smaller patients. Slightly better lesion sensitivity was seen for large patients in images reconstructed using SSS-1i (CHO-SNR=2.23/spl plusmn/0.29) compared to MC-SSS (2.08/spl plusmn/0.27) and uncorrected images (2.02/spl plusmn/0.23).

The effects of pulse pile-up on point-source measurements performed on different PET scanning devices

We have quantified resolution and contrast loss as a function of noise-equivalent count rate on the GSO-based Philips Allegro and the pixelated NaI CPET2 using previously developed methodology involving the NEMA NU 2-2001 70 cm test phantom and a Na-22 point source. We found that at peak NEC, neither system showed significant loss of resolution. However, after accounting for event loss due to dead time, loss of contrast was seen in both systems.