Identifying Defects without a priori Knowledge in a Room-Temperature Semiconductor Detector Using Physics Inspired Machine Learning Model

Room-temperature semiconductor radiation detectors (RTSD) such as CdZnTe are popular in Computed Tomography (CT) imaging and other applications. Transport properties and material defects with respect to electron and hole transport often need to be characterized, which is a labor intensive process. However, these defects often vary from one RTSD to another and are not known a priori during characterization of the material. In recent years, physics-inspired machine learning (PI-ML) models have been developed for the RTSDs which have the ability to characterize the defects in a RTSD by discretizing it volumetrically. These learning models capture the heterogeneity of the defects in the RTSD-which arises due to the fabrication process and the energy bands of elements in the RTSD. In those models, the different defects of RTSD-trapping, detrapping and recombination for electrons and holes-are present. However, these defects are often unknown. In this work, we show the capabilities of a PI-ML model which has been developed considering all the material defects to identify certain defects which are present (or absent). Additionally, these models can identify the defects over the volume of the RTSD in a discretized manner.

3D-Quantitated Single Photon Emission Computed Tomography/Computed Tomography: Impact on intended Management Compared to Lung Perfusion Scan in Marginal Candidates for Pulmonary Resection

OBJECTIVES

Based on previous studies, single-photon emission computed tomography/computed tomography (SPECT/CT) has been proven more accurate and reproducible than planar lung perfusion scintigraphy to assess lobar perfusion. However, the impact of 3D-quantitated SPECT/CT on intended management in functionally marginal candidates for pulmonary resection is unknown. The evaluation of this impact was the main aim of this study.

METHODS

Consecutive candidates for lung resection underwent preoperative evaluation according to ERS/ESTS Algorithm and underwent preoperative lung perfusion imaging. The lobar contribution to the total lung perfusion was estimated using established planar scintigraphic methods and 3-dimensional quantitative SPECT/CT method (CT Pulmo3D and xSPECT-Quant, Siemens). The difference in estimated lobar perfusion with resulting changes in predicted postoperative (ppo) lung function and extent of lung resection were analyzed to reveal possible changes in operability. In-hospital outcome was assessed.

RESULTS

One hundred twenty patients (46 females) were enrolled. The mean age (±SD) of patients was 68 ± 9 years, target lesions were in upper lobes in 57.7% and in lower lobes in 33.5%. The median FEV1 (forced expiratory volume in 1 second) was 70.5% (IQR 52-84) and median DLCO (diffusion capacity of lung for carbon monoxide) was 56.6% [47.1-67.4]. The planar posterior oblique method, compared to 3D-quantitated SPECT/CT, underestimated the perfusion of upper lobes by a median difference of 5% (right [2-9], left [2.5-8]; P = <.0001), while it overestimated the perfusion of lower lobes (left by 4% [2-7], right by 6% [2-9]; P = <.0001). In contrast to planar scintigraphy-based evaluation, 4 patients (3.3%), all with upper lobe lesions, were classified as inoperable when 3D-quantitated SPECT/CT was used for calculation of the ppo lung function.

CONCLUSIONS

In selected patients with upper lobe lesions, 3D-quantitated SPECT/CT would have changed the treatment strategy from operable to inoperable. Importantly, postoperative mortality in this particular subgroup was disproportionally high. 3D-quantitated SPECT/CT shall be further evaluated as it might improve preoperative risk stratification in functionally marginal candidates.

Method for comparison of data driven gating algorithms in emission tomography

Objective.Multiple algorithms have been proposed for data driven gating (DDG) in single photon emission computed tomography (SPECT) and have successfully been applied to myocardial perfusion imaging (MPI). Application of DDG to acquisition types other than SPECT MPI has not been demonstrated so far, as limitations and pitfalls of current methods are unknown.Approach.We create a comprehensive set of phantoms simulating the influence of different motion artifacts, view angles, moving objects, contrast, and count levels in SPECT. We perform Monte Carlo simulation of the phantoms, allowing the characterization of DDG algorithms using quantitative metrics derived from the data and evaluate the Center of Light (COL) and Laplacian Eigenmaps methods as sample DDG algorithms.Main results.View angle, object size, count rate density, and contrast influence the accuracy of both DDG methods. Moreover, the ability to extract the respiratory motion in the phantom was shown to correlate with the contrast of the moving feature to the background, the signal to noise ratio, and the noise in the data.Significance.We showed that reporting the average correlation to an external physical reference signal per acquisition is not sufficient to characterize DDG methods. Assessing DDG methods on a view-by-view basis using the simulations and metrics from this work could enable the identification of pitfalls of current methods, and extend their application to acquisitions beyond SPECT MPI.

Learning-based physical models of room-temperature semiconductor detectors with reduced data

Room-temperature semiconductor radiation detectors (RTSD) have broad applications in medical imaging, homeland security, astrophysics and others. RTSDs such as CdZnTe, CdTe are often pixelated, and characterization of these detectors at micron level can benefit 3-D event reconstruction at sub-pixel level. Material defects alongwith electron and hole charge transport properties need to be characterized which requires several experimental setups and is labor intensive. The current state-of-art approaches characterize each detector pixel, considering the detector in bulk. In this article, we propose a new microscopic learning-based physical models of RTSD based on limited data compared to what is dictated by the physical equations. Our learning models uses a physical charge transport considering trapping centers. Our models learn these material properties in an indirect manner from the measurable signals at the electrodes and/or free and/or trapped charges distributed in the RTSD for electron-hole charge pair injections in the material. Based on the amount of data used during training our physical model, our algorithm characterizes the detector for charge drifts, trapping, detrapping and recombination coefficients considering multiple trapping centers or as a single equivalent trapping center. The RTSD is segmented into voxels spatially, and in each voxel, the material properties are modeled as learnable parameters. Depending on the amount of data, our models can characterize the RTSD either completely or in an equivalent manner.

ROC study and SUV threshold using quantitative multi-modal SPECT for bone imaging

Background

We investigated the clinical performance of a quantitative multi-modal SPECT/CT reconstruction platform for yielding radioactivity concentrations of bone imaging with ^99mTc-methylene diphosphonate (MDP) or ^99mTc-dicarboxypropane diphosphonate (DPD). The novel reconstruction incorporates CT-derived tissue information while preserving the delineation of tissue boundaries. We assessed image-based reader concordance and confidence, and determined lesion classification and SUV thresholds from ROC analysis.

Methods

Seventy-two cancer patients were scanned at three US and two German clinical sites, each contributing two experienced board-certified nuclear medicine physicians as readers. We compared four variants of the reconstructed data resulting from the Flash3D (F3D) and the xSPECT Bone™ (xB) iterative reconstruction methods and presented images to the readers with and without a fused CT, resulting in four combinations. We used an all-or-none approach for inclusion, compiling results only when a reader completed all reads in a subset. After the final read, we conducted a “surrogate truth” reading, presenting all data to each reader. For any remaining discordant lesions, we conducted a consensus read. We next undertook ROC analysis to determine SUV thresholds for differentiating benign and lesional uptake.

Results

On a five-point rating scale of image quality, xB was deemed better by almost two points in resolution and one point better in overall acceptance compared to F3D. The absolute agreement of the rendered decision between the nine readers was significantly higher with CT information either inside the reconstruction (xB, xBCT) or simply through image fusion (F3DCT): 0.70 (xBCT), 0.67 (F3DCT), 0.64 (xB), and 0.46 (F3D). The confidence level to characterize the lesion was significantly higher (3.03x w/o CT, 1.32x w/CT) for xB than for F3D. There was high correlation between xB and F3D scores for lesion detection and classification, but lesion detection confidence was 41% higher w/o CT, and 21% higher w/CT for xB compared to F3D. Without CT, xB had 6.6% higher sensitivity, 7.1% higher specificity, and 6.9% greater AUC compared to F3D, and similarly with CT-fusion. The overall SUV-criterion (SUV_c) of xB (12) exceeded that for xSPECT Quant™ (xQ; 9), an approach not using the tissue delineation of xB. SUV critical numbers depended on lesion volume and location. For non-joint lesions > 6 ml, the AUC for xQ and xB was 94%, with SUV_c > 9.28 (xQ) or > 9.68 (xB); for non-joint lesions ≤ 6 ml, AUCs were 81% (xQ) and 88% (xB), and SUV_c > 8.2 (xQ) or > 9.1 (xB). For joint lesions, the AUC was 80% (xQ) and 83% (xB), with SUV_c > 8.61 (xQ) or > 13.4 (xB).

Conclusion

The incorporation of high-resolution CT-based tissue delineation in SPECT reconstruction (xSPECT Bone) provides better resolution and detects smaller lesions (6 ml), and the CT component facilitates lesion characterization. Our approach increases confidence, concordance, and accuracy for readers with a wide range of experience. The xB method retained high reading accuracy, despite the unfamiliar image presentation, having greatest impact for smaller lesions, and better localization of foci relative to bone anatomy. The quantitative assessment yielded an SUV-threshold for sensitively distinguishing benign and malignant lesions. Ongoing efforts shall establish clinically usable protocols and SUV thresholds for decision-making based on quantitative SPECT.

Quantification of 99mTc-DPD concentration in the lumbar spine with SPECT/CT

BACKGROUND

Routine single-photon emission computed tomography (SPECT) currently lacks quantitative information on regional activity concentration (ACC) of the injected tracer (e.g. kBq/ml). Furthermore, little is known on the skeletal absolute concentration of 99mTc-DPD after intravenous injection in bone scintigraphy. The aim of this study is to determine ACC in the healthy lumbar vertebrae of patients using a recently published quantitative SPECT/computed tomography (CT) protocol.

METHODS

Lumbar vertebrae ACC estimates were performed in 50 female patients (mean age 69.88 ± 13.73 years) who had been administered 562.84 ± 102.33 MBq of 99mTc-DPD and had undergone SPECT acquisition 4 h after the injection. The SPECT/CT system was calibrated against a well counter. Images were reconstructed with Flash3D. ACC and CT tissue density were measured in volumes of interest drawn over the spongious bone tissue of the three lower lumbar vertebral bodies when these exhibited no focal CT or SPECT pathology.

RESULTS

Average ACC measured in the normal spongious bone tissue was 48.15 ± 13.66 kBq/ml (95% confidence interval (CI) 45.81 to 50.50 kBq/ml). This corresponds to a mean standardised uptake value (SUV) of (5.91 ± 1.54) (95% CI (5.64 to 6.17) SUV). SUV correlated significantly with Hounsfield units (HU) (r = 0.678, p < 0.0001). Significant negative correlations were observed between age and HU (r = -0.650, p < 0.0001) and between age and SUV (r = -0.385, p < 0.0001).

CONCLUSIONS

The SUVs determined for 99mTc-DPD uptake 4 h post injection are in the same range as those reported for [18F]fluoride in positron emission tomography. The strong correlation of SUV with bone CT density underlines the physiological significance of this variable. Our data suggest further investigation of the potential value of ACC measurement in the diagnosis of pathological conditions such as osteoporosis or in following up osseous metastases under therapy.

Quantitative accuracy of clinical 99mTc SPECT/CT using ordered-subset expectation maximization with 3-dimensional resolution recovery, attenuation, and scatter correction

We present a calibration method of a clinical SPECT/CT device for quantitative (99m)Tc SPECT. We use a commercially available reconstruction package including ordered-subset expectation maximization (OSEM) with depth-dependent 3-dimensional resolution recovery (OSEM-3D), CT-based attenuation correction, and scatter correction. We validated the method in phantom studies and applied it to images from patients injected with (99m)Tc-diphosponate.

METHODS

The following 3 steps were performed to derive absolute quantitative values from SPECT reconstructed images. In step 1, we used simulations to characterize the SPECT/CT system and derive emission recovery values for various imaging parameter settings. We simulated spheres of varying diameters and focused on the dependencies of activity estimation errors on structure size and position, pixel size, count density, and reconstruction parameters. In step 2, we cross-calibrated our clinical SPECT/CT system with the well counter using a large cylinder phantom. This step provided the mapping from image counts to kBq/mL. And in step 3, correction factors from steps 1 and 2 were applied to reconstructed images. We used a cylinder phantom with variable-sized spheres for verification of the method. For in vivo validation, SPECT/CT datasets from 16 patients undergoing (99m)Tc-diphosponate SPECT/CT examinations of the pelvis including the bladder were acquired. The radioactivity concentration in the patients' urine served as the gold standard. Mean quantitative accuracy and SEs were calculated.

RESULTS

In the phantom experiments, the mean accuracy in quantifying radioactivity concentration in absolute terms was within 3.6% (SE, 8.0%), with a 95% confidence interval between -19.4% and +12.2%. In the patient studies, the mean accuracy was within 1.1% (SE, 8.4%), with a 95% confidence interval between -15.4% and +17.5%.

CONCLUSION

Current commercially available SPECT/CT technology using OSEM-3D reconstruction, scatter correction, and CT-based attenuation correction allows quantification of (99m)Tc radioactivity concentration in absolute terms within 3.6% in phantoms and 1.1% in patients with a focus on the bladder. This opens up the opportunity of SPECT quantitation entering the routine clinical arena. Still, the imprecision caused by unavoidable measurement errors is a dominant factor for absolute quantitation in a clinical setup.

Effect of CT-based attenuation correction on uptake ratios in skeletal SPECT

The AIM of this study was to determine the clinical relevance of compensating SPECT data for patient specific attenuation by the use of CT data simultaneously acquired with SPECT/CT when analyzing the skeletal uptake of polyphosphonates (DPD). Furthermore, the influence of misregistration between SPECT and CT data on uptake ratios was investigated.

METHODS

Thirty-six data sets from bone SPECTs performed on a hybrid SPECT/CT system were retrospectively analyzed. Using regions of interest (ROIs), raw counts were determined in the fifth lumbar vertebral body, its facet joints, both anterior iliacal spinae, and of the whole transversal slice. ROI measurements were performed in uncorrected (NAC) and attenuation-corrected (AC) images. Furthermore, the ROI measurements were also performed in AC scans in which SPECT and CT images had been misaligned by 1 cm in one dimension beforehand (ACX, ACY, ACZ).

RESULTS

After AC, DPD uptake ratios differed significantly from the NAC values in all regions studied ranging from 32% for the left facet joint to 39% for the vertebral body. AC using misaligned pairs of patient data sets led to a significant change of whole-slice uptake ratios whose differences ranged from 3,5 to 25%. For ACX, the average left-to-right ratio of the facet joints was by 8% and for the superior iliacal spines by 31% lower than the values determined for the matched images (p < 0.05).

CONCLUSIONS

AC significantly affects DPD uptake ratios. Furthermore, misalignment between SPECT and CT may introduce significant errors in quantification, potentially also affecting left-to-right ratios. Therefore, at clinical evaluation of attenuation-corrected scans special attention should be given to possible misalignments between SPECT and CT.