Anatomically derived attenuation coefficients for use in quantitative single photon emission tomography studies of the thorax.

Elimination of errors due to poor attenuation correction is an essential part of any quantitative single photon emission tomography (SPET) technique. Attenuation coefficients (mu Tc) for use in attenuation correction of SPET data were determined using technetium 99m and cobalt 57 flood sources and using topographical information obtained from computed tomography (CT) scans and magnetic resonance (MR) images. In patients with carcinoma of the bronchus, the mean attenuation coefficient for 99mTc was 0.096 cm-1 when determined across a transverse section of the thorax at the level of the tumour by means of a 57Co flood source (13 patients) and 0.093 and 0.074 cm-1 as determined from CT scans for points in the centre of the tumour and contralateral normal lung, respectively (21 patients). In 18 patients with breast tumours, the mean attenuation coefficient for 99mTc was 0.110 and 0.076 cm-1 when determined from MRI cross-sections for points in the centre of the tumour and normal contralateral lung, respectively. This indicates significant overcorrection for attenuation when the conventional value of 0.12 cm-1 is used. A value in the range 0.08-0.09 cm-1 would be more appropriate for SPET studies of the thorax. An alternative approach to quantitative region of interest (ROI) analysis is to perform attenuation correction appropriate to the centre of each ROI (using topographical information derived from CT or MRI) on non-attenuation-corrected reconstructions.     

Accuracy of numerically produced compensators.

A feasibility study is performed to assess the utility of a computer numerically controlled (CNC) mill to produce compensating filters for conventional clinical use and for the delivery of intensity-modulated beams. A computer aided machining (CAM) software is used to assist in the design and construction of such filters. Geometric measurements of stepped and wedged surfaces are made to examine the accuracy of surface milling. Molds are milled and filled with molten alloy to produce filters, and both the molds and filters are examined for consistency and accuracy. Results show that the deviation of the filter surfaces from design does not exceed 1.5%. The effective attenuation coefficient is measured for CadFree, a cadmium-free alloy, in a 6 MV photon beam. The effective attenuation coefficients at the depth of maximum dose (1.5 cm) and at 10 cm in solid water phantom are found to be 0.546 cm-1 and 0.522 cm-1, respectively. Further attenuation measurements are made with Cerrobend to assess the variations of the effective attenuation coefficient with field size and source-surface distance. The ability of the CNC mill to accurately produce surfaces is verified with dose profile measurements in a 6 MV photon beam. The test phantom is composed of a 10 degrees polystyrene wedge and a 30 degrees polystyrene wedge, presenting both a sharp discontinuity and sloped surfaces. Dose profiles, measured at the depth of compensation (10 cm) beneath the test phantom and beneath a flat phantom, are compared to those produced by a commercial treatment planning system. Agreement between measured and predicted profiles is within 2%, indicating the viability of the system for filter production.     

Lower accuracy of Tl-201 SPECT in women is not improved by size-based normal databases or Wiener filtering

BACKGROUND: We have shown that the diagnostic accuracy of quantitative single photon emission computed tomography (SPECT) thallium 201 myocardial perfusion imaging is lower in women than in men and that much of the difference can be explained by the smaller size of the left ventricle in women. Therefore attempts at improving the accuracy of myocardial perfusion imaging in women should focus on the problem of lower accuracy in patients with small chamber size. We evaluated two strategies for this: size- and gender-based normal databases and inverse filtering with the Wiener filter. METHODS AND RESULTS: We identified 618 patients undergoing exercise SPECT TI-201 who either had a low pre-test probability of coronary artery disease or had catheterization-documented disease. Their images were analyzed on the basis of gender and chamber size: both gender and size- and gender-based normal databases were created. The studies were analyzed quantitatively, and the accuracy was evaluated by use of the area under the receiver operating characteristic (ROC) curve. Chamber size was significantly lower in women (size index 69+/-22 women vs 96+/-28 men; P < .0001). The accuracy of myocardial perfusion imaging was lower in women compared with men (ROC area: 0.92+/-0.01 men vs 0.85+/-0.03 women; P = .03), and there was an even greater difference in accuracy between patients with large versus small chamber size (ROC area: 0.94+/-0.01 large vs 0.81+/-0.03 small; P < .001). There was no improvement in the diagnostic accuracy either in women or in patients with small chamber size when a size- and gender-based normal database, Wiener filter, or the Wiener filter with a size- and gender-based normal database was used. CONCLUSION: The left ventricular chamber size in women is smaller than that in men. There is a significant difference in the accuracy of quantitative SPECT TI-201 between men and women and an even greater difference between patients with large versus small chamber size. Neither size- and gender-based databases nor Wiener filtering significantly improves accuracy in women or in patients with small chamber size.     

Semiquantitative analysis of SPECT images using 99Tc(m) -HMPAO in the treatment of brain perfusion after the attenuation correction by the Chang method and the application of the Butterworth filter

The aim of this work was to optimize the parameters for semiquantitative analysis with regard to the average number of counts per pixel in tomographic projections. The analysed studies were divided into three groups, proportionally, to average the counts per pixel. The analysis of results proved the dependence (P<0.05) between the average geometric contrast of images with the maximum counts per pixel higher than 500, reconstructed with the cut-off frequency fc=0.50fn, and the standard image. Also, the same dependence was found between the group with an average of 200 counts per pixel and the standard image at the cut-off frequency fc=0.40fn. The analysis of the attenuation coefficient, c, showed significant differences. The attenuation coefficient c=0.12 cm-1 as accepted for further studies. Optimization of the matrix order of interpolation filter proved increasing of contrast with decreasing size of the matrix. In conclusion, according to the average number of counts per pixel in a projection the parameters of image reconstruction, particularly the cut-off frequency fc of the Butterworth filter, should be changed. Additionally, an attenuation correction with coefficient c=0.12 cm-1 and interpolation with matrix size equal 7x7 should be applied.     

Esophageal source measurement of Tc-99m attenuation coefficients for use in left ventricular volume determinations

A linear attenuation coefficient for water (mu = .15 cm-1) at 140 keV has been used in the determination of left ventricular volumes (LVV) by attenuation-corrected equilibrium methods. This theoretical value ignores the effect of Compton scatter and thus may be too high for human LVV determinations. The effective attenuation coefficient, mu', of the human chest was determined in ten normal volunteers using a Tc-99m esophageal source imaged with a gamma camera. Values for mu' at 30 degrees LAO in end-expiration, quiet breathing, and end-inspiration were .125 +/- .006 cm-1, .125 +/- .005 cm-1, and .113 +/- .007 cm-1, respectively (95% confidence interval). Values of mu' at 45 degrees LAO were .122 +/- .006 cm-1, .119 +/- .007 cm-1, and .099 +/- .009 cm-1, respectively, for the same conditions. The measured value of mu' for the source in a water phantom was .127 +/- .001 cm-1. This suggests that a value of mu' of .125 cm-1 may be appropriate for use in determining LVV in patients.     

The effect of source size on the buildup factor calculation of absolute volume

A general scheme for generating attenuation-corrected images for use in absolute volume measurements has been developed. The technique is based on a buildup factor approach for Compton scatter compensation and requires anterior and posterior count-rate measurements. The scatter correction requires evaluation of the attenuation factor (1-(1-e-mu d)B(infinity), where mu is the linear attenuation coefficient in cm-1 and B(infinity) is the buildup factor at infinite depth. The attenuation factors for four different source sizes using 99mTcO4- and a 20% scintillation camera energy window are reported. The results indicated that B(infinity) was constant while mu varied as a function of size (S) in pixels according to mu = 0.151 cm-1 X exp(-1.18 X 10(-4) pixels-1 X S). Once the appropriate value of mu was determined, a pair of anterior-posterior count rate equations was used to generate attenuation-corrected count rate data for use in absolute volume measurements. The method was validated by calculating three separate phantom volumes. The results showed that the method provides less than +/- 6.0% error for volume determinations at all investigated depths.     

Quantitation of iodine-123-beta-CIT dopamine receptor uptake in a phantom model.

OBJECTIVE: The purpose of this study was to determine the effects of technical factors such as collimation and filtration on the measurement of 123I-beta-CIT uptake in the striatum. METHODS: All SPECT studies were performed using a brain phantom containing striata within a bone- and tissue-equivalent skull. The effects of collimator resolution and septal penetration were assessed from 99mTc and 123I studies containing variable activities in the striata and background regions. Optimum attenuation coefficients (mu) were determined from studies containing uniform activity in the brain. RESULTS: For 99mTc, mu was 0.095 cm-1 and 0.07 cm-1 for parallel-hole and fanbeam collimators, respectively. For 123I, these values dropped to 0.09 cm-1 and 0.00 cm-1 (zero) for medium-energy and fanbeam collimators, respectively. Striatal uptake was significantly underestimated, particularly for medium-energy and general-purpose collimators. With 99mTc, fanbeam collimation gave a 50% increase in the measured striatal uptake, compared to medium-energy collimation. However, with 123I, this gain was eliminated by septal penetration and scatter. Increasing transaxial slice thickness, ROI size and decreasing filter cutoff frequency all degraded apparent striatal uptake. CONCLUSION: Partial volume effects, combined with the averaging effects of increasing slice thickness and ROI size, are the most significant factors affecting measurement of striatal uptake of 123I-beta-CIT. The increased resolution of low-energy high-resolution collimators, compared to a medium-energy collimator, is offset by the increased septal penetration and scatter.     

Quantitation in planar renal scintigraphy: which mu value should be used?

The attenuation coefficient value mu used by different authors for quantitation in planar renal scintigraphy varies greatly, from the theoretical value of 0.153 cm-1 (appropriate for scatter-free data) down to 0.099 cm-1 (empirical value assumed to compensate for both scatter and attenuation). For a 6-cm-deep kidney, such variations introduce up to 30% differences in absolute measurement of kidney activity. Using technetium-99m phantom studies, we determined the mu values that would yield accurate kidney activity quantitation for different energy windows corresponding to different amounts of scatter, and when using different image analysis approaches similar to those used in renal quantitation. With the 20% energy window, it was found that the mu value was strongly dependent on the size of the region of interest (ROI) and on whether background subtraction was performed: the mu value thus varied from 0.119 cm-1 (loose ROI, no background subtraction) to 0.150 cm-1 (kidney ROI and background subtraction). When using data from an energy window that could be considered scatter-free, the mu value became almost independent of the image analysis scheme. It is concluded that: (1) when performing background subtraction, which implicitly reduces the effect of scatter, the mu value to be used for accurate quantitation is close to the theoretical mu value; (2) if the acquired data were initially corrected for scatter, the appropriate mu value would then be the theoretical mu value, whatever the image analysis scheme.     

Optimum processing protocols for volume determination of the liver and spleen from SPECT imaging with technetium-99m sulfur colloid

A study of the effects of processing parameters on the determination of liver and spleen volume from SPECT data was performed. A method for volume determination using a threshold algorithm was calibrated against phantoms and applied to 60 patient studies. Good reproducibility was found using different projections and computing the volume on separate days. Variations of the measured volumes with the threshold value, reconstruction filter cutoff frequency and attenuation correction were investigated. Reconstruction parameters producing best image quality were also determined. A threshold of 25% of the maximum value in the organ was determined from phantom studies. Changes of 1% around this value yielded changes of 2-3% in the computed volume. No significant change was noted as cutoff frequencies varied between 0.4 and 0.85 of Nyquist (0.031 to 0.066 cycles/cm) for a third order Butterworth filter. Attenuation correction produced a decrease of 9% and 6% in liver and spleen measured volume respectively. Best image quality was obtained with 0.4 Nyquist (0.031 cycles/cm) cutoff frequency for third order Butterworth filter and attenuation correction. It is concluded that optimal parameters must be determined for any processing protocol, and must then be adhered to in future applications to insure clinical accuracy, especially those parameters demonstrating the most quantitative and qualitative sensitivity.     

Simulating technetium-99m cerebral perfusion studies with a three-dimensional Hoffman brain phantom: Collimator and filter selection in SPECT neuroimaging

The choice of a collimator and the selection of a filter can affect the quality of clinical SPECT images of the brain. The compromises that 4 different collimators make between spatial resolution and sensitivity were studied by imaging a three-dimensional Hoffman brain phantom. The planar data were acquired with each collimator on a three-headed SPECT system and were reconstructed with both a standard Butterworth filter and a Wiener pre-filter. The reconstructed images were then evaluated by specialists in nuclear medicine and were also quantitatively analyzed with specific regions of interest (ROI) in the brain. All observers preferred the Wiener filter reconstructed images regardless of the collimator used to acquire the planar images. With this filter, the ultrahigh-resolution fan-beam collimator was the most subjectively preferable and quantitatively produced the highest contrast ratios. The findings support suggestions that higher resolution collimators are preferable to higher sensitivity collimators, and indicate that fan-beam collimators are preferable to parallel-hole collimators for clinical SPECT studies of cerebral perfusion. The results also suggest that the Wiener filter enhances the quality of SPECT brain images regardless of which collimator is used to acquire the data.     

Pre- to postnatal reduction in ultrasound attenuation coefficient of the liver

A recent study showed the ultrasound attenuation coefficient of fetal liver between 26 and 40 weeks of gestation to be 26% higher than after birth. To test the hypothesis that ultrasound attenuation is sensitive to fetal liver glycogen concentration, the livers of 24 fetuses were examined at 5 MHz just prior to and just after birth. The mean pre- to post-delivery reduction in attenuation coefficient was 0.08 dB cm-1 MHz-1 +/- 0.02 (SEM), or 17% of the post-delivery mean. This is consistent with the increase in attenuation measured by others in liver homogenate when glycogen was added. An increase in measurement accuracy, correlation with glycogen content, and, possibly, control for biological variability will be required to make predictions in individual cases, as opposed to these averages. A simple test of glycogen content would be of value scientifically and in prenatal and postnatal management.     

Common artifacts in PET myocardial perfusion images due to attenuation-emission misregistration: clinical significance, causes, and solutions.

Misregistration between attenuation and emission images causes artifactual abnormalities on cardiac PET images that result in false-positive defects. This study determines the frequency and mechanisms of misregistration artifacts, identifies their predictors, and validates a method for their routine clinical identification, prevention, or correction. METHODS: We performed 1177 consecutive diagnostic myocardial perfusion PET studies using 1 of 3 protocols: (a). 3 initial consecutive measured attenuation correction (MAC) scans, followed by resting and dipyridamole emission scans; (b). an initial MAC scan (early MAC), followed by emission scans; and (c). a MAC attenuation scan obtained after emission scans (late MAC). Emission images were manually shifted to obtain coregistration with attenuation and reconstructed again using shifted emission data that eliminated artifactual defects. Measurements on PET images included heart size, heart and diaphragm displacement after dipyridamole, objective quantitative misregistration of attenuation and emission images, and size or severity of image defects before and after shifting emission images. RESULTS: Of 1,177 rest-dipyridamole PET perfusion studies, 252 (21.4%) had artifactual defects due to attenuation-emission misregistration. By shifting emission images, quantitative severity and size of misregistration and artifactual defects significantly decreased (P < 0.001) with visual normalization. Artifactual defects were predicted by horizontal plane misregistration (odds ratio [OR] = 1.545, confidence intervals [CI] = 1.113-2.145, P = 0.009), body mass index (OR = 2.659, CI = 1.032-6.855, P = 0.043), and whole heart area in the horizontal plane at rest (OR = 1.096, CI = 1.018-1.179, P = 0.015). Quantitative misregistration was predicted by diaphragm displacement between rest and dipyridamole (P = 0.001, CI = 0.158-0.630), body mass index (P = 0.005, CI = 0.202-1.124), and whole heart area in the horizontal plane at rest (P = 0.004, CI = -0.144 to -0.028). Diaphragm displacement was significantly larger for obese compared with lean patients (P = 0.027) during the initial 10 min of the imaging protocol. CONCLUSION: Misregistration of attenuation and emission images is common in cardiac PET imaging and causes artifactual defects predicted by diaphragmatic displacement, body mass index, and heart size. Multiattenuation imaging sequences and manual, visually optimized coregistration of attenuation and emission images substantially eliminate artifacts for reliably identifying mild perfusion defects of early nonobstructive coronary atherosclerosis as the basis for intense lifestyle and pharmacologic treatment.     

Scatter, spatial resolution, and quantitative recovery in high resolution SPECT.

The potential use of single photon emission CT (SPECT) for quantification depends on its physical performance characteristics. We investigated the performance of a high resolution four-head brain SPECT scanner (Neuro-Spect; Summit/Hitachi). With an attenuation coefficient of 0.11 cm-1 and the Chang correction method, the calibration factor of the scanner was 515 (cpm/ml)/(microCi/ml) and showed only a minimal but systematic dependence on object size. Without scatter, the resolution was 4.7 mm (full width at half-maximum); in a scatter medium, the resolution was 5.3-10.0 mm with high resolution collimation and 7.7-18.8 mm with general purpose collimation, depending on filtering. A recovery coefficient of close to 100% was measured in the center of spheres with a size of at least 20 mm placed in a cylindrical water-filled phantom. In lesions of this size, scatter was 20%. We conclude from our measurements that the investigated high resolution SPECT offers significant improvements in resolution, scatter, and recovery, which will improve both the quality of brain studies and the measurement of quantitative parameters such as the cortex/white matter ratio.     

Assessment of polyp and mass histopathology by intravenous contrast-enhanced CT colonography

RATIONALE AND OBJECTIVES: We sought to demonstrate that intravenous contrast-enhanced CT colonography (CTC) can distinguish colonic adenomas from carcinomas. METHODS: Supine intravenous contrast-enhanced CTC with colonoscopic and/or surgical correlation was performed on 25 patients with colonic adenomas or carcinomas. Standard deviation of mean polyp CT attenuation was computed and assessed using ANOVA and receiver-operating characteristic analyses. RESULTS: Colonoscopy confirmed 32 polyps or masses 1 to 8 cm in size. The standard deviations of CT attenuation were carcinomas (n = 13; 36 +/- 6 HU; range 28-48 HU) and adenomas (n = 19; 49 +/- 14 HU; range 31-100 HU) (P = 0.005). At a standard deviation threshold of 42 HU, the sensitivity and specificity for classifying a polyp or mass as a carcinoma were 92% and 79%, respectively. The area under the receiver-operating characteristic curve was 0.89 +/- 0.06 (95% confidence interval 0.73-0.96). CONCLUSIONS: Measurement of the standard deviation of CT attenuation on intravenous contrast-enhanced CTC permits histopathologic classification of polyps 1 cm or larger as carcinomas versus adenomas. The presence of ulceration or absence of muscular invasion in carcinomas creates overlap with adenomas, reducing the specificity of carcinoma classification.     

Integration of diffusion-weighted MRI data and a simple mathematical model to predict breast tumor cellularity during neoadjuvant chemotherapy

Diffusion-weighted magnetic resonance imaging data obtained early in the course of therapy can be used to estimate tumor proliferation rates, and the estimated rates can be used to predict tumor cellularity at the conclusion of therapy. Six patients underwent diffusion-weighted magnetic resonance imaging immediately before, after one cycle, and after all cycles of neoadjuvant chemotherapy. Apparent diffusion coefficient values were calculated for each voxel and for a whole tumor region of interest. Proliferation rates were estimated using the apparent diffusion coefficient data from the first two time points and then used with the logistic model of tumor growth to predict cellularity after therapy. The predicted number of tumor cells was then correlated to the corresponding experimental data. Pearson's correlation coefficient for the region of interest analysis yielded 0.95 (P = 0.004), and, after applying a 3 × 3 mean filter to the apparent diffusion coefficient data, the voxel-by-voxel analysis yielded a Pearson correlation coefficient of 0.70 ± 0.10 (P < 0.05).     

Computer-assisted detection for CT colonography: external validation

AIM: To externally validate a computer-assisted detection (CAD) system for computed tomography (CT) colonography, using data from a single centre uninvolved with the software development. MATERIALS AND METHODS: Twenty-five multi-detector CT colonography examinations of patients with validated polyps accumulated at a single centre were examined by two readers who used endoscopic and histopathological data to identify polyp coordinates. A CAD system that had been developed using data from elsewhere, and had not previously encountered the present data, was then applied to the data at sphericity filter settings of 0.75 and 0.50 and identified potential polyps. True-positive, false-negative, and false-positive counts were determined by comparison with the known polyp coordinates. RESULTS: Twenty-five patients had 57 polyps, median size 6mm (range 1-15mm). Per-patient sensitivity for the CAD system was 96% (24 of 25). The CAD system detected 44 (77%) polyps at sphericity setting 0.75 and 49 (86%) polyps at sphericity 0.50: the additional five polyps detected all measured 5mm or less. Sphericity of 0.75 resulted in a median of 10 (one to 34) easily dismissed false-positive prompts per patient and a median of 4 (zero to 15) that needed three-dimensional rendering before dismissal. This rose to 32 (16 to 99) and 11 (three to 35), respectively, at sphericity 0.5. CONCLUSIONS: A per-patient sensitivity of 96% was found for the CAD system (in patients with a median polyp diameter of 6mm) using external validation, a more stringent test than either internal cross-validation or temporal validation. Decreasing sphericity increases sensitivity for small polyps at the expense of decreased specificity.     

The application of a masked orbiting transmission source for attenuation correction in PET

A new technique for attenuation correction in positron emission tomography is introduced and evaluated. Transmission scans are performed with a point source of 68Ge encapsulated in a lead collimator that masks the source into a fan beam in the scanning plane. The source orbits the patient section at the edge of the slice defining collimator. Only events acquired by detector pairs that are collinear with the source are used to calculate the attenuation coefficients. Events from detector pairs that are nearly collinear are rejected, while those from detector pairs that are far from collinear may be used to acquire a simultaneous emission scan. The coincident event rate per unit source activity is over twice that of rod and ring sources. This technique is compared with calculated outline and ring source attenuation correction techniques in a pie phantom. The linear attenuation coefficient for water was measured as 0.096 cm-1, and 0.094 cm-1 when the water contained 12 kBq/cc 68Ga, compared with 0.085 cm-1 for a ring source. Cerebral glucose utilization rates in a normal volunteer reconstructed with transmission scans performed pre- and postinjection of fluorodeoxyglucose show no significant differences. However, values of cortical glucose utilization average 12% above those measured with the fitted outline method in the highest cuts because of the obliqueness of the skull to the planes examined.     

Correction for attenuation in technetium-99m-HMPAO SPECT brain imaging.

We present a correction technique that uses the effective bone and tissue attenuation coefficients to compensate 99mTc-HMPAO brain SPECT projections for attenuation. Transverse images of a human skull filled with a uniform mixture of 99mTc and gelatin have a greater count density at the center with respect to the periphery when corrected for attenuation with the effective water/tissue coefficient of 0.12 cm-1. An attenuation coefficient of 0.09 cm-1 produces uniform images at the expense of a reduced count density. Additional experiments with phantoms wrapped with aluminum (to simulate bone) indicate that the greater count density at the image center is a result of increased attenuation at the edges of the projections where there is a greater path length through the aluminum (or bone). SPECT projections explicitly corrected for both bone and soft-tissue attenuation result in images of improved uniformity and increased count density.     

Attenuation-corrected 99mTc-MIBI SPECT in overweight patients with chronic ischaemic dysfunction: a comparison to NH3 PET and implications for the diagnosis of myocardial viability

OBJECTIVE: We determined the value of attenuation correction (AC) of myocardial perfusion estimation with (99m)Tc-MIBI SPECT in overweight patients by comparison of uncorrected (filtered back-projection (FBP) and corrected (an iterative algorithm with a measured attenuation coefficients map (FL-AC)) (99m)Tc-MIBI relative uptake to perfusion data obtained in the same patients with NH3 PET. In addition, the impact of attenuation correction for the assessment of myocardial viability with (99m)Tc-MIBI SPECT was determined using FDG PET as the reference method. METHODS: Thirty consecutive overweight patients (BMI=28+/-4) with left ventricular dysfunction underwent a resting (99m)Tc-MIBI SPECT and a PET study (NH3 and FDG). (99m)Tc-MIBI SPECT scans were reconstructed without attenuation correction (FBP) and with attenuation correction (FL-AC). The left ventricle was divided into 16 segments, in which the relative uptake was quantified using circumferential profiles. A relative uptake > or = 60% was considered consistent with viable myocardium for FDG and MIBI. RESULTS: The absolute difference between (99m)Tc-MIBI SPECT and NH3 PET uptakes was less pronounced in the inferior (12+/-10% vs. 17+/-12%, P<0.001), anteroseptal (12+/-11% vs. 16+/-12%, P=0.009) and septal (15+/-12% vs. 18+/-14%, P=0.003) regions (FL-AC vs. FBP, respectively). The sensitivity of MIBI for diagnosing myocardial viability increased from 83 to 100% (P=0.034), without loss in specificity. CONCLUSION: Attenuation correction improves myocardial perfusion estimation by (99m)Tc-MIBI SPECT in the inferior, anteroseptal and septal regions and increases its sensitivity for the diagnosis of myocardial viability.     

Imaging Biomarkers of Hypothyroidism on Lung Cancer Screening CT

PurposeTo assess the feasibility of attenuation and size measurement of the thyroid gland as an imaging biomarker for hypothyroidism in patients undergoing lung cancer screening (LCS) with low dose CT.Materials and MethodsWith institutional review board (IRB) approval, we retrospectively reviewed all patients with LCS CT between September 1, 2016 and March 31, 2020, who had at least 1 thyroid-stimulating hormone (TSH) test within 90 days of the patient's most recent screening CT. Hypothyroid patients were identified through billing diagnosis and/or elevated TSH or those on treatment with thyroxine; normal patients were identified as those without a diagnosis of hypothyroidism and normal TSH. For each hypothyroid patient, an age- and sex-matched normal control was included. The diameters and attenuation of both lobes of the thyroid gland were measured for each case; patients in whom the thyroid gland could not be seen to measure were excluded.ResultsA total of 304 patients were included. The areas under the receiver operating characteristic curve for size and attenuation of the left lobe were 0.774 (95% confidence interval [CI] 0.714-0.825) and 0.812 (95% CI 0.759-0.861), respectively; and for the right lobe were 0.776 (95% CI 0.719-0.827) and 0.794 (95% CI 0.740-0.847), respectively. We developed a decision tree algorithm to predict hypothyroidism combining the minimum size and attenuation of either lobe of the thyroid gland, with sensitivity, specificity, and accuracy of 76%, 87%, and 82%, respectively.ConclusionSize and attenuation of the thyroid gland can be used to identify potential hypothyroid patients undergoing LCS.