Head and neck squamous cell carcinoma: evaluation of iodine overlay maps and low-energy virtual mono-energetic images acquired with spectral detector CT.

AIM: To evaluate the diagnostic value of spectral detector computed tomography (SDCT)-derived iodine overlay maps and low-energy virtual mono-energetic images (VMI) for the initial locoregional assessment of primary, therapy-naive head and neck cancer. MATERIALS AND METHODS: Fifty-six patients with histologically confirmed untreated squamous cell carcinoma of the head and neck who underwent SDCT of the neck for staging purposes were included in this retrospective study. Attenuation, image noise as well as signal- and contrast-to-noise ratios (S-/CNR) in VMI40-70keV were obtained from region of interest (ROI)-based measurements in the tumour and important anatomical landmarks (sternocleidomastoid muscle, subcutaneous fat, thyroid gland, submandibular gland, carotid artery, and jugular vein). Tumour conspicuity and delineation, as well as subjective image quality, were rated for conventional images, VMI40-70keV, and iodine overlay maps using five-point Likert scales. RESULTS: The CNR of the tumour versus the floor of the mouth and the CNR of the tumour versus the sternocleidomastoid muscle was significantly higher in VMI40keV in comparison to conventional images (10.0 ± 7.3 versus 3.8 ± 3.3 and 11.3 ± 7.6 versus 3.6 ± 2.8; p<0.05 each). This was supported by qualitative results, as tumour conspicuity and delineation received superior ratings in iodine overlay maps and VMI40keV compared to conventional images (5 [3-5] and 5 [4-5] versus 3 [2-5]; 5 [2-5] and 5 [3-5] versus 3 [2-4], respectively, all p<0.05). VMI40keV yielded the highest score among all included image reconstructions for overall image quality (p<0.05 all). CONCLUSION: Iodine overlay maps and low-energy VMI derived from SDCT improve initial assessment of primary squamous cell carcinoma of the head and neck compared to conventional images.

Automated Color-Coding of Lesion Changes in Contrast-Enhanced 3D T1-Weighted Sequences for MRI Follow-up of Brain Metastases.

BACKGROUND AND PURPOSE: MR imaging is the technique of choice for follow-up of patients with brain metastases, yet the radiologic assessment is often tedious and error-prone, especially in examinations with multiple metastases or subtle changes. This study aimed to determine whether using automated color-coding improves the radiologic assessment of brain metastases compared with conventional reading. MATERIALS AND METHODS: One hundred twenty-one pairs of follow-up examinations of patients with brain metastases were assessed. Two radiologists determined the presence of progression, regression, mixed changes, or stable disease between the follow-up examinations and indicated subjective diagnostic certainty regarding their decisions in a conventional reading and a second reading using automated color-coding after an interval of 8 weeks. RESULTS: The rate of correctly classified diagnoses was higher (91.3%, 221/242, versus 74.0%, 179/242, P < .01) when using automated color-coding, and the median Likert score for diagnostic certainty improved from 2 (interquartile range, 2-3) to 4 (interquartile range, 3-5) (P < .05) compared with the conventional reading. Interrater agreement was excellent (κ = 0.80; 95% CI, 0.71-0.89) with automated color-coding compared with a moderate agreement (κ = 0.46; 95% CI, 0.34-0.58) with the conventional reading approach. When considering the time required for image preprocessing, the overall average time for reading an examination was longer in the automated color-coding approach (91.5 [SD, 23.1] seconds versus 79.4 [SD, 34.7 ] seconds, P < .001). CONCLUSIONS: Compared with the conventional reading, automated color-coding of lesion changes in follow-up examinations of patients with brain metastases significantly increased the rate of correct diagnoses and resulted in higher diagnostic certainty.

Automated Detection and Segmentation of Brain Metastases in Malignant Melanoma: Evaluation of a Dedicated Deep Learning Model.

BACKGROUND AND PURPOSE: Malignant melanoma is an aggressive skin cancer in which brain metastases are common. Our aim was to establish and evaluate a deep learning model for fully automated detection and segmentation of brain metastases in patients with malignant melanoma using clinical routine MR imaging. MATERIALS AND METHODS: Sixty-nine patients with melanoma with a total of 135 brain metastases at initial diagnosis and available multiparametric MR imaging datasets (T1-/T2-weighted, T1-weighted gadolinium contrast-enhanced, FLAIR) were included. A previously established deep learning model architecture (3D convolutional neural network; DeepMedic) simultaneously operating on the aforementioned MR images was trained on a cohort of 55 patients with 103 metastases using 5-fold cross-validation. The efficacy of the deep learning model was evaluated using an independent test set consisting of 14 patients with 32 metastases. Manual segmentations of metastases in a voxelwise manner (T1-weighted gadolinium contrast-enhanced imaging) performed by 2 radiologists in consensus served as the ground truth. RESULTS: After training, the deep learning model detected 28 of 32 brain metastases (mean volume, 1.0 [SD, 2.4] cm3) in the test cohort correctly (sensitivity of 88%), while false-positive findings of 0.71 per scan were observed. Compared with the ground truth, automated segmentations achieved a median Dice similarity coefficient of 0.75. CONCLUSIONS: Deep learning-based automated detection and segmentation of brain metastases in malignant melanoma yields high detection and segmentation accuracy with false-positive findings of <1 per scan.

Evaluating anemia using contrast-enhanced spectral detector CT of the chest in a large cohort of 522 patients.

OBJECTIVES: The blood of patients with anemia demonstrates distinctly lower attenuation in unenhanced CT images. However, the frequent usage of intravenous contrast hampers evaluation of anemia. Spectral detector computed tomography (SDCT) allows for reconstruction of virtual non-contrast images (VNC) from contrast-enhanced data (CE). The purpose of this study was to evaluate whether VNC allow for prediction of anemia. METHODS: Five hundred twenty-two patients with CE-SDCT of the chest and accessible serum hemoglobin (HbS) were retrospectively included. Patients were assigned to three groups (severe anemia, moderate/mild anemia, and healthy) based on recent lab tests (≤ 7 days) for HbS following gender and the WHO definition of anemia. CT attenuation was determined using two ROI in the left ventricular lumen and one ROI in the descending thoracic aorta. ROI were placed on CE and copied to VNC. ANOVA, linear regression, and receiver operating characteristics were used for statistic evaluation. RESULTS: Average HbS was 11.6 ± 2.4 g/dl. Attenuation on VNC showed significant differences between healthy patients, patients with mild/moderate anemia, and severely anemic patients (all p ≤ 0.05). Applying cutoffs of 39.2/37.6 HU and 33.6/32.7 HU allowed to differentiate between healthy, mild/moderately, and severely anemic men/women (AUC 0.857/0.833 and 0.879/0.932). A linear relationship between HbS and attenuation on VNC was established (r2 = 0.54, HbS = - 0.875 + 0.329 × HU). CONCLUSIONS: An approximation of HbS and presence of anemia can be conducted based on simple attenuation measurements in contrast-enhanced SDCT examinations enabled by VNC imaging. KEY POINTS: • While the attenuation of blood is a previously described biomarker for anemia in non-contrast images, virtual non-contrast images from spectral detector CT circumvent this limitation and allow for diagnosis of anemia in contrast-enhanced scans. • Attenuation of blood in virtual non-contrast images derived from spectral detector CT shows a moderate correlation to serum hemoglobin levels. • Presence of anemia be estimated in virtual non-contrast images using proposed cutoffs of 39.2 HU and 37.6 HU for men and women, respectively, to differentiate between healthy and anemic patients.

Virtual monoenergetic images from spectral detector CT as a surrogate for conventional CT images: Unaltered attenuation characteristics with reduced image noise.

OBJECTIVE: This study aimed to identify the energy level of virtual monoenergetic images (VMI) that closest represents conventional images (CI) in order to demonstrate that these images provide improved image quality in terms of noise and Signal-to-noise ratio (SD/SNR) while attenuation values (HU) remain unaltered as compared to CI. METHODS: 60 and 30 patients with contrast-enhanced (CE) and non-enhanced (NCE) spectral detector CT (SDCT) of the abdomen were included in this retrospective, IRB-approved study. CI and VMI of 66-74 keV as well as quantitative iodine maps were reconstructed (Q-IodMap). Two regions of interest were placed in each: pulmonary trunk, abdominal aorta, portal vein, liver, pancreas, renal cortex left/right, psoas muscle, (filled) bladder and subcutaneous fat. For each reconstruction, HU and SD were averaged. ΔHU and SNR (SNR = HU/SD) were calculated. Q-IodMap were considered as confounder for ΔHU. In addition, two radiologists compared VMI of 72 keV and CI in a forced-choice approach regarding image quality. RESULTS: In NCE studies, no significant differences for any region was found. In CE studies, VMI72keV images showed lowest ΔHU (HUliver CI/VMI72keV: 104 ±â€¯18/103 ±â€¯17, p ≥ 0.05). Iodine containing voxels as indicated by Q-IodMap resulted in an over- and underestimation of attenuation in lower and higher VMI energies, respectively. Image noise was lower in VMI images (e.g. muscle: CI/ VMI72keV: 15.3 ±â€¯3.3/12.3 ±â€¯2.9 HU, p ≤ 0.05). Hence, SNR was significantly higher in VMI72keV compared to CI (e.g. liver 3.8 ±â€¯0.6 vs 3.0 ±â€¯0.8, p ≤ 0.05). In visual analysis, VMI72keV were preferred over CI at all times. CONCLUSIONS: VMI72keV show improved SD/SNR characteristics while the attenuation remains unaltered as compared to CI.

Improved visualization of hypodense liver lesions in virtual monoenergetic images from spectral detector CT: Proof of concept in a 3D-printed phantom and evaluation in 74 patients.

OBJECTIVES: The well-known boost of iodine associated-attenuation in low-keV virtual monoenergetic images (VMI_low) is frequently used to improve visualization of lesions and structures taking up contrast media. This study aimed to evaluate this concept in reverse. Hence to investigate if increased attenuation within the liver allows for improved visualization of little or not-enhancing lesions. METHODS: A 3D-printed phantom mimicking the shape of a human liver exhibiting a lesion in its center was designed and printed. Both, parenchyma- and lesion-mimic were filled with different solutions exhibiting 80/100/120HU and 0/15/40/60HU, respectively. Further, a total of 74 contrast-enhanced studies performed on a spectral detector CT scanner (SDCT) were included in this retrospective study. Patients had MRI or follow-up proven cysts and/or hypodense metastases. VMI of 40-200 keV as well as conventional images (CI) were reconstructed. ROI were placed in lesion and parenchyma(-mimics) on CI and transferred to VMI. Signal- and contrast-to-noise ratio were calculated (S-/CNR). Further, two radiologists independently evaluated image quality. Data was statistically assessed using ANOVA or Wilcoxon-test. RESULTS: In phantoms, S/CNR was significantly higher in VMI_low. The cyst-mimic in highly attenuating parenchyma-mimic on CI yielded a CNR of 6.4 ± 0.8; using VMI_40 keV, mildly hypodense lesion-mimic in poorly attenuating parenchyma-mimic exhibited a similar CNR (5.8 ± 0.9; p ≤ 0.05). The same tendency was observed in patients (cyst in CI/metastasis in VMI_40 keV: 4.4 ± 1.2/3.9 ± 1.8; p ≤ 0.05). Qualitative analysis indicated a benefit of VMI_40 keV (p ≤ 0.05). CONCLUSIONS: VMI_low from SDCT allow for an improved visualization of hypodense focal liver lesions exploiting the concept of contrast blooming in reverse.

Differentiation of Hemorrhage from Iodine Using Spectral Detector CT: A Phantom Study.

BACKGROUND AND PURPOSE: Conventional CT often cannot distinguish hemorrhage from iodine extravasation following reperfusion therapy for acute ischemic stroke. We investigated the potential of spectral detector CT in differentiating these lesions. MATERIALS AND METHODS: Centrifuged blood with increasing hematocrit (5%-85%) was used to model hemorrhage. Pure blood, blood-iodine mixtures (75/25, 50/50, and 25/75 ratios), and iodine solutions (0-14 mg I/mL) were scanned in a phantom with attenuation ranging from 12 to 75 HU on conventional imaging. Conventional and virtual noncontrast attenuation was compared and investigated for correlation with calculation of relative virtual noncontrast attenuation. Values for all investigated categories were compared using the Mann-Whitney U test. Sensitivity and specificity of virtual noncontrast, relative virtual noncontrast, conventional CT attenuation, and iodine quantification for hemorrhage detection were determined with receiver operating characteristic analysis. RESULTS: Conventional image attenuation was not significantly different among all samples containing blood (P > .05), while virtual noncontrast attenuation showed a significant decrease with a decreasing blood component (P < .01) in all blood-iodine mixtures. Relative virtual noncontrast values were significantly different among all investigated categories (P < .01), with correct hemorrhagic component size estimation for all categories within a 95% confidence interval. Areas under the curve for hemorrhage detection were 0.97, 0.87, 0.29, and 0.16 for virtual noncontrast, relative virtual noncontrast, conventional CT attenuation, and iodine quantification, respectively. A ≥10-HU virtual noncontrast, ≥20-HU virtual noncontrast, ≥40% relative virtual noncontrast, and combined ≥10-HU virtual noncontrast and ≥40% relative virtual noncontrast attenuation threshold had a sensitivity/specificity for detecting hemorrhage of 100%/23%, 89%/95%, 100%/82%, and 100%/100%, respectively. CONCLUSIONS: Spectral detector CT can accurately differentiate blood from iodinated contrast in a phantom setting.

Artifact reduction from dental implants using virtual monoenergetic reconstructions from novel spectral detector CT.

OBJECTIVES: Image quality in head and neck imaging is often severely hampered by artifacts arising from dental implants. This study evaluates metal artifact (MA) reduction using virtual monoenergetic images (VMI) compared to conventional CT images (CI) from spectral-detector computed tomography (SDCT). METHODS: 38 consecutive patients with dental implants were included in this retrospective study. All examinations were performed using a SDCT (IQon, Philips, Best, The Netherlands). Images were reconstructed as conventional images (CI) and as VMI in a range of 40-200 keV (10 keV increment). Quantitative image analysis was performed ROI-based by measurement of attenuation (HU) and standard deviation in most pronounced hypo- and hyperdense artifact, fat and soft tissue with presence of artifacts. Qualitatively, extent of artifact reduction, assessment of soft palate and cheeks were rated on 5-point Likert-scales by two radiologists. Statistical data evaluation included ANOVA and Wilcoxon-test with correction for multiple comparisons; interrater-agreement was determined by intraclass-correlation coefficient (ICC). RESULTS: The hypo- and hyperattenuating artifacts showed an increase and decrease of HU-values in VMIhigh (CI/VMI200 keV: -218.7/-174.4 HU, p = 0.1; and 309.8/119.2, p ≤ 0.05, respectively). Artifacts in the fat, as depicted by image noise did also decrease in VMIhigh (CI/VMI200 keV: 23.9/16.4, p ≤ 0.05). Qualitatively, hyperdense artifacts were decreased significantly in VMI ≥100 keV (e.g. CI/VMI200 keV: 2(1-3)/3(1-5), p ≤ 0.05). Artifact reduction resulted in improved assessment of the soft palate and cheeks (e.g. CI/VMI200 keV: 2(1-4)/3(1-5) and 2(1-5)/3(1-5), p ≤ 0.05). Overall interrater agreement was good (ICC = 0.77). CONCLUSIONS: Virtual monoenergetic images from SDCT reduce metal artifacts from dental implants and improve diagnostic assessment of surrounding soft tissue.

Reduction of artifacts caused by orthopedic hardware in the spine in spectral detector CT examinations using virtual monoenergetic image reconstructions and metal-artifact-reduction algorithms.

OBJECTIVE: Aim of this study was to assess the artifact reduction in patients with orthopedic hardware in the spine as provided by (1) metal-artifact-reduction algorithms (O-MAR) and (2) virtual monoenergetic images (MonoE) as provided by spectral detector CT (SDCT) compared to conventional iterative reconstruction (CI). METHODS: In all, 28 consecutive patients with orthopedic hardware in the spine who underwent SDCT-examinations were included. CI, O-MAR and MonoE (40-200 keV) images were reconstructed. Attenuation (HU) and noise (SD) were measured in order to calculate signal-to-noise ratio (SNR) of paravertebral muscle and spinal canal. Subjective image quality was assessed by two radiologists in terms of image quality and extent of artifact reduction. RESULTS: O-MAR and high-keV MonoE showed significant decrease of hypodense artifacts in terms of higher attenuation as compared to CI (CI vs O-MAR, 200 keV MonoE: -396.5HU vs. -115.2HU, -48.1HU; both p ≤ 0.001). Further, artifacts as depicted by noise were reduced in O-MAR and high-keV MonoE as compared to CI in (1) paravertebral muscle and (2) spinal canal-CI vs. O-MAR/200 keV: (1) 34.7 ± 19.0 HU vs. 26.4 ± 14.4 HU, p ≤ 0.05/27.4 ± 16.1, n.s.; (2) 103.4 ± 61.3 HU vs. 72.6 ± 62.6 HU/60.9 ± 40.1 HU, both p ≤ 0.001. Subjectively both O-MAR and high-keV images yielded an artifact reduction in up to 24/28 patients. CONCLUSION: Both, O-MAR and high-keV MonoE reconstructions as provided by SDCT lead to objective and subjective artifact reduction, thus the combination of O-MAR and MonoE seems promising for further reduction.