Contrast-enhanced Computed Tomography Data Research Articles

Liver tumour segmentation using contrast-enhanced multi-detector CT data: performance benchmarking of three semiautomated methods

Automatic tumour segmentation and volumetry is useful in cancer staging and treatment outcome assessment. This paper presents a performance benchmarking study on liver tumour segmentation for three semiautomatic algorithms: 2D region growing with knowledge-based constraints (A1), 2D voxel classification with propagational learning (A2) and Bayesian rule-based 3D region growing (A3). CT data from 30 patients were studied, and 47 liver tumours were isolated and manually segmented by experts to obtain the reference standard. Four datasets with ten tumours were used for algorithm training and the remaining 37 tumours for testing. Three evaluation metrics, relative absolute volume difference (RAVD), volumetric overlap error (VOE) and average symmetric surface distance (ASSD), were computed based on computerised and reference segmentations. A1, A2 and A3 obtained mean/median RAVD scores of 17.93/10.53%, 17.92/9.61% and 34.74/28.75%, mean/median VOEs of 30.47/26.79%, 25.70/22.64% and 39.95/38.54%, and mean/median ASSDs of 2.05/1.41 mm, 1.57/1.15 mm and 4.12/3.41 mm, respectively. For each metric, we obtained significantly lower values of A1 and A2 than A3 (P < 0.01), suggesting that A1 and A2 outperformed A3. Compared with the reference standard, the overall performance of A1 and A2 is promising. Further development and validation is necessary before reliable tumour segmentation and volumetry can be widely used clinically.

Tracer kinetics analysis of dynamic contrast‐enhanced CT and MR data in patients with squamous cell carcinoma of the upper aerodigestive tract: comparison of the results

(i) To evaluate the feasibility of tracer kinetics analysis of dynamic contrast-enhanced (DCE) CT and T2-weighted MR data of squamous cell carcinoma (SCCA) of the upper aerodigestive tract. (ii) To compare functional parameters derived by both modalities and examine the interchangeability of them as well as the intra- and inter-rater agreement. Dynamic contrast-enhanced-CT and MR images of 23 patients with SCCA were postprocessed using a distributed-parameter (DP) tracer kinetic model. The evaluated parameters included blood flow (F), intravascular blood volume (v(1)), extravascular extracellular blood volume (v(2)), intravascular mean transit time (t(1)), lag time (t(0)), permeability surface area product (PS) and extraction ratio (E). Mean perfusion values, based on region-of-interest analysis, of the tumors and the healthy muscle tissue were compared and correlated. Inter-rater and intra-rater variability were assessed. Interchangeability of the tumor functional parameters was tested using Pearson's correlation coeficients and Bland-Altman plots. The mean values in tumor and healthy muscle tissues were significantly different for each modality (0.0001< or =P< or =0.03). The mean values of all tumor perfusion parameters apart from v(2) and E were significantly different (0.001< or =P< or =0.009) between the two modalities. The intra-rater variability was good to very good for all parameters. The inter-rater variability was moderate to good. Bland-Altman plots of F, t(1), t(0), and v(2) showed moderate interchangeability. There was a proportionality error in v(1) and PS graphs. The estimation of functional parameters in SCCA is feasible using DCE-CT and -MR with a DP model. The parameters are mostly significantly different and the interchangeability of them is limited.

Effect of Contrast Media on Megavoltage Photon Beam Dosimetry

The purpose of this study was to quantify changes in photon beam dosimetry caused by using contrast media during computed tomography (CT) simulation and determine if the resulting changes are clinically significant. The effect of contrast on dosimetry was first examined for a single 6-MV photon beam incident on a plane phantom with a structure of varying electron densities (ρ e) and thickness. Patient studies were then undertaken in which CT data sets were collected with and without contrast for 6 typical patients. Three patients received IV contrast (Optiray-240™) only and 3 received IV plus oral (Gastrograffin™) contrast. Each patient was planned using conformal multifield techniques in accordance with the department standards. Two methods were used to compare the effect of contrast on dosimetry for each patient. The phantom analysis showed that the change in dose at the isocenter for a single 10 × 10 cm 2 6-MV photon beam traversing 10 cm of a contrast-enhanced structure with ρ e 1.22 was 7.0% (1.22 was the highest average ρ e observed in the patient data). As a result of using contrast, increases in ρ e were observed in structures for the 6 patients studied. Consequently, when using contrast-enhanced CT data for multifield planning, increases in dose at the isocenter and in critical structures were observed up to 2.1% and 2.5%, respectively. Planning on contrast-enhanced CT images may result in an increase in dose of up to 2.1% at the isocenter, which would generally be regarded as clinically insignificant. If, however, a critical organ is in close proximity to the planning target volume (PTV) and is planned to receive its maximum allowable dose, planning on contrast-enhanced CT images may result in that organ receiving dose beyond the recommended tolerance. In these instances, pre-contrast CT data should be used for dosimetry.

Liver Tumor segmentation in CT images using probabilistic

Liver tumors segmentation is an important prerequisite for planning of surgical interventions. For clinical applicability, the segmentation approach must be able to cope with the high variation in shape and gray-value appearance of the liver. We present a fully automatic 3D segmentation method for the liver tumors from contrast-enhanced CT data. The method consists of two main stages. First an initial histogram and statistical distribution functions are created, and from them a new image is created where, in each voxel, a weighted function is attached in accordance with the probability of the voxel grey level. Next, we use the active contour method on the new image, where the active contour evolution is based upon the minimization of variances between the liver tumor and its closest neighborhood.

Assessment of regional left ventricular function by Dual Source Computed Tomography: Interobserver variability and validation to laevocardiography

Objective Assessment of left ventricular function is possible in contrast-enhanced cardiac CT data sets. However, rapid ventricular motion especially in systole can lead to artifacts. Dual Source Computed Tomography (DSCT) has high temporal resolution which effectively limits motion artifact. We therefore assessed the accuracy of DSCT to detect regional left ventricular wall motion abnormalities in comparison to invasive cine angiocardiography. Methods We analyzed DSCT data sets of 50 patients (39 male, 11 female, mean age: 61 ± 10 years) which were acquired after intravenous injection of 55–70 mL contrast agent (rotation time: 330 ms, collimation: 2 mm × 64 mm × 0.6 mm, 120 kV, 380 mAs, ECG-correlated tube current modulation). 10 data sets consisting of transaxial slices with a slice thickness of 1.5 mm, an increment of 1.0 mm and a matrix of 256 × 256 pixels were reconstructed at 10 time instants during the cardiac cycle (0–90% in 10% increments). The data sets were analyzed visually by two independent readers, using standard left ventricular planes, concerning regional wall motion abnormalities. DSCT was verified in a blinded fashion against cine ventriculography performed during cardiac catheterization (RAO and LAO projection), using a 7-segment model. Analysis was performed on a per-patient (presence of at least one hypo-, a- or dyskinetic segment) and on a per-segment basis. Results Concerning the presence of a wall motion abnormality, the two observers agreed in 340/350 segments (97%) and 48/50 patients (96%). In invasive cine angiocardiography, 22 of 50 patients displayed at least one segment with abnormal contraction. To detect these patients, DSCT showed a sensitivity of 95% (21/22), specificity of 96% (27/28), positive predictive value of 95% and negative predictive value of 96%. Out of a total of 350 left ventricular segments, 66 segments had abnormal contraction in cine angiocardiography (34 hypokinetic, 26 akinetic, 6 dyskinetic). For detection of these segments, DSCT had a sensitivity of 88% (58/66), specificity of 98% (278/284), positive predictive value of 91% (58/64) and negative predictive value of 97% (278/286). Conclusion DSCT allows the detection of regional wall motion abnormalities with high interobserver agreement as well as high sensitivity and specificity. Whereas sensitivity and positive predictive value were higher in a per-patient- in comparison to a per-segment-based analysis, specificity, negative predictive value and interobserver agreement did not differ considerably between both analyzing methods.

Impact of whole-body imaging on treatment decision to radio-frequency ablation in patients with malignant liver tumors: comparison of [18F]fluorodeoxyglucose-PET/computed tomography, PET and computed tomography

The correct staging of patients with malignant liver tumors before radio-frequency ablation (RFA) is mandatory for successful treatment. Our study aimed to compare the influence on decision to perform RFA of whole-body fluorodeoxyglucose (FDG)-PET/computed tomography (CT) with whole-body contrast-enhanced CT (CE-CT) and PET alone. Fifty-eight patients with known hepatic malignancies (23, liver metastases 35) received FDG-PET/CT before RFA planned with curative intention. CT and PET data were each read separately, PET/CT fusion data were read in consensus afterward by a third reader group. The diagnostic accuracy of CE-CT, PET alone, and PET/CT to identify patients eligible for RFA was compared and the impact on decision was analyzed. The McNemar test with Bonferroni correction was used to test for significant differences. The accuracy and sensitivity to detect correctly intrahepatic and extrahepatic tumor were 94 and 97% for CT, 75 and 54% for PET, and 97 and 95% for PET/CT. The differences between CT and PET, as well as between PET/CT and PET, were statistically significant, but there was no significant difference between PET/CT and CT alone (P>0.65). PET alone, CE-CT, and PET/CT correctly identified 32, 55, and 57 patients, respectively. Again, PET/CT showed no significant advantage over CE-CT. Both imaging methods performed significantly better than PET alone (P<0.0001). Forty-three (74%) of 58 patients underwent RFA with curative intention. Whole-body imaging changed patient management in 26% of the patients planned for curative intended RFA, yet there was no significant difference between CE-CT and PET/CT.

Dynamic contrast-enhanced CT of head and neck tumors: perfusion measurements using a distributed-parameter tracer kinetic model. Initial results and comparison with deconvolution-based analysis

The objective of this work was to evaluate the feasibility of a two-compartment distributed-parameter (DP) tracer kinetic model to generate functional images of several physiologic parameters from dynamic contrast-enhanced CT data obtained of patients with extracranial head and neck tumors and to compare the DP functional images to those obtained by deconvolution-based DCE-CT data analysis. We performed post-processing of DCE-CT studies, obtained from 15 patients with benign and malignant head and neck cancer. We introduced a DP model of the impulse residue function for a capillary-tissue exchange unit, which accounts for the processes of convective transport and capillary-tissue exchange. The calculated parametric maps represented blood flow (F), intravascular blood volume (v1), extravascular extracellular blood volume (v2), vascular transit time (t1), permeability–surface area product (PS), transfer ratios k12 and k21, and the fraction of extracted tracer (E). Based on the same regions of interest (ROI) analysis, we calculated the tumor blood flow (BF), blood volume (BV) and mean transit time (MTT) by using a modified deconvolution-based analysis taking into account the extravasation of the contrast agent for PS imaging. We compared the corresponding values by using Bland–Altman plot analysis. We outlined 73 ROIs including tumor sites, lymph nodes and normal tissue. The Bland–Altman plot analysis revealed that the two methods showed an accepted degree of agreement for blood flow, and, thus, can be used interchangeably for measuring this parameter. Slightly worse agreement was observed between v1 in the DP model and BV but even here the two tracer kinetic analyses can be used interchangeably. Under consideration of whether both techniques may be used interchangeably was the case of t1 and MTT, as well as for measurements of the PS values. The application of the proposed DP model is feasible in the clinical routine and it can be used interchangeably for measuring blood flow and vascular volume with the commercially available reference standard of the deconvolution-based approach. The lack of substantial agreement between the measurements of vascular transit time and permeability–surface area product may be attributed to the different tracer kinetic principles employed by both models and the detailed capillary tissue exchange physiological modeling of the DP technique.

Isolation of the left atrial surface from cardiac multi-detector CT images based on marker controlled watershed segmentation

The delineation of left atrium (LA) and pulmonary veins (PVs) anatomy from high resolution images holds importance for atrial fibrillation (AF) investigation and treatment. In this study, a semiautomatic segmentation procedure for LA and PVs inner surface from contrast enhanced CT data was developed. The procedure consists of a three dimensional marker controlled watershed segmentation applied to the external morphological gradient, followed by variable threshold surface extraction from the original intensity image. A preliminary anisotropic non-linear filtering was implemented to improve the S/N ratio of CT images. The performance of segmentation was evaluated on cardiac CT scans of 12 AF patients both qualitatively and quantitatively. The qualitative evaluation by expert radiologist assessed the segmentation as overall successful in all patients and capable of extracting both the LA body and the connected vascular trees. The quantitative validation, by computing discrepancy measures with respect to a manually segmented gold standard, indicated an average of about 90% of voxels correctly classified and an average border mismatch lower than 1.5 voxels (1.2 mm). The accurate extraction of the inner LA-PVs walls provided by this method, along with the minimal required human intervention, should facilitate the use of anatomical atrial models for the non-pharmacological treatment of AF.

Determination of 3-dimensional Zonal Renal Volumes Using Contrast-enhanced Computed Tomography

To determine the accuracy with which total and cortical renal volumes may be measured in vivo with contrast-enhanced computed tomography (CT) in an animal model. Seven female Yorkshire pigs were scanned in vivo using both a 64-channel multidetector row computed tomography (MDCT) scanner (Siemens Medical Solutions, Erlangen, Germany) and a C-arm CT scanner (Siemens AXIOM). Kidneys were scanned in corticomedullary and nephrographic phases after contrast injection. Three-dimensional volumetric analysis of CT data was performed to determine renal cortical and overall renal volumes, and renal lengths. We measured renal weights, water displacement volumes, and overall and cortical renal volumes of resected kidneys. Overall and cortical renal volumes by CT showed excellent correlation (R2 values varying from 0.77 to 0.92) with anatomic measurements of renal volume. There was poor correlation of anatomic renal volumes with renal lengths by CT. Three-dimensional analysis of contrast-enhanced CT provides accurate measurement of overall and cortical renal volumes in vivo.

Clinical evaluation of bone-subtraction CT angiography (BSCTA) in head and neck imaging

Fifty-one patients were examined with bone subtraction CT angiography (BSCTA). Data were acquired on 4- and 64-slice spiral CT systems. The postprocessing method is based on fully automatic registration of nonenhanced and contrast-enhanced CT data and subsequent selective bone removal. Vascular structures and brain tissue are retained with the original CTA noise level. Image quality and delineation of the pathologic process were assessed, and artifacts were introduced by the bone removal process recorded.

Localizing Calcifications in Cardiac CT Data Sets Using a New Vessel Segmentation Approach

The new generation of multislice computed tomography (CT) scanners allows for the acquisition of high-resolution images of the heart. Based on that image data, the heart can be analyzed in a noninvasive way-improving the diagnosis of cardiovascular malfunctions on one hand, and the planning of an eventually necessary intervention on the other. One important parameter for the evaluation of the severity of a coronary artery disease is the number and localization of calcifications (hard plaques). This work presents a method for localizing these calcifications by employing a newly developed vessel segmentation approach. This extraction technique has been developed for, and tested with, contrast-enhanced CT data sets of the heart. The algorithm provides enough information to compute the vessel diameter along the extracted segment. An approach for automatically detecting calcified regions that combines diameter information and gray value analysis is presented. In addition, specially adapted methods for the visualization of these analysis results are described.

Optimizing localization accuracy in head and neck, and brain radiotherapy

The purpose of this study was to investigate the impact on localization of utilizing contrast-enhanced CT scans and the formal input of a radiologist in the planning process. 25 head and neck/brain patients had pre- and post-contrast CT scans in the treatment position. Radiotherapy treatment was planned on the unenhanced CT images as per standard practice. Retrospectively, their scans (unenhanced and enhanced) were re-contoured by two oncologists and a radiologist. These new contours were compared with the original unenhanced treatment contours and differences in contour volume, geographical isocentre position and tolerance coverage of the associated planning target volumes (PTVs) were evaluated using the original plans. The use of contrast enhanced CT data during localization by the oncologist shows little change in gross tumour volumes (GTVs) or PTVs, geographical position or tolerance coverage for the targets in the brain studied here. Larger changes in mean volume are seen for the head and neck cases alone. Changes are greater and statistically significant (p < 0.05, Wilcoxon signed rank test) for localization by the radiologist. Furthermore, when comparing the original PTV marked by the oncologist with a new PTV re-contoured by the oncologist, but based on a GTV marked-up by the radiologist, again statistically significant (p < 0.01) changes in percentage volume are noted. Intraoperator precision is good, percentage volume differences being of the order 3-6%. PTVs also show improved standard deviations compared with GTVs. Geographic shifts are generally within our departmental tolerance levels for daily patient setup. Comparing precision of unenhanced data with enhanced, mean percentage volume changes are smaller, but not statistically significant. The use of enhanced scan data for localization has little effect on size, geographical position or tolerance coverage of PTVs marked up by the oncologists in this study. However, more important is the input from a radiologist. Statistically significant differences due to mark-up on enhanced scans by the radiologist are shown. Furthermore, significant differences are also seen between PTVs based on oncologist-generated GTVs, and those based on radiologist-generated GTVs.

Clinical evaluation of bone-subtraction CT angiography (BSCTA) in head and neck imaging

Fifty-one patients were examined with bone subtraction CT angiography (BSCTA). Data were acquired on 4-and 64-slice spiral CT systems. The post-processing method is based on fully automatic registration of non-enhanced and contrast-enhanced CT data and subsequent selective bone removal. Vascular structures and brain tissue are retained with the original CTA noise level. Image quality and delineation of the pathologic process were assessed and artifacts introduced by the bone removal process recorded. The bone subtraction algorithm worked successfully in all examinations. The processing time was 6 min on average. Image quality was rated excellent in 20 (39%), good in 26 (51%) and acceptable in 5 (10%) patients. Ophthalmic arteries were visible in 12 (24%) patients bilaterally, in 13 (25%) patients unilaterally and in 26 (51%) patients at least at the origin. BSCTA improved visualization of the infraclinoid ICA and the vertebral arteries. The depiction of stenosis of the extracranial ICA and supraclinoid aneurysms was not significantly improved. In patients with suspicion of sinus thrombosis, BSCTA and conventional CTA yielded similar results. To conclude, BSCTA improves the visualization of vessels with close contact to bone and can improve the diagnostic accuracy and therapy planning of infraclinoid aneurysms.

Analysis of dynamic contrast enhanced MRI

The advances in MRI image acquisition have led to the increasing use of parametric or calculated images, which are designed to display physiological features of tissues rather than simply anatomical structure. One common type of dynamic imaging data is produced when rapidly repeated sequential imaging sequences are performed during injection of an intravenous contrast media [1–8]. Although many of the applications of dynamic contrast enhanced imaging were developed using MRI in the brain they are now being commonly applied using both MRI and CT. These techniques now have both research and clinical applications throughout the body and in many types of pathological tissue. In this article we will describe the analysis techniques which can be applied to dynamic contrast enhanced data. We will start with the specific application of dynamic susceptibility contrast imaging of perfusion using MRI (DSC-MRI) since a number of technical aspects of the data collection and analysis are unique to this approach. We will then discuss the techniques available for the analysis of dynamic relaxivity contrast enhanced MRI (DRC-MRI), which are also directly applicable to dynamic contrast enhanced CT data.

Daily ultrasound-based image-guided targeting for radiotherapy of upper abdominal malignancies

Purpose Development and implementation of a strategy to use a stereotactic ultrasound (US)-based image-guided targeting device (BAT) to align intensity-modulated radiotherapy (IMRT) target volumes accurately in the upper abdomen. Because the outlines of such targets may be poorly visualized by US, we present a method that uses adjacent vascular guidance structures as surrogates for the target position. We assessed the potential for improvement of daily repositioning and the feasibility of daily application. Methods and materials A total of 62 patients were treated by sequential tomotherapeutic IMRT between October 2000 and June 2003 for cholangiocarcinoma and gallbladder carcinoma ( n = 10), hepatocellular carcinoma ( n = 10), liver metastases ( n = 11), pancreatic carcinoma ( n = 20), neuroblastoma ( n = 3), and other abdominal and retroperitoneal tumors ( n = 8). The target volumes (TVs) and organs at risk were delineated in contrast-enhanced CT data sets. Additionally, vascular guidance structures in close anatomic relation to the TV, or within the TV, were delineated. Throughout the course of IMRT, US BAT images were acquired during daily treatment positioning. In addition to the anatomic structures typically used for US targeting (e.g., the TV and dose-limiting organs at risk), CT contours of guidance structures were superimposed onto the real-time acquired axial and sagittal US images, and target position adjustments, as indicated by the system, were performed accordingly. We report the BAT-derived distribution of shifts in the three principal room axes compared with a skin-mark–based setup, as well as the time required to perform BAT alignment. The capability of the presented method to improve target alignment was assessed in 15 patients by comparing the organ and fiducial position between the respective treatment simulation CT with a control CT study after US targeting in the CT suite. Results A total of 1,337 BAT alignments were attempted. US images were not useful in 56 setups (4.2%), mainly because of limited visibility due to daily variations in colonic and gastric air. US imaging was facilitated in intrahepatic tumors and asthenic patients. The mean ± SD shift from the skin mark position was 4.9 ± 4.35, 6.0 ± 5.31, and 6.0 ± 6.7 mm in the x, y, and z direction, respectively. The mean magnitude vector of three-dimensional alignment correction was 11.4 ± 7.6 mm. The proportion of daily alignments corrected by a magnitude of >10, >15, and >20 mm was 48.9%, 25.1%, and 12.7%, respectively. The magnitude of shifts in the principal directions, as well as the three-dimensional vector of displacement, was statistically significant (test against the zero hypothesis) at p <0.0001. The guidance structures that were the most valuable for identification of the TV position were the branches of the portal vein, hepatic artery, and dilated bile ducts in intrahepatic lesions and the aorta, celiac trunk, superior mesenteric artery, and extrahepatic aspects of the portal vein system in retroperitoneal and extrahepatic lesions. The mean total setup time was 4.6 min. The correlation of BAT targeting with target setup error assessment in the control CT scans in 15 patients revealed setup error reduction in 14 of 15 alignments. The average setup error reduction, assessed as a reduction in the length of setup error three-dimensional magnitude vector, was 54.4% ± 26.9%, with an observed mean magnitude of residual setup error of 4.6 ± 3.4 mm. The sole worsening of an initial setup was by a magnitude of <2 mm. US targeting resulted in statistically significant improvements in patient setup ( p = 0.03). Conclusion Daily US-guided BAT targeting for patients with upper abdominal tumors was feasible in the vast majority of attempted setups. This method of US-based image-guided tumor targeting has been successfully implemented in clinical routine. The observed improved daily repositioning accuracy might allow for individualized reduction of safety margins and optional dose escalation. Compared with the established application of the BAT device for prostate radiotherapy, in which the target can be directly visualized, the TV in the present study was predominantly positioned relative to guidance vascular structures in close anatomic relation. We perceived an enormous potential in improved and individualized patient positioning for fractionated radiotherapy and also for stereotactic extracranial radiotherapy and radiosurgery, especially for tumors of the liver and pancreas.