Computed tomography-based calcium scoring in cadaver leg arteries: Influence of dose, reader, and reconstruction algorithm

PURPOSE

Computed tomography (CT) might be a good diagnostic test to accurately quantify calcium in vascular beds but there are multiple factors influencing the quantification. The aim of this study was to investigate the influence of different computed tomography protocol settings in the quantification of calcium in the lower extremities using modified Agatston and volume scores.

METHODS

Fresh-frozen human legs were scanned at different tube current protocols and reconstructed at different slice thickness. Two different iterative reconstruction protocols for conventional CT images were compared. Calcium was manually scored using modified Agatston and volume scores. Outcomes were statistically analyzed using Wilcoxon signed-rank tests and mean absolute and relative differences were plotted in Bland-Altman plots.

RESULTS

Of the 20 legs, 16 had CT detectable calcifications. Differences between thick and thin slice reconstruction protocols were 129 Agatston units and 125% for Agatston and 78.4 mm³ and 57.8% for volume (all p ≤ 0.001). No significant differences were found between low and high tube current protocols. Differences between iDose⁴ and IMR reconstruction protocols for modified Agatston were 34.2 Agatston units and 17.7% and the volume score 33.5 mm³ and 21.2% (all p ≤ 0.001).

CONCLUSIONS

Slice thickness reconstruction and reconstruction method protocols influenced the modified Agatston and volume scores in leg arteries, but tube current and different observers did not have an effect. This data emphasizes the need for standardized quantification of leg artery calcifications. Possible implications are in the development of a more universal quantification method, independent of the type of scan and vasculature.

Emphysema quantification using chest CT: influence of radiation dose reduction and reconstruction technique

BACKGROUND

Computed tomography (CT) emphysema quantification is affected by both radiation dose (i.e. image noise) and reconstruction technique. At reduced dose, filtered back projection (FBP) results in an overestimation of the amount of emphysema due to higher noise levels, while the use of iterative reconstruction (IR) can result in an underestimation due to reduced noise. The objective of this study was to determine the influence of dose reduction and hybrid IR (HIR) or model-based IR (MIR) on CT emphysema quantification.

METHODS

Twenty-two patients underwent inspiratory chest CT scan at routine radiation dose and at 45%, 60% and 75% reduced radiation dose. Acquisitions were reconstructed with FBP, HIR and MIR. Emphysema was quantified using the 15th percentile of the attenuation curve and the percentage of voxels below -950 HU. To determine whether the use of a different percentile or HU threshold is more accurate at reduced dose levels and with IR, additional measurements were performed using different percentiles and HU thresholds to determine the optimal combination.

RESULTS

Dose reduction resulted in a significant overestimation of emphysema, while HIR and MIR resulted in an underestimation. Lower HU thresholds with FBP at reduced dose and higher HU thresholds with HIR and MIR resulted in emphysema percentages comparable to the reference. The 15th percentile quantification method showed similar results as the HU threshold method.

CONCLUSIONS

This within-patients study showed that CT emphysema quantification is significantly affected by dose reduction and IR. This can potentially be solved by adapting commonly used thresholds.

Image quality of conventional images of dual-layer SPECTRAL CT: A phantom study

PURPOSE

Spectral CT using a dual layer detector offers the possibility of retrospectively introducing spectral information to conventional CT images. In theory, the dual-layer technology should not come with a dose or image quality penalty for conventional images. In this study, we evaluate the influence of a dual-layer detector (IQon Spectral CT, Philips Healthcare) on the image quality of conventional CT images, by comparing these images with those of a conventional but otherwise technically comparable single-layer CT scanner (Brilliance iCT, Philips Healthcare), by means of phantom experiments.

METHODS

For both CT scanners, conventional CT images were acquired using four adult scanning protocols: (a) body helical, (b) body axial, (c) head helical, and (d) head axial. A CATPHAN 600 phantom was scanned to conduct an assessment of image quality metrics at equivalent (CTDI) dose levels. Noise was characterized by means of noise power spectra (NPS) and standard deviation (SD) of a uniform region, and spatial resolution was evaluated with modulation transfer functions (MTF) of a tungsten wire. In addition, contrast-to-noise ratio (CNR), image uniformity, CT number linearity, slice thickness, slice spacing, and spatial linearity were measured and evaluated. Additional measurements of CNR, resolution and noise were performed in two larger phantoms.

RESULTS

The resolution levels at 50%, 10%, and 5% MTF of the iCT and IQon showed small, but significant differences up to 0.25 lp/cm for body scans, and up to 0.2 lp/cm for head scans in favor of the IQon. The iCT and IQon showed perfect CT linearity for body scans, but for head scans both scanners showed an underestimation of the CT numbers of materials with a high opacity. Slice thickness was slightly overestimated for both scanners. Slice spacing was comparable and reconstructed correctly. In addition, spatial linearity was excellent for both scanners, with a maximum error of 0.11 mm. CNR was higher on the IQon compared to the iCT for both normal and larger phantoms with differences up to 0.51. Spatial resolution did not change with phantom size, but noise levels increased significantly. For head scans, IQon had a noise level that was significantly lower than the iCT, on the other hand IQon showed noise levels significantly higher than the iCT for body scans. Still, these differences were well within the specified range of performance of iCT scanners.

CONCLUSIONS

At equivalent dose levels, this study showed similar quality of conventional images acquired on iCT and IQon for medium-sized phantoms and slightly degraded image quality for (very) large phantoms at lower tube voltages on the IQon. Accordingly, it may be concluded that the introduction of a dual-layer detector neither compromises image quality of conventional images nor increases radiation dose for normal-sized patients, and slightly degrades dose efficiency for large patients at 120 kVp and lower tube voltages.

Contrast agent concentration optimization in CTA using low tube voltage and dual-energy CT in multiple vendors: a phantom study

We investigated the feasibility and extent to which iodine concentration can be reduced in computed tomography angiography imaging of the aorta and coronary arteries using low tube voltage and virtual monochromatic imaging of 3 major dual-energy CT (DECT) vendors. A circulation phantom was imaged with dual source CT (DSCT), gemstone spectral imaging (GSI) and dual-layer spectral detector CT (SDCT). For each scanner, a reference scan was acquired at 120 kVp using routine iodine concentration (300 mg I/ml). Subsequently, scans were acquired at lowest possible tube potential (70, 80, 80 kVp, respectively), and DECT-mode (80/150Sn, 80/140 and 120 kVp, respectively) in arterial phase after administration of iodine (300, 240, 180, 120, 60, 30 mg I/ml). Objective image quality was evaluated using attenuation, CNR and dose corrected CNR (DCCNR) measured in the aorta and left main coronary artery. Average DCCNR at reference was 227.0, 39.7 and 60.2 for DSCT, GSI and SDCT. Maximum iodine concentration reduction without loss of DCCNR was feasible down to 180 mg I/ml (40% reduced) for DSCT (DCCNR 467.1) and GSI (DCCNR 46.1) using conventional CT low kVp, and 120 mg I/ml (60% reduced) for SDCT (DCCNR 171.5) using DECT mode. Low kVp scanning and DECT allows for 40-60% iodine reduction without loss in image quality compared to reference. Optimal scan protocol and to which extent varies per vendor. Further patient studies are needed to extend and translate our findings to clinical practice.

The Effects of Iodine Attenuation on Pulmonary Nodule Volumetry using Novel Dual-Layer Computed Tomography Reconstructions

OBJECTIVES

To assess the effect of iodine attenuation on pulmonary nodule volumetry using virtual non-contrast (VNC) and mono-energetic reconstructions.

METHODS

A consecutive series of patients who underwent a contrast-enhanced chest CT scan were included. Images were acquired on a novel dual-layer spectral CT system. Conventional reconstructions as well as VNC and mono-energetic images at different keV levels were used for nodule volumetry.

RESULTS

Twenty-four patients with a total of 63 nodules were included. Conventional reconstructions showed a median (interquartile range) volume and diameter of 174 (87 - 253) mm3 and 6.9 (5.4 - 9.9) mm, respectively. VNC reconstructions resulted in a significant volume reduction of 5.5% (2.6 - 11.2%; p<0.001). Mono-energetic reconstructions showed a correlation between nodule attenuation and nodule volume (Spearman correlation 0.77, (0.49 - 0.94)). Lowering the keV resulted in increased volumes while higher keV levels resulted in decreased pulmonary nodule volumes compared to conventional CT.

CONCLUSIONS

Novel dual-layer spectral CT offers the possibility to reconstruct VNC and mono-energetic images. Those reconstructions show that higher pulmonary nodule attenuation results in larger nodule volumes. This may explain the reported underestimation in nodule volume on non-contrast enhanced compared to contrast-enhanced acquisitions.

KEY POINTS

• Pulmonary nodule volumes were measured on virtual non-contrast and mono-energetic reconstructions • Mono-energetic reconstructions showed that higher attenuation results in larger volumes • This may explain the reported nodule volume underestimation on non-contrast enhanced CT • Mostly metastatic pulmonary nodules were evaluated, results might differ for benign nodules.

Radiation dose reduction for CT assessment of urolithiasis using iterative reconstruction: A prospective intra-individual study

Objective

To assess the performance of hybrid (HIR) and model-based iterative reconstruction (MIR) in patients with urolithiasis at reduced-dose computed tomography (CT).

Methods

Twenty patients scheduled for unenhanced abdominal CT for follow-up of urolithiasis were prospectively included. Routine dose acquisition was followed by three low-dose acquisitions at 40%, 60% and 80% reduced doses. All images were reconstructed with filtered back projection (FBP), HIR and MIR. Urolithiasis detection rates, gall bladder, appendix and rectosigmoid evaluation and overall subjective image quality were evaluated by two observers.

Results

74 stones were present in 17 patients. Half the stones were not detected on FBP at the lowest dose level, but this improved with MIR to a sensitivity of 100%. HIR resulted in a slight decrease in sensitivity at the lowest dose to 72%, but outperformed FBP. Evaluation of other structures with HIR at 40% and with MIR at 60% dose reductions was comparable to FBP at routine dose, but 80% dose reduction resulted in non-evaluable images.

Conclusions

CT radiation dose for urolithiasis detection can be safely reduced by 40 (HIR)–60 (MIR) % without affecting assessment of urolithiasis, possible extra-urinary tract pathology or overall image quality.

Key Points

• Iterative reconstruction can be used to substantially lower the radiation dose.

• This allows for radiation reduction without affecting sensitivity of stone detection.

• Possible extra-urinary tract pathology evaluation is feasible at 40–60% reduced dose.

Electronic supplementary material

The online version of this article (doi:10.1007/s00330-017-4929-2) contains supplementary material, which is available to authorized users.

Radiation dose reduction in pediatric great vessel stent computed tomography using iterative reconstruction: A phantom study

BACKGROUND

To study dose reduction using iterative reconstruction (IR) for pediatric great vessel stent computed tomography (CT).

METHODS

Five different great vessel stents were separately placed in a gel-containing plastic holder within an anthropomorphic chest phantom. The stent lumen was filled with diluted contrast gel. CT acquisitions were performed at routine dose, 52% and 81% reduced dose and reconstructed with filtered back projection (FBP) and IR. Objective image quality in terms of noise, signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) as well as subjective image quality were evaluated.

RESULTS

Noise, SNR and CNR were improved with IR at routine and 52% reduced dose, compared to FBP at routine dose. The lowest dose level resulted in decreased objective image quality with both FBP and IR. Subjective image quality was excellent at all dose levels.

CONCLUSION

IR resulted in improved objective image quality at routine dose and 52% reduced dose, while objective image quality deteriorated at 81% reduced dose. Subjective image quality was not affected by dose reduction.

Accuracy of bone mineral density quantification using dual-layer spectral detector CT: a phantom study

OBJECTIVES

To investigate the accuracy of bone mineral density (BMD) quantification using dual-layer spectral detector CT (SDCT) at various scan protocols.

METHODS

Two validated anthropomorphic phantoms containing inserts of 50-200 mg/cm³ calcium hydroxyapatite (HA) were scanned using a 64-slice SDCT scanner at various acquisition protocols (120 and 140 kVp, and 50, 100 and 200 mAs). Regions of interest (ROIs) were placed in each insert and mean attenuation profiles at monochromatic energy levels (90-200 keV) were constructed. These profiles were fitted to attenuation profiles of pure HA and water to calculate HA concentrations. For comparison, one phantom was scanned using dual energy X-ray absorptiometry (DXA).

RESULTS

At both 120 and 140 kVp, excellent correlations (R = 0.97, P < 0.001) were found between true and measured HA concentrations. Mean error for all measurements at 120 kVp was -5.6 ± 5.7 mg/cm³ (-3.6 ± 3.2%) and at 140 kVp -2.4 ± 3.7 mg/cm³ (-0.8 ± 2.8%). Mean measurement errors were smaller than 6% for all acquisition protocols. Strong linear correlations (R² ≥ 0.970, P < 0.001) with DXA were found.

CONCLUSIONS

SDCT allows for accurate BMD quantification and potentially opens up the possibility for osteoporosis evaluation and opportunistic screening in patients undergoing SDCT for other clinical indications. However, patient studies are needed to extend and translate our findings.

KEY POINTS

• Dual-layer spectral detector CT allows for accurate bone mineral density quantification. • BMD measurements on SDCT are strongly linearly correlated to DXA. • SDCT, acquired for several indications, may allow for evaluation of osteoporosis. • This potentially opens up the possibility for opportunistic osteoporosis screening.

Feasibility and accuracy of dual-layer spectral detector computed tomography for quantification of gadolinium: a phantom study

Objectives

The aim of this study was to evaluate the feasibility and accuracy of dual-layer spectral detector CT (SDCT) for the quantification of clinically encountered gadolinium concentrations.

Methods

The cardiac chamber of an anthropomorphic thoracic phantom was equipped with 14 tubular inserts containing different gadolinium concentrations, ranging from 0 to 26.3 mg/mL (0.0, 0.1, 0.2, 0.4, 0.5, 1.0, 2.0, 3.0, 4.0, 5.1, 10.6, 15.7, 20.7 and 26.3 mg/mL). Images were acquired using a novel 64-detector row SDCT system at 120 and 140 kVp. Acquisitions were repeated five times to assess reproducibility. Regions of interest (ROIs) were drawn on three slices per insert. A spectral plot was extracted for every ROI and mean attenuation profiles were fitted to known attenuation profiles of water and pure gadolinium using in-house-developed software to calculate gadolinium concentrations.

Results

At both 120 and 140 kVp, excellent correlations between scan repetitions and true and measured gadolinium concentrations were found (R > 0.99, P < 0.001; ICCs > 0.99, CI 0.99–1.00). Relative mean measurement errors stayed below 10% down to 2.0 mg/mL true gadolinium concentration at 120 kVp and below 5% down to 1.0 mg/mL true gadolinium concentration at 140 kVp.

Conclusion

SDCT allows for accurate quantification of gadolinium at both 120 and 140 kVp. Lowest measurement errors were found for 140 kVp acquisitions.

Key Points

• Gadolinium quantification may be useful in patients with contraindication to iodine.

• Dual-layer spectral detector CT allows for overall accurate quantification of gadolinium.

• Interscan variability of gadolinium quantification using SDCT material decomposition is excellent.

New horizons in cardiac CT

Until recently, cardiovascular computed tomography angiography (CCTA) was associated with considerable radiation doses. The introduction of tube current modulation and automatic tube potential selection as well as high-pitch prospective ECG-triggering and iterative reconstruction offer the ability to decrease dose with approximately one order of magnitude, often to sub-millisievert dose levels. In parallel, advancements in computational technology have enabled the measurement of fractional flow reserve (FFR) from CCTA data (FFRCT). This technique shows potential to replace invasively measured FFR to select patients in need of coronary intervention. Furthermore, developments in scanner hardware have led to the introduction of dual-energy and photon-counting CT, which offer the possibility of material decomposition imaging. Dual-energy CT reduces beam hardening, which enables CCTA in patients with a high calcium burden and more robust myocardial CT perfusion imaging. Future-generation CT systems will be capable of counting individual X-ray photons. Photon-counting CT is promising and may result in a substantial further radiation dose reduction, vastly increased spatial resolution, and the introduction of a whole new class of contrast agents.

Dose reduction with iterative reconstruction for coronary CT angiography: a systematic review and meta-analysis

OBJECTIVE

To investigate the achievable radiation dose reduction for coronary CT angiography (CCTA) with iterative reconstruction (IR) in adults and the effects on image quality.

METHODS

PubMed and EMBASE were searched, and original articles concerning IR for CCTA in adults using prospective electrocardiogram triggering were included. Primary outcome was the effective dose using filtered back projection (FBP) and IR. Secondary outcome was the effect of IR on objective and subjective image quality.

RESULTS

The search yielded 1616 unique articles, of which 10 studies (1042 patients) were included. The pooled routine effective dose with FBP was 4.2 mSv [95% confidence interval (CI) 3.5-5.0]. A dose reduction of 48% to a pooled effective dose of 2.2 mSv (95% CI 1.3-3.1) using IR was reported. Noise, contrast-to-noise ratio and subjective image quality were equal or improved in all but one study, whereas signal-to-noise ratio was decreased in two studies with IR at reduced dose.

CONCLUSION

IR allows for CCTA acquisition with an effective dose of 2.2 mSv with preserved objective and subjective image quality.

Achievable dose reduction using iterative reconstruction for chest computed tomography: A systematic review

OBJECTIVES

Iterative reconstruction (IR) allows for dose reduction with maintained image quality in CT imaging. In this systematic review the reported effective dose reductions for chest CT and the effects on image quality are investigated.

METHODS

A systematic search in PubMed and EMBASE was performed. Primary outcome was the reported local reference and reduced effective dose and secondary outcome was the image quality with IR. Both non contrast-enhanced and enhanced studies comparing reference dose with reduced dose were included.

RESULTS

24 studies were included. The median number of patients per study was 66 (range 23-200) with in total 1806 patients. The median reported local reference dose of contrast-enhanced chest CT with FBP was 2.6 (range 1.5-21.8) mSv. This decreased to 1.4 (range 0.4-7.3) mSv at reduced dose levels using IR. With non contrast-enhanced chest CT the dose decreased from 3.4 (range 0.7-7.8) mSv to 0.9 (range 0.1-4.5) mSv. Objective mage quality and diagnostic confidence and acceptability remained the same or improved with IR compared to FBP in most studies while data on diagnostic accuracy was limited.

CONCLUSION

Radiation dose can be reduced to less than 2 mSv for contrast-enhanced chest CT and non contrast-enhanced chest CT is possible at a submillisievert dose using IR algorithms.

The impact of a new model-based iterative reconstruction algorithm on prosthetic heart valve related artifacts at reduced radiation dose MDCT

To assess the impact of hybrid iterative reconstruction (IR) and novel model-based iterative reconstruction (IMR) and dose reduction on prosthetic heart valve (PHV) related artifacts and objective image quality. One transcatheter and two mechanical PHVs were embedded in diluted contrast-gel, inserted in an anthropomorphic phantom and imaged stationary with retrospectively ECG-gated computed tomography. Eight acquisitions were obtained of each PHV at 120 kV, 600 mAs (routine), 300 and 150 mAs (reduced dose). Data were reconstructed with filtered back projection (FBP), IR and IMR. Hypodense and hyperdense artifact volumes were quantified using two threshold filters. Signal-to-noise (SNR) and contrast-to-noise (CNR) ratios were calculated. Artifact volumes differed significantly between reconstruction algorithms for all PHVs (P < 0.005). Compared to FBP, IR decreased overall hypodense and hyperdense artifact volumes; at 150 mAs by 53 and 20 % (IR) and 67 and 23 % (IMR), respectively and significantly increased SNR and CNR at all doses (P < 0.012). Even at reduced dose, IMR resulted in higher image quality than routine dose FBP and IR. Iterative reconstruction and particularly IMR significantly reduce PHV-related artifacts and improve objective image quality in non-pulsatile conditions, even in reduced-dose images. Also, this study suggests that IMR allows for more radiation dose reduction in comparison to hybrid IR while maintaining high image quality.

The effects of computed tomography with iterative reconstruction on solid pulmonary nodule volume quantification

Background

The objectives of this study were to evaluate the influence of iterative reconstruction (IR) on pulmonary nodule volumetry with chest computed tomography (CT).

Methods

Twenty patients (12 women and 8 men, mean age 61.9, range 32–87) underwent evaluation of pulmonary nodules with a 64-slice CT-scanner. Data were reconstructed using filtered back projection (FBP) and IR (Philips Healthcare, iDose⁴-levels 2, 4 and 6) at similar radiation dose. Volumetric nodule measurements were performed with semi-automatic software on thin slice reconstructions. Only solid pulmonary nodules were measured, no additional selection criteria were used for the nature of nodules. For intra-observer and inter-observer variability, measurements were performed once by one observer and twice by another observer. Algorithms were compared using the concordance correlation-coefficient (p_c) and Friedman-test, and post-hoc analysis with the Wilcoxon-signed ranks-test with Bonferroni-correction (significance-level p<0.017).

Results

Seventy-eight nodules were present including 56 small nodules (volume<200 mm³, diameter<8 mm) and 22 large nodules (volume≥200 mm³, diameter≥8 mm). No significant differences in measured pulmonary nodule volumes between FBP, iDose⁴-levels 2, 4 and 6 were found in both small nodules and large nodules. FBP and iDose⁴-levels 2, 4 and 6 were correlated with p_c-values of 0.98 or higher for both small and large nodules. P_c-values of intra-observer and inter-observer variability were 0.98 or higher.

Conclusions

Measurements of solid pulmonary nodule volume measured with standard-FBP were comparable with IR, regardless of the IR-level and no significant differences between measured volumes of both small and large solid nodules were found.

Iterative reconstruction techniques for computed tomography part 2: initial results in dose reduction and image quality

OBJECTIVES

To present the results of a systematic literature search aimed at determining to what extent the radiation dose can be reduced with iterative reconstruction (IR) for cardiopulmonary and body imaging with computed tomography (CT) in the clinical setting and what the effects on image quality are with IR versus filtered back-projection (FBP) and to provide recommendations for future research on IR.

METHODS

We searched Medline and Embase from January 2006 to January 2012 and included original research papers concerning IR for CT.

RESULTS

The systematic search yielded 380 articles. Forty-nine relevant studies were included. These studies concerned: the chest(n = 26), abdomen(n = 16), both chest and abdomen(n = 1), head(n = 4), spine(n = 1), and no specific area (n = 1). IR reduced noise and artefacts, and it improved subjective and objective image quality compared to FBP at the same dose. Conversely, low-dose IR and normal-dose FBP showed similar noise, artefacts, and subjective and objective image quality. Reported dose reductions ranged from 23 to 76 % compared to locally used default FBP settings. However, IR has not yet been investigated for ultra-low-dose acquisitions with clinical diagnosis and accuracy as endpoints.

CONCLUSION

Benefits of IR include improved subjective and objective image quality as well as radiation dose reduction while preserving image quality. Future studies need to address the value of IR in ultra-low-dose CT with clinically relevant endpoints.

KEY POINTS

• Iterative reconstruction improves image quality of CT images at equal acquisition parameters. • IR preserves image quality compared to normal-dose filtered back-projection. • The reduced radiation dose made possible by IR is advantageous for patients. • IR has not yet been investigated with clinical diagnosis and accuracy as endpoints.

Iterative reconstruction improves evaluation of native aortic and mitral valves by retrospectively ECG-gated thoracoabdominal CTA

OBJECTIVES

To compare native aortic (AV) and mitral valve (MV) image quality on limited-dose retrospectively ECG-gated CTA of the thoracoabdominal aorta reconstructed with iterative reconstruction (IR) and filtered back projection (FBP).

METHODS

Fifty patients underwent routine care retrospectively ECG-gated thoracoabdominal limited-dose 256-slice CTA. At 30 % (systole) and 75 % (diastole) of the R-R interval AV and MV were reconstructed using FBP and IR. Objective image quality [density and noise (SD of density measurement)] was measured. Two independent observers scored subjective valve image quality using four-point Likert scales.

RESULTS

IR significantly decreased image noise, but did not alter the aorta and interventricular septum density. Interobserver variability was moderate to good. Valve image quality was scored at least moderate in most cases. IR scored one or two Likert scale points higher than FBP in 10 (first observer) and 27 (second observer) scores. Conversely, IR scored one Likert scale point lower than FBP in 1 (first observer) and 4 (second observer) scores.

CONCLUSIONS

Limited-dose retrospectively ECG-gated thoracoabdominal CTA enables moderate to excellent evaluation of AV and MV in most patients, in addition to the primary diagnostic question. Image quality is further improved by IR.

Fast noise reduction in computed tomography for improved 3-D visualization

Computed tomography (CT) has a trend towards higher resolution and higher noise. This development has increased the interest in anisotropic smoothing techniques for CT, which aim to reduce noise while preserving structures of interest. However, existing smoothing techniques are slow, which makes clinical application difficult. Furthermore, the published methods have limitations with respect to preserving small details in CT data. This paper presents a widely applicable speed optimized framework for anisotropic smoothing techniques. A second contribution of this paper is an extension to an existing smoothing technique aimed at better preserving small structures of interest in CT data. Based on second-order image structure, the method first determines an importance map, which indicates potentially relevant structures that should be preserved. Subsequently an anisotropic diffusion process is started. The diffused data is used in most parts of the images, while structures with significant second-order information are preserved. The method is qualitatively evaluated against an anisotropic diffusion method without structure preservation in an observer study to assess the improvement of 3-D visualizations of CT series and quantitatively by determining the reduction of the difference between low and high dose CT scans of in vitro carotid plaques.

Filter learning: application to suppression of bony structures from chest radiographs

A novel framework for image filtering based on regression is presented. Regression is a supervised technique from pattern recognition theory in which a mapping from a number of input variables (features) to a continuous output variable is learned from a set of examples from which both input and output are known. We apply regression on a pixel level. A new, substantially different, image is estimated from an input image by computing a number of filtered input images (feature images) and mapping these to the desired output for every pixel in the image. The essential difference between conventional image filters and the proposed regression filter is that the latter filter is learned from training data. The total scheme consists of preprocessing, feature computation, feature extraction by a novel dimensionality reduction scheme designed specifically for regression, regression by k-nearest neighbor averaging, and (optionally) iterative application of the algorithm. The framework is applied to estimate the bone and soft-tissue components from standard frontal chest radiographs. As training material, radiographs with known soft-tissue and bone components, obtained by dual energy imaging, are used. The results show that good correlation with the true soft-tissue images can be obtained and that the scheme can be applied to images from a different source with good results. We show that bone structures are effectively enhanced and suppressed and that in most soft-tissue images local contrast of ribs decreases more than contrast between pulmonary nodules and their surrounding, making them relatively more pronounced.

Local noise weighted filtering for emphysema scoring of low-dose CT images

Computed tomography (CT) has become the new reference standard for quantification of emphysema. The most popular measure of emphysema derived from CT is the pixel index (PI), which expresses the fraction of the lung volume with abnormally low intensity values. As PI is calculated from a single, fixed threshold on intensity, this measure is strongly influenced by noise. This effect shows up clearly when comparing the PI score of a high-dose scan to the PI score of a low-dose (i.e., noisy) scan of the same subject. In this paper, the noise variance (NOVA) filter is presented: a general framework for (iterative) nonlinear filtering, which uses an estimate of the spatially dependent noise variance in an image. The NOVA filter iteratively estimates the local image noise and filters the image. For the specific purpose of emphysema quantification of low-dose CT images, a dedicated, noniterative NOVA filter is constructed by using prior knowledge of the data to obtain a good estimate of the spatially dependent noise in an image. The performance of the NOVA filter is assessed by comparing characteristics of pairs of high-dose and low-dose scans. The compared characteristics are the PI scores for different thresholds and the size distributions of emphysema bullae. After filtering, the PI scores of high-dose and low-dose images agree to within 2%-3% points. The reproducibility of the high-dose bullae size distribution is also strongly improved. NOVA filtering of a CT image of typically 400 x 512 x 512 voxels takes only a couple of minutes which makes it suitable for routine use in clinical practice.

A computer-aided diagnosis system for detection of lung nodules in chest radiographs with an evaluation on a public database

A computer algorithm for nodule detection in chest radiographs is presented. The algorithm consists of four main steps: (i) image preprocessing; (ii) nodule candidate detection; (iii) feature extraction; (iv) candidate classification. Two optional extensions to this scheme are tested: candidate selection and candidate segmentation. The output of step (ii) is a list of circles, which can be transformed into more detailed contours by the extra candidate segmentation step. In addition, the candidate selection step (which is a classification step using a small number of features) can be used to reduce the list of nodule candidates before step (iii). The algorithm uses multi-scale techniques in several stages of the scheme: Candidates are found by looking for local intensity maxima in Gaussian scale space; nodule boundaries are detected by tracing edge points found at large scales down to pixel scale; some of the features used for classification are taken from a multi-scale Gaussian filterbank. Experiments with this scheme (with and without the segmentation and selection steps) are carried out on a previously characterized, publicly available database, that contains a large number of very subtle nodules. For this database, counting as detections only those nodules that were indicated with a confidence level of 50% or more, radiologists previously detected 70% of the nodules. For our algorithm, it turns out that the selection step does have an added value for the system, while segmentation does not lead to a clear improvement. With the scheme with the best performance, accepting on average two false positives per image results in the identification of 51% of all nodules. For four false positives, this increases to 67%. This is close to the previously reported 70% detection rate of the radiologists.