Usefulness of computerized method for lung nodule detection on digital chest radiographs using similar subtraction images from different patients

Purpose

The purpose of this study is to evaluate the usefulness of a novel computerized method to select automatically the similar chest radiograph for image subtraction in the patients who have no previous chest radiographs and to assist the radiologists’ interpretation by presenting the “similar subtraction image” from different patients.

Materials and methods

Institutional review board approval was obtained, and the requirement for informed patient consent was waived. A large database of approximately 15,000 normal chest radiographs was used for searching similar images of different patients. One hundred images of candidates were selected according to two clinical parameters and similarity of the lung field in the target image. We used the correlation value of chest region in the 100 images for searching the most similar image. The similar subtraction images were obtained by subtracting the similar image selected from the target image. Thirty cases with lung nodules and 30 cases without lung nodules were used for an observer performance test. Four attending radiologists and four radiology residents participated in this observer performance test.

Results

The AUC for all radiologists increased significantly from 0.925 to 0.974 with the CAD (P = .004). When the computer output images were available, the average AUC for the residents was more improved (0.960 vs. 0.890) than for the attending radiologists (0.987 vs. 0.960).

Conclusion

The novel computerized method for lung nodule detection using similar subtraction images from different patients would be useful to detect lung nodules on digital chest radiographs, especially for less experienced readers.     

Influence of radiation dose and iterative reconstruction algorithms for measurement accuracy and reproducibility of pulmonary nodule volumetry: A phantom study

Purpose

To evaluate the influence of radiation dose settings and reconstruction algorithms on the measurement accuracy and reproducibility of semi-automated pulmonary nodule volumetry.

Materials and methods

CT scans were performed on a chest phantom containing various nodules (10 and 12 mm; +100, −630 and −800 HU) at 120 kVp with tube current–time settings of 10, 20, 50, and 100 mAs. Each CT was reconstructed using filtered back projection (FBP), iDose⁴ and iterative model reconstruction (IMR). Semi-automated volumetry was performed by two radiologists using commercial volumetry software for nodules at each CT dataset. Noise, contrast-to-noise ratio and signal-to-noise ratio of CT images were also obtained. The absolute percentage measurement errors and differences were then calculated for volume and mass. The influence of radiation dose and reconstruction algorithm on measurement accuracy, reproducibility and objective image quality metrics was analyzed using generalized estimating equations.

Results

Measurement accuracy and reproducibility of nodule volume and mass were not significantly associated with CT radiation dose settings or reconstruction algorithms (p > 0.05). Objective image quality metrics of CT images were superior in IMR than in FBP or iDose⁴ at all radiation dose settings (p < 0.05).

Conclusion

Semi-automated nodule volumetry can be applied to low- or ultralow-dose chest CT with usage of a novel iterative reconstruction algorithm without losing measurement accuracy and reproducibility.     

Detectability of simulated pulmonary nodules on chest radiographs: Comparison between irradiation side sampling indirect flat-panel detector and computed radiography

Objective

To compare the detectability of simulated pulmonary nodules on chest radiographs between an irradiation side sampling indirect flat-panel detector (ISS-FPD) and computed radiography (CR).

Materials and methods

This study was an observer performance study. Simulated pulmonary nodules of 8 mm in diameter were superimposed on an anthropomorphic chest phantom. Chest radiographs were acquired under 2 exposure levels (4 and 3.2 mAs) with the ISS-FPD and the CR. Six thoracic radiologists evaluated all 40 images (10 patterns × 2 different exposure doses × 2 different systems) for the presence or absence of a lesion over each of 12 defined areas on a 3-megapixel monochrome liquid-crystal display. Receiver operating characteristic (ROC) curves were obtained for observation in predefined 480 areas. A jackknife method was used for statistical analysis. Differences with a P value of <0.05 were considered significant.

Results

The analysis of the observer detection of simulated pulmonary nodules showed larger areas under the ROC curve (AUC) by the ISS-FPD than by the CR. There was a statistically significant difference between the two systems at 3.2 mAs (P = 0.0330).

Conclusion

The ISS-FPD was superior to the CR for the detection of simulated pulmonary nodules at 3.2 mAs.     

Standard versus inverted digital luminescence radiography in detecting pulmonary nodules: A ROC analysis

Objective

Observer performance tests were conducted to compare the effect of presenting digital luminescence radiography (DLR) monitor images in positive mode (“bones black”) or negative mode (“bones white”) in the detectability of subtle lung nodules.

Methods

Five radiologists independently reviewed digital radiographs of 55 patients with either (a) a single, small (6–12 mm), non- calcified peripheral nodule confirmed by chest CT (n = 47) or (b) normal finding (n = 8) confirmed by chest CT, respectively. Mean size of pulmonary nodules was 8.3 mm (range 6–12 mm, median 7 mm). Radiographs were displayed either in standard mode (bones white) or inverse intensity (bones black). A total of 550 observations resulted. For the evaluation ROC analysis was performed using a specialized computer algorithm.

Results

The standard presentation mode showed a sensitivity of 61.7% and a specificity of 72.5%, the inverse presentation mode a sensitivity of 68.1% and a specificity of 75.0%. ROC- analysis showed that the difference between the standard mode (Az- value 0.694) and the inverse mode (Az- value 0.810) was significant in favour of the inverse presentation mode (p = 0.001). This finding was especially observed in more experienced radiologist (Az- values 0.716 vs. 0.909, P < 0.001).

Conclusion

Our study demonstrates a significant advantage of the inverse mode in the detection of small pulmonary nodules compared with the commonly used negative mode when interpreted by more experienced radiologists.

Advance in knowledge

Inverse intensity images (“bones black”) may have some advantages in the detection of small pulmonary nodules in experienced readers when presented on a stand-alone display.     

Quantitative assessment of the influence of anatomic noise on the detection of subtle lung nodule in digital chest radiography using fractal-feature distance

Purpose

To confirm whether or not the influence of anatomic noise on the detection of nodules in digital chest radiography can be evaluated by the fractal-feature distance.

Materials and methods

We used the square images with and without a simulated nodule which were generated in our previous observer performance study; the simulated nodule was located on the upper margin of a rib, the inside of a rib, the lower margin of a rib, or the central region between two adjoining ribs. For the square chest images, fractal analysis was conducted using the virtual volume method. The fractal-feature distances between the considered and the reference images were calculated using the pseudo-fractal dimension and complexity, and the square images without the simulated nodule were employed as the reference images. We compared the fractal-feature distances with the observer's confidence level regarding the presence of a nodule in plain chest radiograph.

Results

For all square chest images, the relationships between the length of the square boxes and the mean of the virtual volumes were linear on a log–log scale. For all types of the simulated nodules, the fractal-feature distance was the highest for the simulated nodules located on the central region between two adjoining ribs and was the lowest for those located in the inside of a rib. The fractal-feature distance showed a linear relation to an observer's confidence level.

Conclusion

The fractal-feature distance would be useful for evaluating the influence of anatomic noise on the detection of nodules in digital chest radiography.     

A comparison between the electronic magnification (EM) and true magnification (TM) of breast phantom images using a CDMAM phantom

Purpose

To provide a comparison between the image quality of electronically magnified (EM) and geometric, or true, magnification (TM) mammographic images.

Materials and methods

One Computed Radiography (CR), one Digital Radiography (DR) and two screen–film (S–F) imaging systems were investigated. A Contrast-Detail Mammography (CDMAM) phantom was used as a test object. Three contact images and three sets of TM images with a magnification factor of 1.8 were taken on all systems. Software was used to zoom the contact images by a factor of 1.8 to produce EM images. Two observers evaluated all of the images. An Image Quality Figure and contrast detail curve were used to analyze the observer data and Mann–Whitney U-tests were performed to determine the statistical significance of the results.

Results

No significant differences were found between soft copy and hard copy for any imaging modality. No significant difference in contrast detail detectability (CDD) was seen between EM images from the two digital systems and TM images on S–F systems. The results for the DR EM images and S–F TM images also showed no differences. The CDD of DR TM images was significantly better than both EM and S–F TM images.

Conclusion

Digitally zoomed images offer the same level of CDD as S–F TM images, and so may be viably used in their place. DR systems offer greater CDD than conventional S–F images, when comparing the TM images. This implies that doses can be greatly reduced for TM views using DR systems, while maintaining acceptable image quality.     

Liquid-Crystal Display Monitors and Cathode-Ray Tube Monitors: A Comparison of Observer Performance in the Detection of Small Solitary Pulmonary Nodules

Objective

To compare observer performance using liquid-crystal display (LCD) and cathode-ray tube (CRT) monitors in the interpretation of soft-copy chest radiographs for the detection of small solitary pulmonary nodules.

Materials and Methods

By reviewing our Medical Center''s radiologic information system, the eight radiologists participating in this study (three board-certified and five resident) retrospectively collected 40 chest radiographs showing a solitary noncalcified pulmonary nodule approximately 1 cm in diameter, and 40 normal chest radiographs. All were obtained using a storage-phosphor system, and CT scans of the same patients served as the gold standard for the presence of a pulmonary nodule. Digital images were displayed on both high-resolution LCD and CRT monitors. The readers were requested to rank each image using a five-point scale (1 = definitely negative, 3 = equivocal or indeterminate, 5 = definitely positive), and the data were interpreted using receiver operating characteristic (ROC) analysis.

Results

The mean area under the ROC curve was 0.8901±0.0259 for the LCD session, and 0.8716±0.0266 for the CRT session (p > 0.05). The reading time for the LCD session was not significantly different from that for the CRT session (37.12 and 41.46 minutes, respectively; p = 0.889).

Conclusion

For detecting small solitary pulmonary nodules, an LCD monitor and a CRT monitor are comparable.     

Local diagnostic reference levels for typical radiographic procedures

Purpose

To establish local diagnostic reference levels (DRL) for typical radiographic examinations in a fully digital imaging institution.

Methods

The initial survey included 6 standard radiographic projections performed in 19 computed radiography (CR) and digital radiography (DR) rooms. Because of the expected difference in the performance, the local reference levels were analysed separately for those 2 modalities. Data of 226 average size adult patients were included in the analysis. Entrance surface dose (ESD) was calculated from the recorded radiographic techniques and tube radiation output measurements. After observing wide variations in the results of the patient survey, the examinations were repeated by using anthropomorphic phantoms. Initial efforts to understand the reasons for dose variations were focused on CR chest, abdomen, pelvis, and lumbar spine examinations.

Results

The average size patient doses for similar examinations were lower in the DR rooms than in the CR rooms by factors that ranged from 1.2 to 3, with the exception of the chest examination. Standardization of the CR exposure index value allowed us to decrease ESD by 21%-30%. Detector sensitivity had an insignificant effect (2%) on ESD; proper collimation lowered the dose by 17%. However, the major effect, up to 46% difference, was found because of antiscatter grids cutoff.

Conclusion

Modality specific local diagnostic reference levels for standard examinations have been established in a large digital imaging department with hybrid modalities. Typically the local reference values were lower than those recommended in Safety Code 35, except for CR chests. Factors that affect the dose variations have been investigated and determined.     

Effect of radiation dose and iterative reconstruction on lung lesion conspicuity at MDCT: Does one size fit all?

Objective

To evaluate the effect of different acquisition parameters and reconstruction algorithms in lung lesions conspicuity in chest MDCT.

Methods

An anthropomorphic chest phantom containing 6 models of lung disease (ground glass opacity, bronchial polyp, solid nodule, ground glass nodule, emphysema and tree-in-bud) was scanned using 80, 100 and 120 kVp, with fixed mAs ranging from 10 to 110. The scans were reconstructed using filtered back projection (FBP) and iterative reconstruction (IR) algorithms. Three blinded thoracic radiologists reviewed the images and scored lesions conspicuity and overall image quality. Image noise and radiation dose parameters were recorded.

Results

All acquisitions with 120 kVp received a score of 3 (acceptable) or higher for overall image quality. There was no significant difference between IR and FBP within each setting for overall image quality (p > 0.05), even though image noise was significantly lower using IR (p < 0.0001). When comparing specific lower radiation acquisition parameters 100 kVp/10 mAs [Effective Dose (ED): 0.238 mSv] vs 120 kVp/10 mAs (ED: 0.406 mSv) vs 80 kVp/40 mAs (ED: 0.434 mSv), we observed significant difference in lesions conspicuity (p < 0.02), as well as significant difference in overall image quality, independent of the reconstruction algorithm (p < 0.02), with higher scores on the 120 kV/10 mAs setting. Tree-in-bud pattern, ground glass nodule and ground glass opacity required lower radiation doses to get a diagnostic score using IR when compared to FBP.

Conclusion

Designing protocols for specific lung pathologies using lower dose acquisition parameters is feasible, and by applying iterative reconstruction, radiologists may have better diagnostic confidence to evaluate some lesions in very low dose settings, preserving acceptable image quality.     

Functional MRI using Fourier decomposition of lung signal: Reproducibility of ventilation- and perfusion-weighted imaging in healthy volunteers

Purpose

To assess the reproducibility of Fourier decomposition (FD) based ventilation- and perfusion-weighted lung MRI.

Methods

Sixteen healthy volunteers were examined on a 1.5 T whole-body MR-scanner with 4–6 sets of coronal slices over the chest volume with a non-contrast enhanced steady-state free precession sequence. The identical protocol was repeated after 24 h. Reconstructed perfusion- and ventilation-weighted images were obtained through non-rigid registration and FD post-processing of images. Analysis of signal in segmented regions of interest was performed for both native and post-processed data. Two blinded chest radiologists rated image quality of perfusion- and ventilation-weighted images using a 3-point scale.

Results

Reproducibility of signal between the two time points was very good with intra-class correlation coefficients of 0.98, 0.94 and 0.86 for native, perfusion- and ventilation-weighted images, respectively. Perfusion- and ventilation-weighted images were of overall good quality with proportions of diagnostic images of 87–95% and 69–75%, respectively. Lung signal decreased from posterior to anterior slices with image quality of ventilation-weighted images in anterior areas rated worse than in posterior or perfusion-weighted images. Inter- and intra-observer agreement of image quality was good for perfusion and ventilation.

Conclusions

The study demonstrates high reproducibility of ventilation- and perfusion-weighted FD lung MRI.     

Can real-time ultrasound elastography using the color score and strain ratio differentiate between benign and malignant solitary thyroid nodules?

Background

Solitary thyroid nodule may represent a multitude of thyroid disorders; therefore, detection of whether these nodules are benign or malignant is crucial for patient’s triage.

Objective

To evaluate the diagnostic performance of the latest generation of real-time ultrasound elastography (USE) in differentiation between benign and malignant solitary thyroid nodules.

Materials and methods

Thirty consecutive patients who were referred for surgical treatment were prospectively examined by real-time USE. Tissue stiffness on real-time USE was determined with light compression using the standard elastography color scoring system according to Rago criteria ranging from 1 (low stiffness over the entire nodule) to 5 (high stiffness over the entire nodule and surrounding tissue). The strain ratio (normal tissue to lesion strain ratio) was calculated. The histopathological examination of these resected nodules was used as the diagnostic standard of reference.

Results

Scores of 1 and 2 with Rago criteria were highly significant seen in benign nodules, whereas, scores of 4 and 5 with Rago criteria were highly significant seen in malignant nodules (p < 0.001) with a sensitivity, specificity and diagnostic accuracy of 78.6%, 78.9% and 78.8% respectively. Additionally, the best strain ratio cut-off value for discrimination between benign and malignant nodules by using receiver operating characteristic analysis was 2.20 (area under the curve of 0.861; p value <0.001) with a consequential sensitivity, specificity and diagnostic accuracy of 85.7%, 90.5% and 88.6% respectively.

Conclusion

Both the color score and the strain ratio are higher in malignant solitary thyroid nodules than those in benign ones. Consequently, real-time USE can be used for the differentiation of benign and malignant solitary thyroid nodules. Eventually, this reduces the number of superfluous surgical procedures on benign thyroid nodules.     

Regional variance of visually lossless threshold in compressed chest CT images: Lung versus mediastinum and chest wall

Objective

To estimate the visually lossless threshold (VLT) for the Joint Photographic Experts Group (JPEG) 2000 compression of chest CT images and to demonstrate the variance of the VLT between the lung and mediastinum/chest wall.

Subjects and methods

Eighty images were compressed reversibly (as negative control) and irreversibly to 5:1, 10:1, 15:1 and 20:1. Five radiologists determined if the compressed images were distinguishable from their originals in the lung and mediastinum/chest wall. Exact tests for paired proportions were used to compare the readers’ responses between the reversible and irreversible compressions and between the lung and mediastinum/chest wall.

Results

At reversible, 5:1, 10:1, 15:1, and 20:1 compressions, 0%, 0%, 3–49% (p < .004, for three readers), 69–99% (p < .001, for all readers), and 100% of the 80 image pairs were distinguishable in the lung, respectively; and 0%, 0%, 74–100% (p < .001, for all readers), 100%, and 100% were distinguishable in the mediastinum/chest wall, respectively. The image pairs were less frequently distinguishable in the lung than in the mediastinum/chest wall at 10:1 (p < .001, for all readers) and 15:1 (p < .001, for two readers). In 321 image comparisons, the image pairs were indistinguishable in the lung but distinguishable in the mediastinum/chest wall, whereas there was no instance of the opposite.

Conclusion

For JPEG2000 compression of chest CT images, the VLT is between 5:1 and 10:1. The lung is more tolerant to the compression than the mediastinum/chest wall.     

Bone suppressed images improve radiologists’ detection performance for pulmonary nodules in chest radiographs

Objectives

To assess the effect of bone suppression imaging on observer performance in detecting lung nodules in chest radiographs.

Materials and methods

Posteroanterior (PA) and lateral digital chest radiographs of 111 (average age 65) patients with a CT proven solitary nodule (median diameter 15 mm), and 189 (average age 63) controls were read by 5 radiologists and 3 residents. Conspicuity of nodules on the radiographs was classified in obvious (n = 32), moderate (n = 32), subtle (n = 29) and very subtle (n = 18). Observers read the PA and lateral chest radiographs without and with an additional PA bone suppressed image (BSI) (ClearRead Bone Suppression 2.4, Riverain Technologies, Ohio) within one reading session. Multi reader multi case (MRMC) receiver operating characteristics (ROC) were used for statistical analysis.

Results

ROC analysis showed improved detection with use of BSI compared to chest radiographs alone (AUC = 0.883 versus 0.855; p = 0.004). Performance also increased at high specificities exceeding 80% (pAUC = 0.136 versus 0.124; p = 0.0007). Operating at a specificity of 90%, sensitivity increased with BSI from 66% to 71% (p = 0.0004). Increase of detection performance was highest for nodules with moderate and subtle conspicuity (p = 0.02; p = 0.03).

Conclusion

Bone suppressed images improve radiologists’ detection performance for pulmonary nodules, especially for those of moderate and subtle conspicuity.     

Dose reduction in chest CT: Comparison of the adaptive iterative dose reduction 3D,adaptive iterative dose reduction,and filtered back projection reconstruction techniques

Objectives

To assess the effectiveness of adaptive iterative dose reduction (AIDR) and AIDR 3D in improving the image quality in low-dose chest CT (LDCT).

Materials and methods

Fifty patients underwent standard-dose chest CT (SDCT) and LDCT simultaneously, performed under automatic exposure control with noise index of 19 and 38 (for a 2-mm slice thickness), respectively. The SDCT images were reconstructed with filtered back projection (SDCT-FBP images), and the LDCT images with FBP, AIDR and AIDR 3D (LDCT-FBP, LDCT-AIDR and LDCT-AIDR 3D images, respectively). On all the 200 lung and 200 mediastinal image series, objective image noise and signal-to-noise ratio (SNR) were measured in several regions, and two blinded radiologists independently assessed the subjective image quality. Wilcoxon's signed rank sum test with Bonferroni's correction was used for the statistical analyses.

Results

The mean dose reduction in LDCT was 64.2% as compared with the dose in SDCT. LDCT-AIDR 3D images showed significantly reduced objective noise and significantly increased SNR in all regions as compared to the SDCT-FBP, LDCT-FBP and LDCT-AIDR images (all, P ≤ 0.003). In all assessments of the image quality, LDCT-AIDR 3D images were superior to LDCT-AIDR and LDCT-FBP images. The overall diagnostic acceptability of both the lung and mediastinal LDCT-AIDR 3D images was comparable to that of the lung and mediastinal SDCT-FBP images.

Conclusions

AIDR 3D is superior to AIDR. Intra-individual comparisons between SDCT and LDCT suggest that AIDR 3D allows a 64.2% reduction of the radiation dose as compared to SDCT, by substantially reducing the objective image noise and increasing the SNR, while maintaining the overall diagnostic acceptability.     

Multidetector CT chest with bronchial and pulmonary angiography determining causes,site and vascular origin of bleeding in patients with hemoptysis

Objective

To assess the role of MDCT chest with bronchial and pulmonary angiography in determining the cause, site of bleeding, and its vascular origin in patients presenting with hemoptysis.

Materials and methods

Fifty patients suffering from hemoptysis were evaluated by MDCT with bronchial and pulmonary angiographic techniques.

Results

MDCT chest with angiography revealed the cause in 84% of cases, the site and vascular origin in 76% of cases presenting with hemoptysis.

Conclusion

MDCT of the chest with bronchial and pulmonary angiography is considered a primary noninvasive imaging modality in the evaluation of patients with hemoptysis. It also serves as a guide for other diagnostic or therapeutic procedures.     

Does PACS improve diagnostic accuracy in chest radiograph interpretations in clinical practice?

Objectives

To assess the impact of a Picture Archiving and Communication System (PACS) on the diagnostic accuracy of the interpretation of chest radiology examinations in a “real life” radiology setting.

Materials and methods

During a period before PACS was introduced to radiologists, when images were still interpreted on film and reported on paper, images and reports were also digitally stored in an image database. The same database was used after the PACS introduction. This provided a unique opportunity to conduct a blinded retrospective study, comparing sensitivity (the main outcome parameter) in the pre and post-PACS periods.We selected 56 digitally stored chest radiograph examinations that were originally read and reported on film, and 66 examinations that were read and reported on screen 2 years after the PACS introduction. Each examination was assigned a random number, and both reports and images were scored independently for pathological findings. The blinded retrospective score for the original reports were then compared with the score for the images (the gold standard).

Results

Sensitivity was improved after the PACS introduction. When both certain and uncertain findings were included, this improvement was statistically significant. There were no other statistically significant changes.

Conclusion

The result is consistent with prospective studies concluding that diagnostic accuracy is at least not reduced after PACS introduction. The sensitivity may even be improved.     

Integration of interactive three-dimensional image post-processing software into undergraduate radiology education effectively improves diagnostic skills and visual-spatial ability

Purpose

Integrating interactive three-dimensional post-processing software into undergraduate radiology teaching might be a promising approach to synergistically improve both visual-spatial ability and radiological skills, thereby reducing students’ deficiencies in image interpretation. The purpose of this study was to test our hypothesis that a hands-on radiology course for medical students using interactive three-dimensional image post-processing software improves radiological knowledge, diagnostic skills and visual-spatial ability.

Materials and methods

A hands-on radiology course was developed using interactive three-dimensional image post-processing software. The course consisted of seven seminars held on a weekly basis. The 25 participating fourth- and fifth-year medical students learnt to systematically analyse cross-sectional imaging data and correlated the two-dimensional images with three-dimensional reconstructions. They were instructed by experienced radiologists and collegiate tutors. The improvement in radiological knowledge, diagnostic skills and visual-spatial ability was assessed immediately before and after the course by multiple-choice tests comprising 64 questions each. Wilcoxon signed rank test for paired samples was applied.

Results

The total number of correctly answered questions improved from 36.9 ± 4.8 to 49.5 ± 5.4 (p < 0.001) which corresponded to a mean improvement of 12.6 (95% confidence interval 9.9–15.3) or 19.8%. Radiological knowledge improved by 36.0% (p < 0.001), diagnostic skills for cross-sectional imaging by 38.7% (p < 0.001), diagnostic skills for other imaging modalities – which were not included in the course – by 14.0% (p = 0.001), and visual-spatial ability by 11.3% (p < 0.001).

Conclusion

The integration of interactive three-dimensional image post-processing software into undergraduate radiology education effectively improves radiological reasoning, diagnostic skills and visual-spatial ability, and thereby even diagnostic skills for imaging modalities not included in the course.     

Comparison of chest radiography,chest digital tomosynthesis and low dose MDCT to detect small ground-glass opacity nodules: an anthropomorphic chest phantom study

Objectives

The purpose of this study was to evaluate the diagnostic performance of chest radiography (CXR), chest digital tomosynthesis (DT) and low dose multidetector computed tomography (LDCT) for the detection of small pulmonary ground-glass opacity (GGO) nodules, using an anthropomorphic chest phantom.

Methods

Artificial pulmonary nodules were placed in a phantom and a total of 40 samples of different nodule settings underwent CXR, DT and LDCT. The images were randomly read by three experienced chest radiologists. Free-response receiver-operating characteristics (FROC) were used.

Results

The figures of merit for the FROC curves averaged for the three observers were 0.41, 0.37 and 0.76 for CXR, DT and LDCT, respectively. FROC analyses revealed significantly better performance of LDCT over CXR or DT for the detection of GGO nodules (P?P?=?0.73).

Conclusion

The diagnostic performance of DT for the detection of pulmonary small GGO nodules was not significantly different from that of CXR, but LDCT performed significantly better than both CXR and DT. DT is not a suitable alternative to CT for small GGO nodule detection, and LDCT remains the method of choice for this purpose.

Key Points

? For GGO nodule detection, DT was not significantly different from CXR. ? DT is not a suitable alternative to CT for GGO nodule detection. ? LDCT is the method of choice for GGO nodule detection.     

Computer-aided detection of lung nodules on multidetector CT in concurrent-reader and second-reader modes: A comparative study

Purpose

To compare the reading times and detection performances of radiologists in concurrent-reader and second-reader modes of computer-aided detection (CAD) for lung nodules on multidetector computed tomography (CT).

Materials and Methods

Fifty clinical multidetector CT datasets containing nodules up to 20 mm in diameter were retrospectively collected. For the detection and rating of non-calcified nodules larger than 4 mm in diameter, 6 radiologists (3 experienced radiologists and 3 resident radiologists) independently interpreted these datasets twice, once with concurrent-reader CAD and once with second-reader CAD. The reference standard of nodules in the datasets was determined by the consensus of two experienced chest radiologists. The reading times and detection performances in the two modes of CAD were statistically compared, where jackknife free-response receiver operating characteristic (JAFROC) analysis was used for the comparison of detection performances.

Results

Two hundreds and seven nodules constituted the reference standard. Reading time was significantly shorter in the concurrent-reader mode than in the second-reader mode, with the mean reading time for the 6 radiologists being 132 s with concurrent-reader CAD and 210 s with second-reader CAD (p < 0.01). JAFROC analysis revealed no significant difference between the detection performances in the two modes, with the average figure-of-merit value for the 6 radiologists being 0.70 with concurrent-reader CAD and 0.72 with second-reader CAD (p = 0.35).

Conclusion

In CAD for lung nodules on multidetector CT, the concurrent-reader mode is more time-efficient than the second-reader mode, and there can be no significant difference between the two modes in terms of detection performance of radiologists.     

Computer-aided detection of pulmonary nodules: a comparative study using the public LIDC/IDRI database

Objectives

To benchmark the performance of state-of-the-art computer-aided detection (CAD) of pulmonary nodules using the largest publicly available annotated CT database (LIDC/IDRI), and to show that CAD finds lesions not identified by the LIDC’s four-fold double reading process.

Methods

The LIDC/IDRI database contains 888 thoracic CT scans with a section thickness of 2.5 mm or lower. We report performance of two commercial and one academic CAD system. The influence of presence of contrast, section thickness, and reconstruction kernel on CAD performance was assessed. Four radiologists independently analyzed the false positive CAD marks of the best CAD system.

Results

The updated commercial CAD system showed the best performance with a sensitivity of 82 % at an average of 3.1 false positive detections per scan. Forty-five false positive CAD marks were scored as nodules by all four radiologists in our study.

Conclusions

On the largest publicly available reference database for lung nodule detection in chest CT, the updated commercial CAD system locates the vast majority of pulmonary nodules at a low false positive rate. Potential for CAD is substantiated by the fact that it identifies pulmonary nodules that were not marked during the extensive four-fold LIDC annotation process.

Key Points

? CAD systems should be validated on public, heterogeneous databases. ? The LIDC/IDRI database is an excellent database for benchmarking nodule CAD. ? CAD can identify the majority of pulmonary nodules at a low false positive rate. ? CAD can identify nodules missed by an extensive two-stage annotation process.