Lung cancer screening with submillisievert chest CT: Potential pitfalls of pulmonary findings in different readers with various experience levels

Impact of Radiologist Experience on AI Annotation Quality in Chest Radiographs: A Comparative Analysis

Background/Objectives: In the burgeoning field of medical imaging and Artificial Intelligence (AI), high-quality annotations for training AI-models are crucial. However, there are still only a few large datasets, as segmentation is time-consuming, experts have limited time. This study investigates how the experience of radiologists affects the quality of annotations. Methods: We randomly collected 53 anonymized chest radiographs. Fifteen readers with varying levels of expertise annotated the anatomical structures of different complexity, pneumonic opacities and central venous catheters (CVC) as examples of pathologies and foreign material. The readers were divided into three groups of five. The groups consisted of medical students (MS), junior professionals (JP) with less than five years of working experience and senior professionals (SP) with more than five years of experience. Each annotation was compared to a gold standard consisting of a consensus annotation of three senior board-certified radiologists. We calculated the Dice coefficient (DSC) and Hausdorff distance (HD) to evaluate annotation quality. Inter- and intrareader variability and time dependencies were investigated using Intraclass Correlation Coefficient (ICC) and Ordinary Least Squares (OLS). Results: Senior professionals generally showed better performance, while medical students had higher variability in their annotations. Significant differences were noted, especially for complex structures (DSC Pneumonic Opacities as mean [standard deviation]: MS: 0.516 [0.246]; SP: 0.631 [0.211]). However, it should be noted that overall deviation and intraclass variance was higher for these structures even for seniors, highlighting the inherent limitations of conventional radiography. Experience showed a positive relationship with annotation quality for VCS and lung but was not a significant factor for other structures. Conclusions: Experience level significantly impacts annotation quality. Senior radiologists provided higher-quality annotations for complex structures, while less experienced readers could still annotate simpler structures with satisfying accuracy. We suggest a mixed-expertise approach, enabling the highly experienced to utilize their knowledge most effectively. With the increase in numbers of examinations, radiology will rely on AI support tools in the future. Therefore, economizing the process of data acquisition and AI-training; for example, by integrating less experienced radiologists, will help to meet the coming challenges.

The Effect of X-ray Dose Photon-Counting Detector Computed Tomography on Nodule Properties in a Lung Cancer Screening Cohort: A Prospective Study

OBJECTIVES

The aim of the study was to evaluate the effect of photon-counting detector (PCD-)CT dose reduction to x-ray equivalent levels on nodule detection, diameter, volume, and density compared to a low-dose reference standard using semiautomated and manual methods.

MATERIALS AND METHODS

Between February and July 2023, 101 prospectively enrolled participants underwent noncontrast same-study low- and chest x-ray-dose CT scans using PCD-CT. Patients who were not referred for lung cancer screening or nodule follow-up, as well as those with nodules smaller than 5 mm in diameter, were excluded. Nodule detection and measurement of nodule diameters and volumes was semiautomatically performed for low- and x-ray-dose scans using computer-aided diagnosis software. Additionally, 2 blinded readers manually measured largest nodule diameters and examined nodule density. Nodules were classified using Lung-RADS v2022. Image quality was assessed with subjective and objective measures.

RESULTS

Mean CTDIvol for x-ray dose scans was 0.11 ± 0.03 mGy, compared to 0.65 ± 0.15 mGy for low-dose images (P < 0.001). One hundred seventy-two nodules larger than 5 mm were detected in 53 of the 101 participants (32 male, 61.6 ± 12.5 years; 21 female, 60.3 ± 12.5 years). The semiautomated method had high overall sensitivity for nodule detection (0.94) on x-ray dose scans, with a higher sensitivity for solid nodules (>0.95) and lower for subsolid nodules (>0.86). Nodules not detected on x-ray dose scans were significantly smaller. Semiautomated measurements underestimated nodule diameter for solid nodules on x-ray dose scans (P = 0.01), but no significant effect for nodule volume was found (P = 0.775). Readers rated nodule density less dense on x-ray dose scans (R1: P < 0.001, R2: P = 0.006). There was no significant difference in nodule diameter for both readers between scan doses (R1: P = 0.141; R2: P = 0.554). There were good to excellent correlations between semiautomated and reader nodule diameters. Agreement and accuracy between low-dose and x-ray dose Lung-RADS classifications across methods were good (Cohens' к = 0.73, 0.62, 0.76 for semiautomated method, R1 and R2; resp. Accuracy: 0.82, 0.78, 0.85). No Lung-RADS classification changes were observed with semiautomated volumetric measurements of nodules.

CONCLUSIONS

Semiautomated nodule detection is highly sensitive in PCD-CT x-ray dose scans. Semiautomated nodule volume measurement is more robust to image quality changes than nodule diameter. Accurate semiautomated and manual nodule measurements are feasible on x-ray dose scans, but nodule density was in tendency underestimated. Nodule classification using Lung-RADS was shown to be accurate on x-ray dose scans.

Ultra-low dose computed tomography protocols using spectral shaping for lung cancer screening: Comparison with low-dose for volumetric LungRADS classification

PURPOSE

To compare Low-Dose Computed Tomography (LDCT) with four different Ultra-Low-Dose Computed Tomography (ULDCT) protocols for PN classification according to the Lung Reporting and Data System (LungRADS).

METHODS

Three hundred sixty-one participants of an ongoing lung cancer screening (LCS) underwent single-breath-hold double chest Computed Tomography (CT), including LDCT (120kVp, 25mAs; CTDIvol 1,62 mGy) and one ULDCT among: fully automated exposure control ("ULDCT₁"); fixed tube-voltage and current according to patient size ("ULDCT₂"); hybrid approach with fixed tube-voltage ("ULDCT₃") and tube current automated exposure control ("ULDCT₄"). Two radiologists (R1, R2) assessed LungRADS 2022 categories on LDCT, and then after 2 weeks on ULDCT using two different kernels (R1: Qr49_{ADMIRE 4}; R2: Br49_{ADMIRE 3}). Intra-subject agreement for LungRADS categories between LDCT and ULDCT was measured by the k-Cohen Index with Fleiss-Cohen weights.

RESULTS

LDCT-dominant PNs were detected in ULDCT in 87 % of cases on Qr49_{ADMIRE 4} and 88 % on Br49_{ADMIRE 3}. The intra-subject agreement was: κULDCT₁ = 0.89 [95 %CI 0.82-0.96]; κULDCT₂ = 0.90 [0.81-0.98]; κULDCT₃ = 0.91 [0.84-0.99]; κULDCT₄ = 0.88 [0.78-0.97] on Qr49_{ADMIRE 4}, and κULDCT₁ = 0.88 [0.80-0.95]; κULDCT₂ = 0.91 [0.86-0.96]; κULDCT₃ = 0.87 [0.78-0.95]; and κULDCT₄ = 0.88 [0.82-0.94] on Br49_{ADMIRE 3}. LDCT classified as LungRADS 4B were correctly identified as LungRADS 4B at ULDCT₃, with the lowest radiation exposure among the tested protocols (median effective doses were 0.31, 0.36, 0.27 and 0.37 mSv for ULDCT₁, ULDCT₂, ULDCT₃, and ULDCT₄, respectively).

CONCLUSIONS

ULDCT by spectral shaping allows the detection and characterization of PNs with an excellent agreement with LDCT and can be proposed as a feasible approach in LCS.

Comparability of Pulmonary Nodule Size Measurements among Different Scanners and Protocols: Should Diameter Be Favorized over Volume?

BACKGROUND

To assess the impact of the lung cancer screening protocol recommended by the European Society of Thoracic Imaging (ESTI) on nodule diameter, volume, and density throughout different computed tomography (CT) scanners.

METHODS

An anthropomorphic chest phantom containing fourteen different-sized (range 3-12 mm) and CT-attenuated (100 HU, -630 HU and -800 HU, termed as solid, GG1 and GG2) pulmonary nodules was imaged on five CT scanners with institute-specific standard protocols (P_S) and the lung cancer screening protocol recommended by ESTI (ESTI protocol, P_E). Images were reconstructed with filtered back projection (FBP) and iterative reconstruction (REC). Image noise, nodule density and size (diameter/volume) were measured. Absolute percentage errors (APEs) of measurements were calculated.

RESULTS

Using P_E, dosage variance between different scanners tended to decrease compared to P_S, and the mean differences were statistically insignificant (p = 0.48). P_S and P_E(REC) showed significantly less image noise than P_E(FBP) (p < 0.001). The smallest size measurement errors were noted with volumetric measurements in P_E(REC) and highest with diametric measurements in P_E(FBP). Volume performed better than diameter measurements in solid and GG1 nodules (p < 0.001). However, in GG2 nodules, this could not be observed (p = 0.20). Regarding nodule density, REC values were more consistent throughout different scanners and protocols.

CONCLUSION

Considering radiation dose, image noise, nodule size, and density measurements, we fully endorse the ESTI screening protocol including the use of REC. For size measurements, volume should be preferred over diameter.

Higher agreement between readers with deep learning CAD software for reporting pulmonary nodules on CT

Purpose

The aim was to evaluate the impact of CAD software on the pulmonary nodule management recommendations of radiologists in a cohort of patients with incidentally detected nodules on CT.

Methods

For this retrospective study, two radiologists independently assessed 50 chest CT cases for pulmonary nodules to determine the appropriate management recommendation, twice, unaided and aided by CAD with a 6-month washout period. Management recommendations were given in a 4-point grade based on the BTS guidelines. Both reading sessions were recorded to determine the reading times per case. A reduction in reading times per session was tested with a one-tailed paired t-test, and a linear weighted kappa was calculated to assess interobserver agreement.

Results

The mean age of the included patients was 65.0 ± 10.9. Twenty patients were male (40 %). For both readers 1 and 2, a significant reduction of reading time was observed of 33.4 % and 42.6 % (p < 0.001, p < 0.001). The linear weighted kappa between readers unaided was 0.61. Readers showed a better agreement with the aid of CAD, namely by a kappa of 0.84. The mean reading time per case was 226.4 ± 113.2 and 320.8 ± 164.2 s unaided and 150.8 ± 74.2 and 184.2 ± 125.3 s aided by CAD software for readers 1 and 2, respectively.

Conclusion

A dedicated CAD system for aiding in pulmonary nodule reporting may help improve the uniformity of management recommendations in clinical practice.

Detection of pulmonary nodules with scoutless fixed-dose ultra-low-dose CT: a prospective study

OBJECTIVES

To determine the accuracy of scoutless, fixed-dose ultra-low-dose (ULD) CT compared to standard-dose (SD) CT for pulmonary nodule detection and semi-automated nodule measurement, across different patient sizes.

METHODS

Sixty-three patients underwent ULD and SD CT. Two readers examined all studies visually and with computer-aided detection (CAD). Nodules detected on SD CT were included in the reference standard by consensus and stratified into 4 categories (nodule category, NODCAT) from the Dutch-Belgian Lung Cancer Screening trial (NELSON). Effects of NODCAT and patient size on nodule detection were determined. For each nodule, volume and diameter were compared between both scans.

RESULTS

The reference standard comprised 173 nodules. For both readers, detection rates on ULD versus SD CT were not significantly different for NODCAT 3 and 4 nodules > 50 mm³ (reader 1: 93% versus 89% (p = 0.257); reader 2: 96% versus 98% (p = 0.317)). For NODCAT 1 and 2 nodules < 50 mm³, detection rates on ULD versus SD CT dropped significantly (reader 1: 66% versus 80% (p = 0.023); reader 2: 77% versus 87% (p = 0.039)). Body mass index and chest circumference did not influence nodule detectability (p = 0.229 and p = 0.362, respectively). Calculated volumes and diameters were smaller on ULD CT (p < 0.0001), without altering NODCAT (84% agreement).

CONCLUSIONS

Scoutless ULD CT reliably detects solid lung nodules with a clinically relevant volume (> 50 mm³) in lung cancer screening, irrespective of patient size. Since detection rates were lower compared to SD CT for nodules < 50 mm³, its use for lung metastasis detection should be considered on a case-by-case basis.

KEY POINTS

• Detection rates of pulmonary nodules > 50 mm³are not significantly different between scoutless ULD and SD CT (i.e. volumes clinically relevant in lung cancer screening based on the NELSON trial), but were different for the detection of nodules < 50 mm³(i.e. volumes still potentially relevant in lung metastasis screening). • Calculated nodule volumes were on average 0.03 mL or 9% smaller on ULD CT, which is below the 20-25% interscan variability previously reported with software-based volumetry. • Even though a scoutless, fixed-dose ULD CT protocol was used (CTDI_vol0.15 mGy), pulmonary nodule detection was not influenced by patient size.

Personalized Chest Computed Tomography: Minimum Diagnostic Radiation Dose Levels for the Detection of Fibrosis, Nodules, and Pneumonia

OBJECTIVES

The purpose of this study was to evaluate the minimum diagnostic radiation dose level for the detection of high-resolution (HR) lung structures, pulmonary nodules (PNs), and infectious diseases (IDs).

MATERIALS AND METHODS

A preclinical chest computed tomography (CT) trial was performed with a human cadaver without known lung disease with incremental radiation dose using tin filter-based spectral shaping protocols. A subset of protocols for full diagnostic evaluation of HR, PN, and ID structures was translated to clinical routine. Also, a minimum diagnostic radiation dose protocol was defined (MIN). These protocols were prospectively applied over 5 months in the clinical routine under consideration of the individual clinical indication. We compared radiation dose parameters, objective and subjective image quality (IQ).

RESULTS

The HR protocol was performed in 38 patients (43%), PN in 21 patients (24%), ID in 20 patients (23%), and MIN in 9 patients (10%). Radiation dose differed significantly among HR, PN, and ID (5.4, 1.2, and 0.6 mGy, respectively; P < 0.001). Differences between ID and MIN (0.2 mGy) were not significant (P = 0.262). Dose-normalized contrast-to-noise ratio was comparable among all groups (P = 0.087). Overall IQ was perfect for the HR protocol (median, 5.0) and decreased for PN (4.5), ID-CT (4.3), and MIN-CT (2.5). The delineation of disease-specific findings was high in all dedicated protocols (HR, 5.0; PN, 5.0; ID, 4.5). The MIN protocol had borderline IQ for PN and ID lesions but was insufficient for HR structures. The dose reductions were 78% (PN), 89% (ID), and 97% (MIN) compared with the HR protocols.

CONCLUSIONS

Personalized chest CT tailored to the clinical indications leads to substantial dose reduction without reducing interpretability. More than 50% of patients can benefit from such individual adaptation in a clinical routine setting. Personalized radiation dose adjustments with validated diagnostic IQ are especially preferable for evaluating ID and PN lesions.

Community-based Lung Cancer Screening Results in Relation to Patient and Radiologist Characteristics: The PROSPR Consortium

Rationale: Lung-RADS classification was developed to standardize reporting and management of lung cancer screening using low-dose computed tomographic (LDCT) imaging. Although variation in Lung-RADS distribution between healthcare systems has been reported, it is unclear if this is explained by patient characteristics, radiologist experience with lung cancer screening, or other factors. Objectives: Our objective was to determine if patient or radiologist factors are associated with Lung-RADS score. Methods: In the Population-based Research to Optimize the Screening Process (PROSPR) Lung consortium, we conducted a study of patients who received their first screening LDCT imaging at one of the five healthcare systems in the PROSPR Lung Research Center from May 1, 2014, through December 31, 2017. Data on LDCT scans, patient factors, and radiologist characteristics were obtained via electronic health records. LDCT scan findings were categorized using Lung-RADS (negative [1], benign [2], probably benign [3], or suspicious [4]). We used generalized estimating equations with a multinomial distribution to compare the odds of Lung-RADS 3, and separately Lung-RADS 4, versus Lung-RADS 1 or 2 and estimated adjusted odds ratios (ORs) and 95% confidence intervals (CIs) for the associations between Lung-RADS assignment and patient and radiologist characteristics. Results: Analyses included 8,556 patients; 24% were assigned Lung-RADS 1, 60% Lung-RADS 2, 10% Lung-RADS 3, and 5% Lung-RADS 4. Age was positively associated with Lung-RADS 3 (OR, 1.02; 95% CI, 1.01-1.03) and 4 (OR, 1.03; 95% CI, 1.01-1.05); chronic obstructive pulmonary disease (COPD) was positively associated with Lung-RADS 4 (OR, 1.78; 95% CI, 1.45-2.20); obesity was inversely associated with Lung-RADS 3 (OR, 0.70; 95% CI, 0.58-0.84) and 4 (OR, 0.58; 95% CI, 0.45-0.75). There was no association between sex, race, ethnicity, education, or smoking status and Lung-RADS assignment. Radiologist volume of interpreting screening LDCT scans, years in practice, and thoracic specialty were also not associated with Lung-RADS assignment. Conclusions: Healthcare systems that are comprised of patients with an older age distribution or higher levels of COPD will have a greater proportion of screening LDCT scans with Lung-RADS 3 or 4 findings and should plan for additional resources to support appropriate and timely management of noted positive findings.

Lung cancer screening by volume computed tomography: thriving to high performance

Low-dose volume CT screening for lung cancer leads to a significant decrease in lung-cancer-related mortality. However, optimisation of the post-screening protocol will be crucial for optimal healthcare.https://bit.ly/2ZkJPpH

Low-dose CT with tin filter combined with iterative metal artefact reduction for guiding lung biopsy

Background

Computed tomography (CT) is currently the imaging modality of choice for guiding pulmonary percutaneous procedures. The use of a tin filter allows low-energy photons to be absorbed which contribute little to image quality but increases the radiation dose that a patient receives. Iterative metal artefact reduction (iMAR) was developed to diminish metal artefacts. This study investigated the impact of using tin filtration combined with an iMAR algorithm on dose reduction and image quality in CT-guided lung biopsy.

Methods

Ninety-nine consecutive patients undergoing CT-guided lung biopsy were randomly assigned to routine-dose CT protocols (groups A and B; without and with iMAR, respectively) or tin filter CT protocols (groups C and D; without or with iMAR, respectively). Subjective image quality was analysed using a 5-point Likert scale. Objective image quality was assessed, and the noise, contrast-to-noise ratio, and figure of merit were compared among the four groups. Metal artefacts were quantified using CT number reduction and metal diameter blurring. The radiation doses, diagnostic performance, and complication rates were also estimated.

Results

The subjective image quality of the two scan types was compared. Images with iMAR reconstruction were superior to those without iMAR reconstruction (group A: 3.49±0.65 vs. group B: 4.63±0.57; P<0.001, and group C: 3.88±0.66 vs. group D: 4.82±0.39; P<0.001). Images taken with a tin filter were found to have a significantly higher figure-of-merit than those taken without a tin filter (group A: 14,041±7,230 vs. group C: 21,866±10,656; P=0.001, and group B: 13,836±6,849 vs. group D: 21,639±9,964; P=0.001). In terms of metal artefact reduction, tin filtration combined with iMAR showed the lowest CT number reduction (116.62±103.48 HU) and metal diameter blurring (0.85±0.30) among the protocols. The effective radiation dose in the tin filter groups was 73.2% lower than that in the routine-dose groups. The complication rate and diagnostic performance (sensitivity, specificity, and overall accuracy) did not differ significantly between the tin filter and routine-dose groups (all P>0.05).

Conclusions

Tin filtration combined with an iMAR algorithm may reduce the radiation dose compared to the routine-dose CT protocol, while maintaining comparable diagnostic accuracy and image quality and producing fewer metal artefacts.

Pneumonia Detection in Chest X-Ray Dose-Equivalent CT: Impact of Dose Reduction on Detectability by Artificial Intelligence

RATIONALE AND OBJECTIVES

There has been a significant increase of immunocompromised patients in recent years due to new treatment modalities for previously fatal diseases. This comes at the cost of an elevated risk for infectious diseases, most notably pathogens affecting the respiratory tract. Because early diagnosis and treatment of pneumonia can help reducing morbidity and mortality, we assessed the performance of a deep neural network in the detection of pulmonary infection in chest X-ray dose-equivalent computed tomography (CT).

MATERIALS AND METHODS

The 100 patients included in this retrospective study were referred to our department for suspicion of pulmonary infection and/or follow-up of known pulmonary nodules. Every patient was scanned with a standard dose (1.43 ± 0.54 mSv) and a 20 times dose-reduced (0.07 ± 0.03 mSv) CT protocol. We trained a deep neural network to perform binary classification (pulmonary consolidation present or not) and assessed diagnostic performance on both standard dose and reduced dose CT images.

RESULTS

The areas under the curve of the deep learning algorithm for the standard dose CT was 0.923 (confidence interval [CI] 95%: 0.905-0.941) and significantly higher than the areas under the curve (0.881, CI 95%: 0.859-0.903) of the reduced dose CT (p = 0.001). Sensitivity and specificity of the standard dose CT was 82.9% and 93.8%, and of the reduced dose CT 71.0% and 93.3%.

CONCLUSION

Pneumonia detection with X-ray dose-equivalent CT using artificial intelligence is feasible and may contribute to a more robust and reproducible diagnostic performance. Dose reduction lowered the performance of the deep neural network, which calls for optimization and adaption of CT protocols when using AI algorithms at reduced doses.

Latest CT technologies in lung cancer screening: protocols and radiation dose reduction

The aim of this review is to provide clinicians and technicians with an overview of the development of CT protocols in lung cancer screening. CT protocols have evolved from pre-fixed settings in early lung cancer screening studies starting in 2004 towards automatic optimized settings in current international guidelines. The acquisition protocols of large lung cancer screening studies and guidelines are summarized. Radiation dose may vary considerably between CT protocols, but has reduced gradually over the years. Ultra-low dose acquisition can be achieved by applying latest dose reduction techniques. The use of low tube current or tin-filter in combination with iterative reconstruction allow to reduce the radiation dose to a submilliSievert level. However, one should be cautious in reducing the radiation dose to ultra-low dose settings since performed studies lacked generalizability. Continuous efforts are made by international radiology organizations to streamline the CT data acquisition and image quality assurance and to keep track of new developments in CT lung cancer screening. Examples like computer-aided diagnosis and radiomic feature extraction are discussed and current limitations are outlined. Deep learning-based solutions in post-processing of CT images are provided. Finally, future perspectives and recommendations are provided for lung cancer screening CT protocols.

Observer Performance for Detection of Pulmonary Nodules at Chest CT over a Large Range of Radiation Dose Levels

Background There is a wide variation in radiation dose levels that can be used with chest CT in order to detect indeterminate pulmonary nodules. Purpose To compare the performance of lower-radiation-dose chest CT with that of routine dose in the detection of indeterminate pulmonary nodules 5 mm or greater. Materials and Methods In this retrospective study, CT projection data from 83 routine-dose chest CT examinations performed in 83 patients (120 kV, 70 quality reference mAs [QRM]) were collected between November 2013 and April 2014. Reference indeterminate pulmonary nodules were identified by two nonreader thoracic radiologists. By using validated noise insertion, five lower-dose data sets were reconstructed with filtered back projection (FBP) or iterative reconstruction (IR; 30 QRM with FBP, 10 QRM with IR, 5 QRM with FBP, 5 QRM with IR, and 2.5 QRM with IR). Three thoracic radiologists circled pulmonary nodules, rating confidence that the nodule was a 5-mm-or-greater indeterminate pulmonary nodule, and graded image quality. Analysis was performed on a per-nodule basis by using jackknife alternative free-response receiver operating characteristic figure of merit (FOM) and noninferiority limit of -0.10. Results There were 66 indeterminate pulmonary nodules (mean size, 8.6 mm ± 3.4 [standard deviation]; 21 part-solid nodules) in 42 patients (mean age, 51 years ± 17; 21 men and 21 women). Compared with the FOM for routine-dose CT (size-specific dose estimate, 6.5 mGy ± 1.8; FOM, 0.86 [95% confidence interval: 0.80, 0.91]), FOM was noninferior for all lower-dose configurations except for 2.5 QRM with IR. The sensitivity for subsolid nodules at 70 QRM was 60% (range, 48%-72%) and was significantly worse at a dose of 5 QRM and lower, whether or not IR was used (P < .05). Diagnostic image quality decreased with decreasing dose (P < .001) and was better with IR at 5 QRM (P < .05). Conclusion CT images reconstructed at dose levels down to 10 quality reference mAs (size-specific dose estimate, 0.9 mGy) had noninferior performance compared with routine dose in depicting pulmonary nodules. Iterative reconstruction improved subjective image quality but not performance at low dose levels. © RSNA, 2020 Online supplemental material is available for this article. See also the editorial by White and Kazerooni in this issue.