Utilizing Deep Learning and Computed Tomography to Determine Pulmonary Nodule Activity in Patients With Nontuberculous Mycobacterial-Lung Disease.

PURPOSE: To develop and evaluate a deep convolutional neural network (DCNN) model for the classification of acute and chronic lung nodules from nontuberculous mycobacterial-lung disease (NTM-LD) on computed tomography (CT). MATERIALS AND METHODS: We collected a data set of 650 nodules (316 acute and 334 chronic) from the CT scans of 110 patients with NTM-LD. The data set was divided into training, validation, and test sets in a ratio of 4:1:1. Bounding boxes were used to crop the 2D CT images down to the area of interest. A DCNN model was built using 11 convolutional layers and trained on these images. The performance of the model was evaluated on the hold-out test set and compared with that of 3 radiologists who independently reviewed the images. RESULTS: The DCNN model achieved an area under the receiver operating characteristic curve of 0.806 for differentiating acute and chronic NTM-LD nodules, corresponding to sensitivity, specificity, and accuracy of 76%, 68%, and 72%, respectively. The performance of the model was comparable to that of the 3 radiologists, who had area under the receiver operating characteristic curve, sensitivity, specificity, and accuracy of 0.693 to 0.771, 61% to 82%, 59% to 73%, and 60% to 73%, respectively. CONCLUSIONS: This study demonstrated the feasibility of using a DCNN model for the classification of the activity of NTM-LD nodules on chest CT. The model performance was comparable to that of radiologists. This approach can potentially and efficiently improve the diagnosis and management of NTM-LD.

Robust deep learning from incomplete annotation for accurate lung nodule detection.

Deep learning plays a significant role in the detection of pulmonary nodules in low-dose computed tomography (LDCT) scans, contributing to the diagnosis and treatment of lung cancer. Nevertheless, its effectiveness often relies on the availability of extensive, meticulously annotated dataset. In this paper, we explore the utilization of an incompletely annotated dataset for pulmonary nodules detection and introduce the FULFIL (Forecasting Uncompleted Labels For Inexpensive Lung nodule detection) algorithm as an innovative approach. By instructing annotators to label only the nodules they are most confident about, without requiring complete coverage, we can substantially reduce annotation costs. Nevertheless, this approach results in an incompletely annotated dataset, which presents challenges when training deep learning models. Within the FULFIL algorithm, we employ Graph Convolution Network (GCN) to discover the relationships between annotated and unannotated nodules for self-adaptively completing the annotation. Meanwhile, a teacher-student framework is employed for self-adaptive learning using the completed annotation dataset. Furthermore, we have designed a Dual-Views loss to leverage different data perspectives, aiding the model in acquiring robust features and enhancing generalization. We carried out experiments using the LUng Nodule Analysis (LUNA) dataset, achieving a sensitivity of 0.574 at a False positives per scan (FPs/scan) of 0.125 with only 10% instance-level annotations for nodules. This performance outperformed comparative methods by 7.00%. Experimental comparisons were conducted to evaluate the performance of our model and human experts on test dataset. The results demonstrate that our model can achieve a comparable level of performance to that of human experts. The comprehensive experimental results demonstrate that FULFIL can effectively leverage an incomplete pulmonary nodule dataset to develop a robust deep learning model, making it a promising tool for assisting in lung nodule detection.

[The application of setting tube voltage and iodine delivery rate in weight-grouped for reducing the radiation and contrast medium dose in coronary CT angiography].

Objective: To evaluate the application value of reducing tube voltage and iodine delivery rate according to body weight in coronary CT angiography (CCTA). Methods: A prospective randomized controlled study. A total of 297 subjects, 172 males and 125 females, aged [M (Q1, Q3)]60.0 (50.0, 68.0) years, who underwent CCTA examination in Peking University Third Hospital due to clinically suspected coronary heart disease from May to December 2022 were included. According to the odd or even visit dates, the subjects were randomly divided into test group (n=156) and control group (n=141). The subjects in both groups were divided into four sub-groups according to body weight: 50-59 group, 60-69 kg group, 70-79 kg group and 80-89 kg group, respectively. The CCTA images were reconstructed with hybrid iterative algorithm(KARL 3D) with levels of 6 and 8, respectively. 100 kVp and iodine flow rate 1.1, 1.3, 1.4 and 1.5 gI/s recommended by the domestic CCTA application guidelines were used in the control group, while the tube voltage and iodine flow rate were reduced in the test group based on the guidelines and body weight:70 kVp and 0.8 g I/s in 50~59 kg group, 80 kVp and 1.0 gI/s in 60~69 kg group, 80 kVp and 1.1 gI/s in70~79 kg group, and 100 kVp and 1.5 gI/s in 80~89 kg group, respectively. The CT values and standard deviation (SD) of aortic root, proximal left anterior descending branch (LAD) and distal right coronary artery (RCA) luminal CCTA, the signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) of coronary artery CT images, subjective coronary scores and effective radiation dose (ED) were compared between the both groups. One-way ANOVA or Wilcoxon test was used to analyze the differences of above indicators between the groups to evaluate the application value of low voltage and low iodine flow rate based on weight in coronary CCTA. Results: CT values of aortic root, LAD proximal CT values and SD values of aortic root [411.4 (377.2, 439.8) HU, (366.3±42.9) HU, 26.5±2.3] in the test group were all higher than those in the control group [379.00 (335.2, 415.9) HU, (355.0±46.9) HU and 24.8±2.3]. The differences were statistically significant (all P<0.05), and the other parameters were not statistically significant (all P>0.05). The total subjective image quality score in test group were superior to those in the control group (all P<0.05). The total ED and contrast agent dosage [2.07 (1.52, 3.28) mSv and (38.28±9.68) ml] in CCTA examination in the test group were lower than those in the control group [3.30(2.32, 4.44) mSv and (45.31±5.63) ml], and the differences were statistically significant (all P<0.05). The dosage of ED and contrast agent in the test group was decreased by 37.3% and 15.5%, respectively. Conclusion: Combined with KARL 3D,it is feasible to reduce contrast medium and ED by setting the tube voltage and iodine flow rate of CCTA according to the weight of the subject, which can further reduce the radiation dose and contrast agent dosage of CCTA.

Effect of X-ray tube on image quality and pancreatic ductal adenocarcinoma conspicuity in pancreatic protocol dual-energy CT.

AIM: To compare the radiation dose, image quality, and conspicuity of pancreatic ductal adenocarcinoma (PDAC) in pancreatic protocol dual-energy computed tomography (CT) between two X-ray tubes mounted in the same CT machine. MATERIAL AND METHODS: This retrospective study comprised 80 patients (median age, 73 years; 45 men) who underwent pancreatic protocol dual-energy CT from January 2019 to March 2022 using either old (Group A, n=41) or new (Group B, n=39) X-ray tubes mounted in the same CT machine. The imaging parameters were completely matched between the two groups, and CT data were reconstructed at 70 and 40 keV. The CT dose-index volume (CTDIvol); CT attenuation of the abdominal aorta, pancreas, and PDAC; background noise; and qualitative scores for the image noise, overall image quality, and PDAC conspicuity were compared between the two groups. RESULTS: The CTDIvol was lower in Group B than Group A (7.9 versus 9.2 mGy; p<0.001). The CT attenuation of all anatomical structures at 70 and 40 keV was comparable between the two groups (p=0.06-0.78). The background noise was lower in Group B than Group A (12 versus 14 HU at 70 keV, p=0.046; and 26 versus 30 HU at 40 keV, p<0.001). Qualitative scores for image noise and overall image quality at 70 and 40 keV and PDAC conspicuity at 40 keV were higher in Group B than Group A (p<0.001-0.045). CONCLUSION: The latest X-ray tube could reduce the radiation dose and improve image quality in pancreatic protocol dual-energy CT.

Optimizing dual-energy CT technique for iodine-based contrast-to-noise ratio, a theoretical study.

BACKGROUND: Dual-energy CT (DECT) systems provide valuable material-specific information by simultaneously acquiring two spectral measurements, resulting in superior image quality and contrast-to-noise ratio (CNR) while reducing radiation exposure and contrast agent usage. The selection of DECT scan parameters, including x-ray tube settings and fluence, is critical for the stability of the reconstruction process and hence the overall image quality. PURPOSE: The goal of this study is to propose a systematic theoretical method for determining the optimal DECT parameters for minimal noise and maximum CNR in virtual monochromatic images (VMIs) for fixed subject size and total radiation dose. METHODS: The noise propagation in the process of projection based material estimation from DECT measurements is analyzed. The main components of the study are the mean pixel variances for the sinogram and monochromatic image and the CNR, which were shown to depend on the Jacobian matrix of the sinograms-to-DECT measurements map. Analytic estimates for the mean sinogram and monochromatic image pixel variances and the CNR as functions of tube potentials, fluence, and VMI energy are derived, and then used in a virtual phantom experiment as an objective function for optimizing the tube settings and VMI energy to minimize the image noise and maximize the CNR. RESULTS: It was shown that DECT measurements corresponding to kV settings that maximize the square of Jacobian determinant values over a domain of interest lead to improved stability of basis material reconstructions. Instances of non-uniqueness in DECT were addressed, focusing on scenarios where the Jacobian determinant becomes zero within the domain of interest despite significant spectral separation. The presence of non-uniqueness can lead to singular solutions during the inversion of sinograms-to-DECT measurements, underscoring the importance of considering uniqueness properties in parameter selection. Additionally, the optimal VMI energy and tube potentials for maximal CNR was determined. When the x-ray beam filter material was fixed at 2 mm of aluminum and the photon fluence for low and high kV scans were considered equal, the tube potential pair of 60/120 kV led to the maximal iodine CNR in the VMI at 53 keV. CONCLUSIONS: Optimizing DECT scan parameters to maximize the CNR can be done in a systematic way. Also, choosing the parameters that maximize the Jacobian determinant over the set of expected line integrals leads to more stable reconstructions due to the reduced amplification of the measurement noise. Since the values of the Jacobian determinant depend strongly on the imaging task, careful consideration of all of the relevant factors is needed when implementing the proposed framework.

Development of a modified 3D region proposal network for lung nodule detection in computed tomography scans: a secondary analysis of lung nodule datasets.

BACKGROUND: Low-dose computed tomography (LDCT) has been shown useful in early lung cancer detection. This study aimed to develop a novel deep learning model for detecting pulmonary nodules on chest LDCT images. METHODS: In this secondary analysis, three lung nodule datasets, including Lung Nodule Analysis 2016 (LUNA16), Lung Nodule Received Operation (LNOP), and Lung Nodule in Health Examination (LNHE), were used to train and test deep learning models. The 3D region proposal network (RPN) was modified via a series of pruning experiments for better predictive performance. The performance of each modified deep leaning model was evaluated based on sensitivity and competition performance metric (CPM). Furthermore, the performance of the modified 3D RPN trained on three datasets was evaluated by 10-fold cross validation. Temporal validation was conducted to assess the reliability of the modified 3D RPN for detecting lung nodules. RESULTS: The results of pruning experiments indicated that the modified 3D RPN composed of the Cross Stage Partial Network (CSPNet) approach to Residual Network (ResNet) Xt (CSP-ResNeXt) module, feature pyramid network (FPN), nearest anchor method, and post-processing masking, had the optimal predictive performance with a CPM of 92.2%. The modified 3D RPN trained on the LUNA16 dataset had the highest CPM (90.1%), followed by the LNOP dataset (CPM: 74.1%) and the LNHE dataset (CPM: 70.2%). When the modified 3D RPN trained and tested on the same datasets, the sensitivities were 94.6%, 84.8%, and 79.7% for LUNA16, LNOP, and LNHE, respectively. The temporal validation analysis revealed that the modified 3D RPN tested on LNOP test set achieved a CPM of 71.6% and a sensitivity of 85.7%, and the modified 3D RPN tested on LNHE test set had a CPM of 71.7% and a sensitivity of 83.5%. CONCLUSION: A modified 3D RPN for detecting lung nodules on LDCT scans was designed and validated, which may serve as a computer-aided diagnosis system to facilitate lung nodule detection and lung cancer diagnosis.

A modified 3D RPN for detecting lung nodules on CT images that exhibited greater sensitivity and CPM than did several previously reported CAD detection models was established.

Development of automatic generation system for lung nodule finding descriptions.

Worldwide, lung cancer is the leading cause of cancer-related deaths. To manage lung nodules, radiologists observe computed tomography images, review various imaging findings, and record these in radiology reports. The report contents should be of high quality and uniform regardless of the radiologist. Here, we propose an artificial intelligence system that automatically generates descriptions related to lung nodules in computed tomography images. Our system consists of an image recognition method for extracting contents-namely, bronchopulmonary segments and nodule characteristics from images-and a natural language processing method to generate fluent descriptions. To verify our system's clinical usefulness, we conducted an experiment in which two radiologists created nodule descriptions of findings using our system. Through our system, the similarity of the described contents between the two radiologists (p = 0.001) and the comprehensiveness of the contents (p = 0.025) improved, while the accuracy did not significantly deteriorate (p = 0.484).

Application of Dual-Layer Spectral-Detector Computed Tomography Angiography in Identifying Symptomatic Carotid Atherosclerosis: A Prospective Observational Study.

BACKGROUND: Dual-layer spectral-detector dual-energy computed tomography angiography (DLCTA) can distinguish components of carotid plaques. Data on identifying symptomatic carotid plaques in patients using DLCTA are not available. METHODS AND RESULTS: In this prospective observational study, patients with carotid plaques were enrolled and received DLCTA. The attenuation for both polyenergetic image and virtual monoenergetic images (40, 70, 100, and 140 keV), as well as Z-effective value, were recorded in the noncalcified regions of plaques. Logistic regression models were used to assess the association between attenuations of DLCTA and the presence of symptomatic carotid plaques. In total, 100 participants (mean±SD age, 64.37±8.31 years; 82.0% were men) were included, and 36% of the cases were identified with the symptomatic group. DLCTA parameters were different between 2 groups (symptomatic versus asymptomatic: computed tomography [CT] 40 keV, 152.63 [interquartile range (IQR), 70.22-259.78] versus 256.78 [IQR, 150.34-408.13]; CT 70 keV, 81.28 [IQR, 50.13-119.33] versus 108.87 [IQR, 77.01-165.88]; slope40-140 keV, 0.91 [IQR, 0.35-1.87] versus 1.92 [IQR, 0.96-3.00]; Z-effective value, 7.92 [IQR, 7.53-8.46] versus 8.41 [IQR, 7.94-8.92]), whereas no difference was found in conventional polyenergetic images. The risk of symptomatic plaque was lower in the highest tertiles of attenuations in CT 40 keV (adjusted odds ratio [OR], 0.243 [95% CI, 0.078-0.754]), CT 70 keV (adjusted OR, 0.313 [95% CI, 0.104-0.940]), Z-effective values (adjusted OR, 0.138 [95% CI, 0.039-0.490]), and slope40-140 keV (adjusted OR, 0.157 [95% CI, 0.046-0.539]), with all P values and P trends <0.05. The areas under the curve for CT 40 keV, CT 70 keV, slope 40 to 140 keV, and Z-effective values were 0.64, 0.61, 0.64, and 0.63, respectively. CONCLUSIONS: Parameters of DLCTA might help assist in distinguishing symptomatic carotid plaques. Further studies with a larger sample size may address the overlap and improve the diagnostic accuracy.

Positive Predictive Values of Abnormality Scores From a Commercial Artificial Intelligence-Based Computer-Aided Diagnosis for Mammography.

OBJECTIVE: Artificial intelligence-based computer-aided diagnosis (AI-CAD) is increasingly used in mammography. While the continuous scores of AI-CAD have been related to malignancy risk, the understanding of how to interpret and apply these scores remains limited. We investigated the positive predictive values (PPVs) of the abnormality scores generated by a deep learning-based commercial AI-CAD system and analyzed them in relation to clinical and radiological findings. MATERIALS AND METHODS: From March 2020 to May 2022, 656 breasts from 599 women (mean age 52.6 ± 11.5 years, including 0.6% [4/599] high-risk women) who underwent mammography and received positive AI-CAD results (Lunit Insight MMG, abnormality score ≥ 10) were retrospectively included in this study. Univariable and multivariable analyses were performed to evaluate the associations between the AI-CAD abnormality scores and clinical and radiological factors. The breasts were subdivided according to the abnormality scores into groups 1 (10-49), 2 (50-69), 3 (70-89), and 4 (90-100) using the optimal binning method. The PPVs were calculated for all breasts and subgroups. RESULTS: Diagnostic indications and positive imaging findings by radiologists were associated with higher abnormality scores in the multivariable regression analysis. The overall PPV of AI-CAD was 32.5% (213/656) for all breasts, including 213 breast cancers, 129 breasts with benign biopsy results, and 314 breasts with benign outcomes in the follow-up or diagnostic studies. In the screening mammography subgroup, the PPVs were 18.6% (58/312) overall and 5.1% (12/235), 29.0% (9/31), 57.9% (11/19), and 96.3% (26/27) for score groups 1, 2, 3, and 4, respectively. The PPVs were significantly higher in women with diagnostic indications (45.1% [155/344]), palpability (51.9% [149/287]), fatty breasts (61.2% [60/98]), and certain imaging findings (masses with or without calcifications and distortion). CONCLUSION: PPV increased with increasing AI-CAD abnormality scores. The PPVs of AI-CAD satisfied the acceptable PPV range according to Breast Imaging-Reporting and Data System for screening mammography and were higher for diagnostic mammography.

Dark-Blood Computed Tomography Angiography Combined With Deep Learning Reconstruction for Cervical Artery Wall Imaging in Takayasu Arteritis.

OBJECTIVE: To evaluate the image quality of novel dark-blood computed tomography angiography (CTA) imaging combined with deep learning reconstruction (DLR) compared to delayed-phase CTA images with hybrid iterative reconstruction (HIR), to visualize the cervical artery wall in patients with Takayasu arteritis (TAK). MATERIALS AND METHODS: This prospective study continuously recruited 53 patients with TAK (mean age: 33.8 ± 10.2 years; 49 females) between January and July 2022 who underwent head-neck CTA scans. The arterial- and delayed-phase images were reconstructed using HIR and DLR. Subtracted images of the arterial-phase from the delayed-phase were then added to the original delayed-phase using a denoising filter to generate the final-dark-blood images. Qualitative image quality scores and quantitative parameters were obtained and compared among the three groups of images: Delayed-HIR, Dark-blood-HIR, and Dark-blood-DLR. RESULTS: Compared to Delayed-HIR, Dark-blood-HIR images demonstrated higher qualitative scores in terms of vascular wall visualization and diagnostic confidence index (all P < 0.001). These qualitative scores further improved after applying DLR (Dark-blood-DLR compared to Dark-blood-HIR, all P < 0.001). Dark-blood DLR also showed higher scores for overall image noise than Dark-blood-HIR (P < 0.001). In the quantitative analysis, the contrast-to-noise ratio (CNR) values between the vessel wall and lumen for the bilateral common carotid arteries and brachiocephalic trunk were significantly higher on Dark-blood-HIR images than on Delayed-HIR images (all P < 0.05). The CNR values were significantly higher for Dark-blood-DLR than for Dark-blood-HIR in all cervical arteries (all P < 0.001). CONCLUSION: Compared with Delayed-HIR CTA, the dark-blood method combined with DLR improved CTA image quality and enhanced visualization of the cervical artery wall in patients with TAK.

Improving image quality of triple-low-protocol renal artery CT angiography with deep-learning image reconstruction: a comparative study with standard-dose single-energy and dual-energy CT with adaptive statistical iterative reconstruction.

AIM: To investigate the improvement in image quality of triple-low-protocol (low radiation, low contrast medium dose, low injection speed) renal artery computed tomography (CT) angiography (RACTA) using deep-learning image reconstruction (DLIR), in comparison with standard-dose single- and dual-energy CT (DECT) using adaptive statistical iterative reconstruction-Veo (ASIR-V) algorithm. MATERIALS AND METHODS: Ninety patients for RACTA were divided into different groups: standard-dose single-energy CT (S group) using ASIR-V at 60% strength (60%ASIR-V), DECT (DE group) with 60%ASIR-V including virtual monochromatic images at 40 keV (DE40 group) and 70 keV (DE70 group), and the triple-low protocol single-energy CT (L group) with DLIR at high level (DLIR-H). The effective dose (ED), contrast medium dose, injection speed, standard deviation (SD), signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) of abdominal aorta (AA), and left/right renal artery (LRA, RRA), and subjective scores were compared among the different groups. RESULTS: The L group significantly reduced ED by 37.6% and 31.2%, contrast medium dose by 33.9% and 30.5%, and injection speed by 30% and 30%, respectively, compared to the S and DE groups. The L group had the lowest SD values for all arteries compared to the other groups (p<0.001). The SNR of RRA and LRA in the L group, and the CNR of all arteries in the DE40 group had highest value compared to others (p<0.05). The L group had the best comprehensive score with good consistency (p<0.05). CONCLUSIONS: The triple-low protocol RACTA with DLIR-H significantly reduces the ED, contrast medium doses, and injection speed, while providing good comprehensive image quality.

A CT deep learning reconstruction algorithm: Image quality evaluation for brain protocol at decreasing dose indexes in comparison with FBP and statistical iterative reconstruction algorithms.

PURPOSE: To characterise the impact of Precise Image (PI) deep learning reconstruction algorithm on image quality, compared to filtered back-projection (FBP) and iDose4 iterative reconstruction for brain computed tomography (CT) phantom images. METHODS: Catphan-600 phantom was acquired with an Incisive CT scanner using a dedicated brain protocol, at six different dose levels (volume computed tomography dose index (CTDIvol): 7/14/29/49/56/67 mGy). Images were reconstructed using FBP, levels 2/5 of iDose4, and PI algorithm (Sharper/Sharp/Standard/Smooth/Smoother). Image quality was assessed by evaluating CT numbers, image histograms, noise, image non-uniformity (NU), noise power spectrum, target transfer function, and detectability index. RESULTS: The five PI levels did not significantly affect the mean CT number. For a given CTDIvol using Sharper-to-Smoother levels, the spatial resolution for all the investigated materials and the detectability index increased while the noise magnitude decreased, slightly affecting noise texture. For a fixed PI level increasing the CTDIvol the detectability index increased, the noise magnitude decreased. From 29 mGy, NU values converged within 1 Hounsfield Unit from each other without a substantial improvement at higher CTDIvol values. CONCLUSIONS: The improved performances of intermediate PI levels in brain protocols compared to conventional algorithms seem to suggest a potential reduction of CTDIvol.

Imaging of intracranial hemorrhage in photon counting computed tomography using virtual monoenergetic images.

PURPOSE: To determine the optimal virtual monoenergetic image (VMI) for detecting and assessing intracranial hemorrhage in unenhanced photon counting CT of the head based on the evaluation of quantitative and qualitative image quality parameters. METHODS: Sixty-three patients with acute intracranial hemorrhage and unenhanced CT of the head were retrospectively included. In these patients, 35 intraparenchymal, 39 intraventricular, 30 subarachnoidal, and 43 subdural hemorrhages were selected. VMIs were reconstructed using all available monoenergetic reconstruction levels (40-190 keV). Multiple regions of interest measurements were used for evaluation of the overall image quality, and signal, noise, signal-to-noise-ratio (SNR), and contrast-to-noise-ratio (CNR) of intracranial hemorrhage. Based on the results of the quantitative analysis, specific VMIs were rated by five radiologists on a 5-point Likert scale. RESULTS: Signal, noise, SNR, and CNR differed significantly between different VMIs (p < 0.001). Maximum CNR for intracranial hemorrhage was reached in VMI with keV levels > 120 keV (intraparenchymal 143 keV, intraventricular 164 keV, subarachnoidal 124 keV, and subdural hemorrhage 133 keV). In reading, no relevant superiority in the detection of hemorrhage could be demonstrated using VMIs above 66 keV. CONCLUSION: For the detection of hemorrhage in unenhanced CT of the head, the quantitative analysis of the present study on photon counting CT is generally consistent with the findings from dual-energy CT, suggesting keV levels just above 120 keV and higher depending on the location of the hemorrhage. However, on the basis of the qualitative analyses, no reliable statement can yet be made as to whether an additional VMI with higher keV is truly beneficial in everyday clinical practice.

Clinical implementation of cinematic rendering.

Cinematic rendering is a recently developed photorealistic display technique for standard volumetric data sets. It has broad-reaching applications in cardiovascular, musculoskeletal, abdominopelvic, and thoracic imaging. It has been used for surgical planning and has emerging use in educational settings. We review the logistics of performing this post-processing step and its integration into existing workflow.

S-Net: an S-shaped network for nodule detection in 3D CT images.

Objective. Accurate and automatic detection of pulmonary nodules is critical for early lung cancer diagnosis, and promising progress has been achieved in developing effective deep models for nodule detection. However, most existing nodule detection methods merely focus on integrating elaborately designed feature extraction modules into the backbone of the detection network to extract rich nodule features while ignore disadvantages of the structure of detection network itself. This study aims to address these disadvantages and develop a deep learning-based algorithm for pulmonary nodule detection to improve the accuracy of early lung cancer diagnosis.Approach. In this paper, an S-shaped network called S-Net is developed with the U-shaped network as backbone, where an information fusion branch is used to propagate lower-level details and positional information critical for nodule detection to higher-level feature maps, head shared scale adaptive detection strategy is utilized to capture information from different scales for better detecting nodules with different shapes and sizes and the feature decoupling detection head is used to allow the classification and regression branches to focus on the information required for their respective tasks. A hybrid loss function is utilized to fully exploit the interplay between the classification and regression branches.Main results. The proposed S-Net network with ResSENet and other three U-shaped backbones from SANet, OSAF-YOLOv3 and MSANet (R+SC+ECA) models achieve average CPM scores of 0.914, 0.915, 0.917 and 0.923 on the LUNA16 dataset, which are significantly higher than those achieved with other existing state-of-the-art models.Significance. The experimental results demonstrate that our proposed method effectively improves nodule detection performance, which implies potential applications of the proposed method in clinical practice.

Low-dose versus conventional CT urography using dual-source CT with different time-current product values and the same tube voltage: image quality and diagnostic performance in various diagnoses.

OBJECTIVES: To compare the image quality and diagnostic performance of low-dose CT urography to that of concurrently acquired conventional CT using dual-source CT. METHODS: This retrospective study included 357 consecutive CT urograms performed by third-generation dual-source CT in a single institution between April 2020 and August 2021. Two-phase CT images (unenhanced phase, excretory phase with split bolus) were obtained with two different tube current-time products (280 mAs for the conventional-dose protocol and 70 mAs for the low-dose protocol) and the same tube voltage (90 kVp) for the two X-ray tubes. Iterative reconstruction was applied for both protocols. Two radiologists independently performed quantitative and qualitative image quality analysis and made diagnoses. The correlation between the noise level or the effective radiation dose and the patients' body weight was evaluated. RESULTS: Significantly higher noise levels resulting in a significantly lower liver signal-to-noise ratio and contrast-to-noise ratio were noted in low-dose images compared to conventional images (P < .001). Qualitative analysis by both radiologists showed significantly lower image quality in low-dose CT than in conventional CT images (P < .001). Patient's body weight was positively correlated with noise and effective radiation dose (P < .001). Diagnostic performance for various diseases, including urolithiasis, inflammation, and mass, was not different between the two protocols. CONCLUSIONS: Despite inferior image quality, low-dose CT urography with 70 mAs and 90 kVp and iterative reconstruction demonstrated diagnostic performance equivalent to that of conventional CT for identifying various diseases of the urinary tract. ADVANCES IN KNOWLEDGE: Low-dose CT (25% radiation dose) with low tube current demonstrated diagnostic performance comparable to that of conventional CT for a variety of urinary tract diseases.

Prospective evaluation of deep learning image reconstruction for Lung-RADS and automatic nodule volumetry on ultralow-dose chest CT.

PURPOSE: To prospectively evaluate whether Lung-RADS classification and volumetric nodule assessment were feasible with ultralow-dose (ULD) chest CT scans with deep learning image reconstruction (DLIR). METHODS: The institutional review board approved this prospective study. This study included 40 patients (mean age, 66±12 years; 21 women). Participants sequentially underwent LDCT and ULDCT (CTDIvol, 0.96±0.15 mGy and 0.12±0.01 mGy) scans reconstructed with the adaptive statistical iterative reconstruction-V 50% (ASIR-V50) and DLIR. CT image quality was compared subjectively and objectively. The pulmonary nodules were assessed visually by two readers using the Lung-RADS 1.1 and automatically using a computerized assisted tool. RESULTS: DLIR provided a significantly higher signal-to-noise ratio for LDCT and ULDCT images than ASIR-V50 (all P < .001). In general, DLIR showed superior subjective image quality for ULDCT images (P < .001) and comparable quality for LDCT images compared to ASIR-V50 (P = .01-1). The per-nodule sensitivities of observers for Lung-RADS category 3-4 nodules were 70.6-88.2% and 64.7-82.4% for DLIR-LDCT and DLIR-ULDCT images (P = 1) and categories were mostly concordant within observers. The per-nodule sensitivities of the computer-assisted detection for nodules ≥4 mm were 72.1% and 67.4% on DLIR-LDCT and ULDCT images (P = .50). The 95% limits of agreement for nodule volume differences between DLIR-LDCT and ULDCT images (-85.6 to 78.7 mm3) was similar to the within-scan nodule volume differences between DLIR- and ASIR-V50-LDCT images (-63.9 to 78.5 mm3), with volume differences smaller than 25% in 88.5% and 92.3% of nodules, respectively (P = .65). CONCLUSION: DLIR enabled comparable Lung-RADS and volumetric nodule assessments on ULDCT images to LDCT images.

Image quality in CT thorax: effect of altering reconstruction algorithm and tube load.

Non-linear properties of iterative reconstruction (IR) algorithms can alter image texture. We evaluated the effect of a model-based IR algorithm (advanced modelled iterative reconstruction; ADMIRE) and dose on computed tomography thorax image quality. Dual-source scanner data were acquired at 20, 45 and 65 reference mAs in 20 patients. Images reconstructed with filtered back projection (FBP) and ADMIRE Strengths 3-5 were assessed independently by six radiologists and analysed using an ordinal logistic regression model. For all image criteria studied, the effects of tube load 20 mAs and all ADMIRE strengths were significant (p < 0.001) when compared to reference categories 65 mAs and FBP. Increase in tube load from 45 to 65 mAs showed image quality improvement in three of six criteria. Replacing FBP with ADMIRE significantly improves perceived image quality for all criteria studied, potentially permitting a dose reduction of almost 70% without loss in image quality.

Multi-instance learning based lung nodule system for assessment of CT quality after small-field-of-view reconstruction.

Small-field-of-view reconstruction CT images (sFOV-CT) increase the pixel density across airway structures and reduce partial volume effects. Multi-instance learning (MIL) is proposed as a weakly supervised machine learning method, which can automatically assess the image quality. The aim of this study was to evaluate the disparities between conventional CT (c-CT) and sFOV-CT images using a lung nodule system based on MIL and assessments from radiologists. 112 patients who underwent chest CT were retrospectively enrolled in this study between July 2021 to March 2022. After undergoing c-CT examinations, sFOV-CT images with small-field-of-view were reconstructed. Two radiologists analyzed all c-CT and sFOV-CT images, including features such as location, nodule type, size, CT values, and shape signs. Then, an MIL-based lung nodule system objectively analyzed the c-CT (c-MIL) and sFOV-CT (sFOV-MIL) to explore their differences. The signal-to-noise ratio of lungs (SNR-lung) and contrast-to-noise ratio of nodules (CNR-nodule) were calculated to evaluate the quality of CT images from another perspective. The subjective evaluation by radiologists showed that feature of minimal CT value (p = 0.019) had statistical significance between c-CT and sFOV-CT. However, most features (all with p < 0.05), except for nodule type, location, volume, mean CT value, and vacuole sign (p = 0.056-1.000), had statistical differences between c-MIL and sFOV-MIL by MIL system. The SNR-lung between c-CT and sFOV-CT had no statistical significance, while the CNR-nodule showed statistical difference (p = 0.007), and the CNR of sFOV-CT was higher than that of c-CT. In detecting the difference between c-CT and sFOV-CT, features extracted by the MIL system had more statistical differences than those evaluated by radiologists. The image quality of those two CT images was different, and the CNR-nodule of sFOV-CT was higher than that of c-CT.

An Evaluation Analysis for Computed Tomography Image Quality of Primary Liver Cancer Lesions Based on Deep Learning Image Reconstruction.

BACKGROUND: Abdominal multi-slice helical computed tomography (CT) and contrast-enhanced scanning have been widely recognized clinically. OBJECTIVE: The impact of the deep learning image reconstruction (DLIR) on the quality of dynamic contrast-enhanced CT imaging of primary liver cancer lesions was evaluated through comparison with the filtered back projection (FBP) and the new generation of adaptive statistical iterative reconstruction-V (ASIR-V). METHODS: We evaluated the image noise of the lesion, fine structures inside the lesion, and diagnostic confidence in 48 liver cancer subjects. The CT values of the solid part of the lesion and the adjacent normal liver tissue and the systolic and diastolic blood pressure (SD) values of the right paravertebral muscle were measured. The muscle SD value was considered as the background noise of the image, and the signal noise ratio (SNR) and contrast signal-to-noise ratio (CNR) of the lesion and normal liver parenchyma were calculated. RESULTS: High consistency in the evaluation of image noise (Kappa = 0.717). The Kappa values for margin/pseudocapsule, fine structure within the lesion, and diagnostic confidence were 0.463, 0.527, and 0.625, respectively. Besides, the differences in SD, SNR and CNR data of reconstructed lesion images among the six groups were statistically significant. CONCLUSION: The contrast-enhanced CT image noise of DLIR-H in the portal venous phase is much lower than that of ASIR-V and FBP in primary liver cancer patients. In terms of the lesion structure display, the new reconstruction algorithm DLIR is superior.