Journal of Computer Assisted Tomography最新文献

Objective: This study aims to evaluate, on one MRI vendor's platform, the impact of deep learning (DL)-based reconstruction techniques on MRI radiomic features compared to conventional image reconstruction techniques.

Methods: Under IRB approval and informed consent, we prospectively collected undersampled coronal T2-weighted MR images of the abdomen (1.5 T; Philips Healthcare) from 17 pediatric and adult subjects and reconstructed them using a conventional image reconstruction technique (compressed sensitivity encoding [C-SENSE]) and two DL-based reconstruction techniques (SmartSpeed [Philips Healthcare, US FDA cleared] and SmartSpeed with Super Resolution [SmartSpeed-SuperRes, not US FDA cleared to date]). Eight regions of interest (ROIs) across organs/tissues (liver, spleen, kidney, pancreas, fat, and muscle) were manually placed. Eighty-six MRI radiomic features were then extracted. Pearson's correlation coefficients (PCCs) and intraclass correlation coefficients (ICCs) were calculated between (A) C-SENSE versus SmartSpeed, and (B) C-SENSE versus SmartSpeed-SuperRes. To quantify the impact from the perspective of the whole MR image, cross-ROI mean PCCs and ICCs were calculated for individual radiomic features. The impact of image reconstruction on individual radiomic features in different organs/tissues was evaluated using ANOVA analyses.

Results: According to cross-ROI mean PCCs, 50 out of 86 radiomic features were highly correlated (PCC, ≥0.8) between SmartSpeed and C-SENSE, whereas only 15 radiomic features were highly correlated between SmartSpeed-SuperRes and C-SENSE reconstructions. According to cross-ROI mean ICCs, 58 out of 86 radiomic features had high agreements (ICC ≥0.75) between SmartSpeed and C-SENSE, whereas only 9 radiomic features had high agreements between SmartSpeed-SuperRes and C-SENSE reconstructions. For SmartSpeed reconstruction, the psoas muscle ROI appeared to be impacted most with the lowest median (IQR) correlation of 0.57 (0.25). The circular liver ROI was impacted most by SmartSpeed-SuperRes (PCC, 0.60 [0.22]). ANOVA analyses suggest that the impact of DL reconstruction algorithms on radiomic features varies significantly among different organs/tissues ( P < 0.001).

Conclusions: MRI radiomic features are significantly altered by DL-based reconstruction compared to a conventional reconstruction technique. The impact of DL reconstruction algorithms on radiomic features varies significantly between different organs/tissues.

Objective: We devised a split-bolus injection and imaging protocol for pulmonary artery and vein separation computed tomography (CT) angiography based on time enhancement curve characterization. Furthermore, we aimed to evaluate the contrast enhancement effect and success rate of blood vessel separation between the pulmonary artery and vein of this proposed protocol.

Methods: In this study, 102 patients (45 patients with the standard protocol and 57 patients with the proposed protocol) who underwent pulmonary arteriovenous computed tomography angiography were included. The CT values of various vessels, CT value difference between the pulmonary trunk and left atrium, and coefficient of variation in pulmonary arteries and veins were obtained from images of the standard and proposed protocols.

Results: The CT values in the proposed protocol for the pulmonary trunk were significantly higher than those in the standard protocol (487.3 [415.5-546.9] HU vs. 293.0 [259.0-350.0] HU, P < 0.01). The CT value difference between the pulmonary trunk and left atrium in the proposed protocol was significantly higher than that in the conventional protocol (211.3 [158.0-265.7] HU vs. 32 [-30.0-55.0] HU, P < 0.01). The coefficient of variation in the proposed protocol was 0.08 (0.06-0.10) and 0.09 (0.08-0.11) in pulmonary arteries and 0.08 (0.06-0.09) and 0.09 (0.07-0.12) in pulmonary veins, respectively.

Conclusions: The proposed protocol achieved separation between the pulmonary artery and vein in many patients, making it useful for the preoperative assessment of individual thoracic anatomy.

Objective: The aim of this study was to explore whether machine learning model based on computed tomography (CT) radiomics and clinical characteristics can differentiate Epstein-Barr virus-associated gastric cancer (EBVaGC) from non-EBVaGC.

Methods: Contrast-enhanced CT images were collected from 158 patients with GC (46 EBV-positive, 112 EBV-negative) between April 2018 and February 2023. Radiomics features were extracted from the volumes of interest. A radiomics signature was built based on radiomics features by the least absolute shrinkage and selection operator logistic regression algorithm. Multivariate analyses were used to identify significant clinicoradiological variables. We developed 6 ML models for EBVaGC, including logistic regression, Extreme Gradient Boosting, random forest (RF), support vector machine, Gaussian Naive Bayes, and K-nearest neighbor algorithm. The area under the receiver operating characteristic curve (AUC), the area under the precision-recall curves (AP), calibration plots, and decision curve analysis were applied to assess the effectiveness of each model.

Results: Six ML models achieved AUC of 0.706-0.854 and AP of 0.480-0.793 for predicting EBV status in GC. With an AUC of 0.854 and an AP of 0.793, the RF model performed the best. The forest plot of the AUC score revealed that the RF model had the most stable performance, with a standard deviation of 0.003 for AUC score. RF also performed well in the testing dataset, with an AUC of 0.832 (95% confidence interval: 0.679-0.951), accuracy of 0.833, sensitivity of 0.857, and specificity of 0.824, respectively.

Conclusions: The RF model based on clinical variables and Rad_score can serve as a noninvasive tool to evaluate the EBV status of gastric cancer.

Objective: This study aimed to evaluate a vendor-specific correction software for apparent diffusion coefficient (ADC) bias due to gradient nonlinearity in breast diffusion-weighted magnetic resonance imaging using an ice-water phantom.

Methods: The phantom consists of 5 plastic tubes with a length of 100 mm and a diameter of 15 mm, filled with distilled water and immersed in an ice-water bath. Diffusion-weighted images were acquired by echo-planar imaging sequence on a 3.0-T scanner. ADC maps with and without correction were calculated using 4 b -values (0, 100, 600, and 800 s/mm 2 ). The mean ADCs were measured using a rectangular profile with 5 × 40 pixels in the anterior-posterior (AP) and a square region of interest with 5 × 5 pixels in the right-left (RL) and superior-inferior (SI) directions on the ADC map. ADC was compared with and without correction using a paired t test. Additionally, ADC of the ice-water phantom was measured at the magnet isocenter.

Results: ADC increased in the AP and RL directions and decreased in the SI direction with increasing distance from the isocenter before correction. After the correction, ADC at the off-center positions in the AP, RL, and SI directions was reduced to within 5% of the expected value. There were significant differences in the ADC at the off-center positions without and with correction ( P < 0.001); however, ADC at the magnet isocenter did not vary after correction (1.08 ± 0.02 × 10 -3 mm 2 /s).

Conclusions: The vendor-specific software corrected the ADC bias due to gradient nonlinearity at the off-center positions in the AP, RL, and SI directions. Therefore, the software will contribute to the accurate ADC assessment in breast DWI.

Objective: The objectives of this research were to assess the effectiveness of computer-assisted detection reading (CADR) and structured reports in monitoring patients with multiple sclerosis (MS) and to evaluate the role of radiology technicians in this context.

Methods: Eighty-seven patients with MS who underwent at least 2 sequential magnetic resonance imaging (MRI) follow-ups analyzed by 2 radiologists and a technician. Progression of disease (POD) was identified through the emergence of T2 fluid-attenuated inversion recovery white matter hyperintensities or contrast enhancements and evaluated both qualitatively (progression vs stability) and quantitatively (count of new white matter hyperintensities).

Results: CADR increased the accuracy by 11%, enhancing interobserver consensus on qualitative progression and saving approximately 2 minutes per examination. Although structured reports did not improve these metrics, it may improve clinical communication and permit technicians to achieve approximately 80% accuracy in MRI readings.

Conclusions: The use of CADR improves the accuracy, agreement, and interpretation time in MRI follow-ups of MS. With the help of computer tools, radiology technicians could represent a significant aid in the follow-up of these patients.

Background: Lung cancer has the highest morbidity and mortality in the world, and immunotherapies have been developed for this disease in recent years. However, activation of the immune system can cause immune-related adverse events (irAEs), and checkpoint inhibitor-related pneumonitis (CIP), can be the most severe and fatal. But few reports have systematically examined the spectrum of imaging findings of this condition. Therefore, the objective of this paper is to investigate the high-resolution computed tomography (HRCT) characteristics of CIP in patients with lung cancer.

Objective: To investigate the HRCT characteristics of CIP in patients with lung cancer.

Methods: HRCT patterns in 41 lung cancer patients who developed CIP after treatment with immune checkpoint inhibitors were retrospectively characterized by interstitial lung disease classification, and their severity was graded. Specific HRCT characteristics related to CIP were identified.

Results: There are 4 types of immunotherapy-induce pneumonitis patterns (organizing pneumonia OP 19 cases, nonspecific interstitial pneumonia NSIP 8 cases, acute interstitial pneumonia AIP 7 cases, 7 cases of undetermined type) and image grade (13 cases of grade 1, 17 cases of grade 2, 11 cases of grade 3, 0 cases of grade 4) were identified. Spatial distribution characteristics of these lesions were noted (17 cases predominantly distributed in tumor-containing lobes, 6 cases predominantly distributed in non-tumor-containing lobes, and no specific predilection in 18 cases). Specific CT imaging features found in CIP included, in the order of prevalence, the following: ground glass opacities (38 cases), subpleural/vertical line (37 cases), interstitial thickening around the bronchovascular bundles (36 cases), reticulation (34 cases), fine reticular shadow (31 cases), consolidation (31 cases), small cystic shadow (24 cases, may not having honeycombing), small nodules (17 cases), bronchiectasis (15 cases), honeycombing (11 cases), mosaic sign (11 cases), and pleural effusion (18 cases).

Conclusion: HRCT of CIP predominantly manifests as ground glass opacities, reticulation, subpleural/vertical line, interstitial thickening around the bronchovascular bundle, and consolidation.

Objective: The aim of this study was to develop and validate an interpretable and highly generalizable multimodal radiomics model for predicting the prognosis of patients with cerebral hemorrhage.

Methods: This retrospective study involved 237 patients with cerebral hemorrhage from 3 medical centers, of which a training cohort of 186 patients (medical center 1) was selected and 51 patients from medical center 2 and medical center 3 were used as an external testing cohort. A total of 1762 radiomics features were extracted from nonenhanced computed tomography using Pyradiomics, and the relevant macroscopic imaging features and clinical factors were evaluated by 2 experienced radiologists. A radiomics model was established based on radiomics features using the random forest algorithm, and a radiomics-clinical model was further trained by combining radiomics features, clinical factors, and macroscopic imaging features. The performance of the models was evaluated using area under the curve (AUC), sensitivity, specificity, and calibration curves. Additionally, a novel SHAP (SHAPley Additive exPlanations) method was used to provide quantitative interpretability analysis for the optimal model.

Results: The radiomics-clinical model demonstrated superior predictive performance overall, with an AUC of 0.88 (95% confidence interval, 0.76-0.95; P < 0.01). Compared with the radiomics model (AUC, 0.85; 95% confidence interval, 0.72-0.94; P < 0.01), there was a 0.03 improvement in AUC. Furthermore, SHAP analysis revealed that the fusion features, rad score and clinical rad score, made significant contributions to the model's decision-making process.

Conclusion: Both proposed prognostic models for cerebral hemorrhage demonstrated high predictive levels, and the addition of macroscopic imaging features effectively improved the prognostic ability of the radiomics-clinical model. The radiomics-clinical model provides a higher level of predictive performance and model decision-making basis for the risk prognosis of cerebral hemorrhage.

Objective: The purpose of this study is to explore the impact of deep learning image reconstruction (DLIR) algorithm on the quantification of radiomic features in ultra-low-dose computed tomography (ULD-CT) compared with adaptive statistical iterative reconstruction-Veo (ASIR-V).

Methods: One hundred eighty-three patients with pulmonary nodules underwent standard-dose computed tomography (SDCT) (4.30 ± 0.36 mSv) and ULD-CT (UL-A, 0.57 ± 0.09 mSv or UL-B, 0.33 ± 0.04 mSv). SDCT was the reference standard using (ASIR-V) at 50% strength (50%ASIR-V). ULD-CT was reconstructed with 50%ASIR-V, DLIR at medium and high strength (DLIR-M, DLIR-H). Radiomics analysis extracted 102 features, and the intraclass correlation coefficient (ICC) quantified reproducibility between ULD-CT and SDCT reconstructed by 50%ASIR-V, DLIR-M, and DLIR-H for each feature.

Results: Among 102 radiomic features, the percentages of reproducibility of 50%ASIR-V, DLIR-M, and DLIR-H were 48.04% (49/102), 49.02% (50/102), and 52.94% (54/102), respectively. Shape and first order features demonstrated high reproducibility across different reconstruction algorithms and radiation doses, with mean ICC values exceeding 0.75. In texture features, DLIR-M and DLIR-H showed improved mean ICC values for pure ground glass nodules (pGGNs) from 0.69 ± 0.23 to 0.75 ± 0.18 and 0.81 ± 0.12, respectively, compared with 50%ASIR-V. Similarly, the mean ICC values for solid nodules (SNs) increased from 0.60 ± 0.19 to 0.66 ± 0.14 and 0.69 ± 0.13, respectively. Additionally, the mean ICC values of texture features for pGGNs and SNs in both ULD-CT groups decreased with reduced radiation dose.

Conclusions: DLIR can improve the reproducibility of radiomic features at ultra-low doses compared with ASIR-V. In addition, pGGNs showed better reproducibility at ultra-low doses than SNs.

Objective: Pulmonary CT angiography (CTA) to detect pulmonary emboli can be performed using conventional dual-source CT with single-energy acquisition at high-pitch (high-pitch conventional CT), which minimizes motion artifacts, or routine-pitch, dual-energy acquisitions (routine-pitch conventional DECT), which maximize iodine signal. We compared iodine signal, radiation dose, and motion artifacts of pulmonary CTA between these conventional CT modalities and dual-source photon-counting detector CT with high-pitch, multienergy acquisitions (high-pitch photon-counting CT).

Methods: Consecutive clinically indicated pulmonary CTA exams were collected. CT number/noise was measured from the main to right lower lobe segmental pulmonary arteries using 120 kV threshold low, 120 kV, and mixed kV (0.6 linear blend) images. Three radiologists reviewed anonymized, randomized exams, rating them using a 4- or 5-point Likert scale (1 = worst, and 4/5 = best) for contrast enhancement in pulmonary arteries, motion artifacts in aortic root to subsegmental pulmonary arteries, lung image quality; pulmonary blood volume (PBV) map image quality (for multienergy or dual-energy exams), and contribution to reader confidence.

Results: One hundred fifty patients underwent high-pitch photon-counting CT (n = 50), high-pitch conventional CT (n = 50), and routine-pitch conventional DECT (n = 50). High-pitch photon-counting CT had lower radiation dose (CTDI vol : 8.1 ± 2.5 vs 9.6 ± 6.8 and 16.2 ± 8.5 mGy, respectively; P < 0.001), and routine-pitch conventional DECT had significantly less contrast ( P < 0.009). CT number and CNR measurements were significantly greater at high-pitch photon-counting CT ( P < 0.001). Across readers, high-pitch photon-counting CT demonstrated significantly higher subjective contrast enhancement in the pulmonary arteries compared to the other modalities (4.7 ± 0.6 vs 4.4 ± 0.7 vs 4.3 ± 0.7; P = 0.011) and lung image quality (3.4 ± 0.5 vs 3.1 ± 0.5 vs 3.1 ± 0.5; P = 0.013). High-pitch photon-counting CT and high-pitch conventional CT had fewer motion artifacts at all levels compared to DECT ( P < 0.001). High-pitch photon-counting CT PBV maps had superior image quality ( P < 0.001) and contribution to reader confidence ( P < 0.001) compared to routine-pitch conventional DECT.

Conclusion: High-pitch photon-counting pulmonary CTA demonstrated higher contrast in pulmonary arteries at lower radiation doses with improved lung image quality and fewer motion artifacts compared to high-pitch conventional CT and routine-pitch conventional dual-energy CT.

Objective: Extracellular volume fraction (fECV) and liver and spleen size have been correlated with liver fibrosis stages and cirrhosis. The purpose of the current study was to determine the predictive value of fECV alone and in conjunction with measurement of liver and spleen size for severity of liver fibrosis.

Methods: This was a retrospective study of 95 subjects (65 with liver biopsy and 30 controls). Spearman rank correlation coefficient was used to assess correlation between radiological markers and fibrosis stage. Receiver operating characteristic analysis was performed to assess the discriminative ability of radiological markers for significant (F2+) and advanced (F3+) fibrosis and cirrhosis (F4), by reporting the area under the curve (AUC).

Results: The cohort had a mean age of 51.4 ± 14.4 years, and 52 were female (55%). There were 36, 5, 6, 9, and 39 in fibrosis stages F0, F1, F2, F3, and F4, respectively. Spleen volume alone showed the highest correlation ( r = 0.552, P < 0.001) and AUCs of 0.823, 0.807, and 0.785 for identification of significant and advanced fibrosis and cirrhosis, respectively. Adding fECV to spleen length improved AUCs (0.764, 0.745, and 0.717 to 0.812, 0.781, and 0.738, respectively) compared with splenic length alone. However, adding fECV to spleen volume did not improve the AUCs for significant or advanced fibrosis or cirrhosis.

Conclusions: Spleen size (measured in length or volume) showed better correlation with liver fibrosis stages compared with fECV. The combination of fECV and spleen length had higher accuracy compared with fECV alone or spleen length alone.