European Radiology Experimental最新文献

Background: In liver computed tomography (CT), tailoring the contrast injection to the patient's specific characteristics is relevant for optimal imaging and patient safety. We evaluated a novel algorithm engineered for personalized contrast injection to achieve reproducible liver enhancement centered on 50 HU.

Methods: From September 2020 to August 31, 2022, CT data from consecutive adult patients were prospectively collected at our multicenter premises. Inclusion criteria consisted of an abdominal CT referral for cancer staging or follow-up. For all examinations, a web interface incorporating data from the radiology information system (patient details and examination information) and radiographer-inputted data (patient fat-free mass, imaging center, kVp, contrast agent details, and imaging phase) were used. Calculated contrast volume and injection rate were manually entered into the CT console controlling the injector. Iopamidol 370 mgI/mL or Iohexol 350 mgI/mL were used, and kVp varied (80, 100, or 120) based on patient habitus.

Results: We enrolled 384 patients (mean age 61.2 years, range 21.1-94.5). The amount of administered iodine dose (gI) was not significantly different across contrast agents (p = 0.700), while a significant increase in iodine dose was observed with increasing kVp (p < 0.001) and in males versus females (p < 0.001), as expected. Despite the differences in administered iodine load, image quality was reproducible across patients with 72.1% of the examinations falling within the desirable range of 40-60 HU.

Conclusion: This study validated a novel algorithm for personalized contrast injection in adult abdominal CT, achieving consistent liver enhancement centered at 50 HU.

Relevance statement: In healthcare's ongoing shift towards personalized medicine, the algorithm offers excellent potential to improve diagnostic accuracy and patient management, particularly for the detection and follow-up of liver malignancies.

Key points: The algorithm achieves reproducible liver enhancement, promising improved diagnostic accuracy and patient management in diverse clinical settings. The real-world study demonstrates this algorithm's adaptability to different variables ensuring high-quality liver imaging. A personalized algorithm optimizes liver CT, improving the visibility, conspicuity, and follow-up of liver lesions.

The growing use of artificial neural network (ANN) tools for computed tomography angiography (CTA) data analysis underscores the necessity for elevated data protection measures. We aimed to establish an automated defacing pipeline for CTA data. In this retrospective study, CTA data from multi-institutional cohorts were utilized to annotate facemasks (n = 100) and train an ANN model, subsequently tested on an external institution's dataset (n = 50) and compared to a publicly available defacing algorithm. Face detection (MTCNN) and verification (FaceNet) networks were applied to measure the similarity between the original and defaced CTA images. Dice similarity coefficient (DSC), face detection probability, and face similarity measures were calculated to evaluate model performance. The CTA-DEFACE model effectively segmented soft face tissue in CTA data achieving a DSC of 0.94 ± 0.02 (mean ± standard deviation) on the test set. Our model was benchmarked against a publicly available defacing algorithm. After applying face detection and verification networks, our model showed substantially reduced face detection probability (p < 0.001) and similarity to the original CTA image (p < 0.001). The CTA-DEFACE model enabled robust and precise defacing of CTA data. The trained network is publicly accessible at www.github.com/neuroAI-HD/CTA-DEFACE . RELEVANCE STATEMENT: The ANN model CTA-DEFACE, developed for automatic defacing of CT angiography images, achieves significantly lower face detection probabilities and greater dissimilarity from the original images compared to a publicly available model. The algorithm has been externally validated and is publicly accessible. KEY POINTS: The developed ANN model (CTA-DEFACE) automatically generates facemasks for CT angiography images. CTA-DEFACE offers superior deidentification capabilities compared to a publicly available model. By means of graphics processing unit optimization, our model ensures rapid processing of medical images. Our model underwent external validation, underscoring its reliability for real-world application.

Background: Renal quantitative measurements are important descriptors for assessing kidney function. We developed a deep learning-based method for automated kidney measurements from computed tomography (CT) images.

Methods: The study datasets comprised potential kidney donors (n = 88), both contrast-enhanced (Dataset 1 CE) and noncontrast (Dataset 1 NC) CT scans, and test sets of contrast-enhanced cases (Test set 2, n = 18), cases from a photon-counting (PC)CT scanner reconstructed at 60 and 190 keV (Test set 3 PCCT, n = 15), and low-dose cases (Test set 4, n = 8), which were retrospectively analyzed to train, validate, and test two networks for kidney segmentation and subsequent measurements. Segmentation performance was evaluated using the Dice similarity coefficient (DSC). The quantitative measurements' effectiveness was compared to manual annotations using the intraclass correlation coefficient (ICC).

Results: The contrast-enhanced and noncontrast models demonstrated excellent reliability in renal segmentation with DSC of 0.95 (Test set 1 CE), 0.94 (Test set 2), 0.92 (Test set 3 PCCT) and 0.94 (Test set 1 NC), 0.92 (Test set 3 PCCT), and 0.93 (Test set 4). Volume estimation was accurate with mean volume errors of 4%, 3%, 6% mL (contrast test sets) and 4%, 5%, 7% mL (noncontrast test sets). Renal axes measurements (length, width, and thickness) had ICC values greater than 0.90 (p < 0.001) for all test sets, supported by narrow 95% confidence intervals.

Conclusion: Two deep learning networks were shown to derive quantitative measurements from contrast-enhanced and noncontrast renal CT imaging at the human performance level.

Relevance statement: Deep learning-based networks can automatically obtain renal clinical descriptors from both noncontrast and contrast-enhanced CT images. When healthy subjects comprise the training cohort, careful consideration is required during model adaptation, especially in scenarios involving unhealthy kidneys. This creates an opportunity for improved clinical decision-making without labor-intensive manual effort.

Key points: Trained 3D UNet models quantify renal measurements from contrast and noncontrast CT. The models performed interchangeably to the manual annotator and to each other. The models can provide expert-level, quantitative, accurate, and rapid renal measurements.

Background: Longitudinal T1 relaxation time is a key imaging biomarker. In addition, T1 values are modulated by the administration of T1 contrast agents used in patients with tumors and metastases. However, in clinical practice, dedicated T1 mapping sequences are often not included in brain MRI protocols. The aim of this study is to address the absence of dedicated T1 mapping sequences in imaging protocol by deriving T1 maps from standard T1-weighted sequences.

Methods: A phantom, composed of 144 solutions of paramagnetic agents at different concentrations, was imaged with a three-dimensional (3D) T1-weighed turbo spin-echo (TSE) sequence designed for brain imaging. The relationship between the T1 values and the signal intensities was established using this phantom acquisition. T1 mapping derived from 3D T1-weighted TSE acquisitions in four healthy volunteers and one patient with brain metastases were established and compared to reference T1 mapping technique. The concentration of Gd-based contrast agents in brain metastases were assessed from the derived T1 maps.

Results: Based on the phantom acquisition, the relationship between T1 values and signal intensity (SI) was found equal to T1 = 0.35 × SI^-1.11 (R² = 0.97). TSE-derived T1 values measured in white matter and gray matter in healthy volunteers were equal to 0.997 ± 0.096 s and 1.358 ± 0.056 s (mean ± standard deviation), respectively. Mean Gd³⁺ concentration value in brain metastases was 94.7 ± 30.0 μM.

Conclusion: The in vivo results support the relevance of the phantom-based approach: brain T1 maps can be derived from T1-weighted acquisitions.

Relevance statement: High-resolution brain T1 maps can be generated, and contrast agent concentration can be quantified and imaged in brain metastases using routine 3D T1-weighted TSE acquisitions.

Key points: Quantitative T1 mapping adds significant value to MRI diagnostics. T1 measurement sequences are rarely included in routine protocols. T1 mapping and concentration of contrast agents can be derived from routine standard scans. The diagnostic value of MRI can be improved without additional scan time.

Background: To assess thoracic vascular computed tomography (CT) contrast enhancement of a novel intravenous tantalum oxide nanoparticle contrast agent (carboxybetaine zwitterionic tantalum oxide, TaCZ) compared to a conventional iodinated contrast agent (Iopamidol) in a rabbit multiphase protocol.

Methods: Five rabbits were scanned inside a human-torso-sized encasement on a clinical CT system at various scan delays after intravenous injection of 540 mg element (Ta or I) per kg of bodyweight of TaCZ or Iopamidol. Net contrast enhancement of various arteries and veins, as well as image noise, were measured. Randomized scan series were reviewed by three independent readers on a clinical workstation and assessed for vascular conspicuity and image artifacts on 5-point Likert scales.

Results: Overall, net vascular contrast enhancement achieved with TaCZ was superior to Iopamidol (p ≤ 0.036 with the exception of the inferior vena cava at 6 s (p = 0.131). Vascular contrast enhancement achieved with TaCZ at delays of 6 s, 40 s, and 75 s was superior to optimum achieved Iopamidol contrast enhancement at 6 s (p ≤ 0.036. Vascular conspicuity was higher for TaCZ in 269 of 300 (89.7%) arterial and 269 of 300 (89.7%) venous vessel assessments, respectively (p ≤ 0.005), with substantial inter-reader reliability (κ = 0.61; p < 0.001) and strong positive monotonic correlation between conspicuity scores and contrast enhancement measurements (ρ = 0.828; p < 0.001).

Conclusion: TaCZ provides absolute and relative contrast advantages compared to Iopamidol for improved visualization of thoracic arteries and veins in a multiphase CT protocol.

Relevance statement: The tantalum-oxide nanoparticle is an experimental intravenous CT contrast agent with superior cardiovascular and venous contrast capacity per injected elemental mass in an animal model, providing improved maximum contrast enhancement and prolonged contrast conspicuity. Further translational research on promising high-Z and nanoparticle contrast agents is warranted.

Key points: There have been no major advancements in intravenous CT contrast agents over decades. Iodinated CT contrast agents require optimal timing for angiography and phlebography. Tantalum-oxide demonstrated increased CT attenuation per elemental mass compared to Iopamidol. Nanoparticle contrast agent design facilitates prolonged vascular conspicuity.

Background: To explore an artificial intelligence (AI) technology employing YOLOv8 for quality control (QC) on elbow joint radiographs.

Methods: From January 2022 to August 2023, 2643 consecutive elbow radiographs were collected and randomly assigned to the training, validation, and test sets in a 6:2:2 ratio. We proposed the anteroposterior (AP) and lateral (LAT) models to identify target detection boxes and key points on elbow radiographs using YOLOv8. These identifications were transformed into five quality standards: (1) AP elbow positioning coordinates (X_A and Y_A); (2) olecranon fossa positioning distance parameters (S₁₇ and S₂₇); (3) key points of joint space (Y₃, Y₄, Y₅ and Y₆); (4) LAT elbow positioning coordinates (X₂ and Y₂); and (5) flexion angle. Models were trained and validated using 2,120 radiographs. A test set of 523 radiographs was used for assessing the agreement between AI and physician and to evaluate clinical efficiency of models.

Results: The AP and LAT models demonstrated high precision, recall, and mean average precision for identifying boxes and points. AI and physicians showed high intraclass correlation coefficient (ICC) in evaluating: AP coordinates X_A (0.987) and Y_A (0.991); olecranon fossa parameters S₁₇ (0.964) and S₂₇ (0.951); key points Y₃ (0.998), Y₄ (0.997), Y₅ (0.998) and Y₆ (0.959); LAT coordinates X₂ (0.994) and Y₂ (0.986); and flexion angle (0.865). Compared to manual methods, using AI, QC time was reduced by 43% for AP images and 45% for LAT images (p < 0.001).

Conclusion: YOLOv8-based AI technology is feasible for QC of elbow radiography with high performance.

Relevance statement: This study proposed and validated a YOLOv8-based AI model for automated quality control in elbow radiography, obtaining high efficiency in clinical settings.

Key points: QC of elbow joint radiography is important for detecting diseases. Models based on YOLOv8 are proposed and perform well in image QC. Models offer objective and efficient solutions for QC in elbow joint radiographs.

Background: Regular disease monitoring with low-dose high-resolution (LD-HR) computed tomography (CT) scans is necessary for the clinical management of people with cystic fibrosis (pwCF). The aim of this study was to compare the image quality and radiation dose of LD-HR protocols between photon-counting CT (PCCT) and energy-integrating detector system CT (EID-CT) in pwCF.

Methods: This retrospective study included 23 pwCF undergoing LD-HR chest CT with PCCT who had previously undergone LD-HR chest CT with EID-CT. An intraindividual comparison of radiation dose and image quality was conducted. The study measured the dose-length product, volumetric CT dose index, effective dose and signal-to-noise ratio (SNR). Three blinded radiologists assessed the overall image quality, image sharpness, and image noise using a 5-point Likert scale ranging from 1 (deficient) to 5 (very good) for image quality and image sharpness and from 1 (very high) to 5 (very low) for image noise.

Results: PCCT used approximately 42% less radiation dose than EID-CT (median effective dose 0.54 versus 0.93 mSv, p < 0.001). PCCT was consistently rated higher than EID-CT for overall image quality and image sharpness. Additionally, image noise was lower with PCCT compared to EID-CT. The average SNR of the lung parenchyma was lower with PCCT compared to EID-CT (p < 0.001).

Conclusion: In pwCF, LD-HR chest CT protocols using PCCT scans provided significantly better image quality and reduced radiation exposure compared to EID-CT.

Relevance statement: In pwCF, regular follow-up could be performed through photon-counting CT instead of EID-CT, with substantial advantages in terms of both lower radiation exposure and increased image quality.

Key points: Photon-counting CT (PCCT) and energy-integrating detector system CT (EID-CT) were compared in 23 people with cystic fibrosis (pwCF). Image quality was rated higher for PCCT than for EID-CT. PCCT used approximately 42% less radiation dose and offered superior image quality than EID-CT.

Background: Patellar instability is a well-known pathology in which kinematics can be investigated using metrics such as tibial tuberosity tracheal groove (TTTG), the bisect offset (BO), and the lateral patellar tilt (LPT). We used dynamic computed tomography (CT) to investigate the patellar motion of healthy subjects in weight-bearing conditions to provide normative values for TTTG, BO, and LPT, as well as to define whether BO and LPT are affected by the morphology of the trochlear groove.

Methods: Dynamic scanning was used to acquire images during weight-bearing in 21 adult healthy volunteers. TTTG, BO, and LPT metrics were computed between 0° and 30° of knee flexion. Sulcus angle, sulcus depth, and lateral trochlear inclination were calculated and used with the TTTG for simple linear regression models.

Results: All metrics gradually decreased during eccentric movement (TTTG, -6.9 mm; BO, -12.6%; LPT, -4.3°). No significant differences were observed between eccentric and concentric phases at any flexion angle for all metrics. Linear regression between kinematic metrics towards full extension showed a moderate fit between BO and TTTG (R² 0.60, β 1.75) and BO and LPT (R² 0.59, β 1.49), and a low fit between TTTG and LPT (R² 0.38, β 0.53). A high impact of the TTTG distance over BO was shown in male participants (R² 0.71, β 1.89) and patella alta individuals (R² 0.55, β 1.91).

Conclusion: We provided preliminary normative values of three common metrics during weight-bearing dynamic CT and showed the substantial impact of lateralisation of the patella tendon over patella displacement.

Relevance statement: These normative values can be used by clinicians when evaluating knee patients using TTTG, BO, and LPT metrics. The lateralisation of the patellar tendon in subjects with patella alta or in males significantly impacts the lateral displacement of the patella.

Key points: Trochlear groove morphology had no substantial impact on motion prediction. The lateralisation of the patellar tendon seems a strong predictor of lateral displacement of the patella in male participants. Participants with patella alta displayed a strong fit between the patellar lateral displacement and tilt. TTTG, BO, and LPT decreased during concentric movement. Concentric and eccentric phases did not show differences for all metrics.

Background

The intricate three-dimensional anatomy of the inner ear presents significant challenges in diagnostic procedures and critical surgical interventions. Recent advancements in deep learning (DL), particularly convolutional neural networks (CNN), have shown promise for segmenting specific structures in medical imaging. This study aimed to train and externally validate an open-source U-net DL general model for automated segmentation of the inner ear from computed tomography (CT) scans, using quantitative and qualitative assessments.

Methods

In this multicenter study, we retrospectively collected a dataset of 271 CT scans to train an open-source U-net CNN model. An external set of 70 CT scans was used to evaluate the performance of the trained model. The model’s efficacy was quantitatively assessed using the Dice similarity coefficient (DSC) and qualitatively assessed using a 4-level Likert score. For comparative analysis, manual segmentation served as the reference standard, with assessments made on both training and validation datasets, as well as stratified analysis of normal and pathological subgroups.

Results

The optimized model yielded a mean DSC of 0.83 and achieved a Likert score of 1 in 42% of the cases, in conjunction with a significantly reduced processing time. Nevertheless, 27% of the patients received an indeterminate Likert score of 4. Overall, the mean DSCs were notably higher in the validation dataset than in the training dataset.

Conclusion

This study supports the external validation of an open-source U-net model for the automated segmentation of the inner ear from CT scans.

Relevance statement

This study optimized and assessed an open-source general deep learning model for automated segmentation of the inner ear using temporal CT scans, offering perspectives for application in clinical routine. The model weights, study datasets, and baseline model are worldwide accessible.

Key Points

• A general open-source deep learning model was trained for CT automated inner ear segmentation.

• The Dice similarity coefficient was 0.83 and a Likert score of 1 was attributed to 42% of automated segmentations.

• The influence of scanning protocols on the model performances remains to be assessed.

Graphical Abstract

Background

We aimed to determine the capabilities of compressed sensing (CS) and deep learning reconstruction (DLR) with those of conventional parallel imaging (PI) for improving image quality while reducing examination time on female pelvic 1.5-T magnetic resonance imaging (MRI).

Methods

Fifty-two consecutive female patients with various pelvic diseases underwent MRI with T1- and T2-weighted sequences using CS and PI. All CS data was reconstructed with and without DLR. Signal-to-noise ratio (SNR) of muscle and contrast-to-noise ratio (CNR) between fat tissue and iliac muscle on T1-weighted images (T1WI) and between myometrium and straight muscle on T2-weighted images (T2WI) were determined through region-of-interest measurements. Overall image quality (OIQ) and diagnostic confidence level (DCL) were evaluated on 5-point scales. SNRs and CNRs were compared using Tukey’s test, and qualitative indexes using the Wilcoxon signed-rank test.

Results

SNRs of T1WI and T2WI obtained using CS with DLR were higher than those using CS without DLR or conventional PI (p < 0.010). CNRs of T1WI and T2WI obtained using CS with DLR were higher than those using CS without DLR or conventional PI (p < 0.003). OIQ of T1WI and T2WI obtained using CS with DLR were higher than that using CS without DLR or conventional PI (p < 0.001). DCL of T2WI obtained using CS with DLR was higher than that using conventional PI or CS without DLR (p < 0.001).

Conclusion

CS with DLR provided better image quality and shorter examination time than those obtainable with PI for female pelvic 1.5-T MRI.

Relevance statement

CS with DLR can be considered effective for attaining better image quality and shorter examination time for female pelvic MRI at 1.5 T compared with those obtainable with PI.

Key Points

• Patients underwent MRI with T1- and T2-weighted sequences using CS and PI.

• All CS data was reconstructed with and without DLR.

• CS with DLR allowed for examination times significantly shorter than those of PI and provided significantly higher signal- and CNRs, as well as OIQ.

Graphical Abstract