Factors influencing the reliability of a CT angiography-based deep learning method for infarct volume estimation.

Objectives: CT angiography (CTA)-based machine learning methods for infarct volume estimation have shown a tendency to overestimate infarct core and final infarct volumes (FIV). Our aim was to assess factors influencing the reliability of these methods. Methods: The effect of collateral circulation on the correlation between convolutional neural network (CNN) estimations and FIV was assessed based on the Miteff system and hypoperfusion intensity ratio (HIR) in 121 patients with anterior circulation acute ischaemic stroke using Pearson correlation coefficients and median volumes. Correlation was also assessed between successful and futile thrombectomies. The timing of individual CTAs in relation to CTP studies was analysed. Results: The strength of correlation between CNN estimated volumes and FIV did not change significantly depending on collateral status as assessed with the Miteff system or HIR, being poor to moderate (r = 0.09-0.50). The strongest correlation was found in patients with futile thrombectomies (r = 0.61). Median CNN estimates showed a trend for overestimation compared to FIVs. CTA was acquired in the mid arterial phase in virtually all patients (120/121). Conclusions: This study showed no effect of collateral status on the reliability of the CNN and best correlation was found in patients with futile thrombectomies. CTA timing in the mid arterial phase in virtually all patients can explain infarct volume overestimation. Advances in knowledge: CTA timing seems to be the most important factor influencing the reliability of current CTA-based machine learning methods, emphasizing the need for CTA protocol optimization for infarct core estimation.

Automatic chest computed tomography image noise quantification using deep learning.

PURPOSE: This study aimed to develop a deep learning (DL) method for noise quantification for clinical chest computed tomography (CT) images without the need for repeated scanning or homogeneous tissue regions. METHODS: A comprehensive phantom CT dataset (three dose levels, six reconstruction methods, amounting to 9240 slices) was acquired and used to train a convolutional neural network (CNN) to output an estimate of local image noise standard deviations (SD) from a single CT scan input. The CNN model consisting of seven convolutional layers was trained on the phantom image dataset representing a range of scan parameters and was tested with phantom images acquired in a variety of different scan conditions, as well as publicly available chest CT images to produce clinical noise SD maps. RESULTS: Noise SD maps predicted by the CNN agreed well with the ground truth both visually and numerically in the phantom dataset (errors of < 5 HU for most scan parameter combinations). In addition, the noise SD estimates obtained from clinical chest CT images were similar to running-average based reference estimates in areas without prominent tissue interfaces. CONCLUSIONS: Predicting local noise magnitudes without the need for repeated scans is feasible using DL. Our implementation trained with phantom data was successfully applied to open-source clinical data with heterogeneous tissue borders and textures. We suggest that automatic DL noise mapping from clinical patient images could be used as a tool for objective CT image quality estimation and protocol optimization.

Intraoperative CBCT imaging in endovascular abdomen aneurysm repair - Optimization of exposure parameters using a stent phantom.

Cone beam computed tomography (CBCT) may provide essential additional image guidance to endovascular abdominal aneurysm repair (EVAR) operations but also significant radiation exposure to patients if scans are not carefully optimized. The purpose of our study was to define the image quality requirements for intraoperative EVAR CBCT imaging and to optimize the CBCT exposure parameters accordingly. A Multi-Energy CT phantom simulating a large patient was used by replacing the central phantom cylinder with a custom water-filled insert including an EVAR stent. Different exposure parameters covering a range of radiation qualities and dose levels were used to define the optimal image quality level regarding stent graft evaluation (compressed, bent, or collapsed). The radiation dose was measured with a calibrated air kerma-area product (KAP) meter and organ doses were calculated based on Monte Carlo simulations and a mathematical patient model. Based on the results, updated exposure parameters with the highest mean energy and lowest dose level available were recommended. With the updated protocol, the radiation exposure could be significantly decreased. The KAP value decreased from 9720 µGy·m2 to 440 µGy·m2 and reference point air kerma from 351 mGy to 16 mGy (a reduction of 96%) and organ doses of the organs in the irradiated region decreased on an average 91%. The new protocol resulted in acceptable clinical image quality based on testing with clinical cases.

Leveraging open dataset and transfer learning for accurate recognition of chronic pulmonary embolism from CT angiogram maximum intensity projection images.

BACKGROUND: Early diagnosis of the potentially fatal but curable chronic pulmonary embolism (CPE) is challenging. We have developed and investigated a novel convolutional neural network (CNN) model to recognise CPE from CT pulmonary angiograms (CTPA) based on the general vascular morphology in two-dimensional (2D) maximum intensity projection images. METHODS: A CNN model was trained on a curated subset of a public pulmonary embolism CT dataset (RSPECT) with 755 CTPA studies, including patient-level labels of CPE, acute pulmonary embolism (APE), or no pulmonary embolism. CPE patients with right-to-left-ventricular ratio (RV/LV) < 1 and APE patients with RV/LV ≥ 1 were excluded from the training. Additional CNN model selection and testing were done on local data with 78 patients without the RV/LV-based exclusion. We calculated area under the receiver operating characteristic curves (AUC) and balanced accuracies to evaluate the CNN performance. RESULTS: We achieved a very high CPE versus no-CPE classification AUC 0.94 and balanced accuracy 0.89 on the local dataset using an ensemble model and considering CPE to be present in either one or both lungs. CONCLUSIONS: We propose a novel CNN model with excellent predictive accuracy to differentiate chronic pulmonary embolism with RV/LV ≥ 1 from acute pulmonary embolism and non-embolic cases from 2D maximum intensity projection reconstructions of CTPA. RELEVANCE STATEMENT: A DL CNN model identifies chronic pulmonary embolism from CTA with an excellent predictive accuracy. KEY POINTS: • Automatic recognition of CPE from computed tomography pulmonary angiography was developed. • Deep learning was applied on two-dimensional maximum intensity projection images. • A large public dataset was used for training the deep learning model. • The proposed model showed an excellent predictive accuracy.

Postoperative computed tomography imaging of pediatric patients with craniosynostosis: radiation dose and image quality comparison between multi-slice computed tomography and O-arm cone-beam computed tomography.

BACKGROUND: When postoperative multi-slice computed tomography (MSCT) imaging of patients with craniosynostosis is used, it is usually performed a few days after surgery in a radiology department. This requires additional anesthesia for the patient. Recently, intraoperative mobile cone-beam CT (CBCT) devices have gained popularity for orthopedic and neurosurgical procedures, which allows postoperative CT imaging in the operating room. OBJECTIVE: This single-center retrospective study compared radiation dose and image quality of postoperative imaging performed using conventional MSCT scanners and O-arm CBCT. MATERIALS AND METHODS: A total of 104 pediatric syndromic and non-syndromic patients who were operated on because of single- or multiple-suture craniosynostosis were included in this study. The mean volumetric CT dose index (CTDIvol) and dose-length product (DLP) values of optimized craniosynostosis CT examinations (58 MSCT and 46 CBCT) were compared. Two surgeons evaluated the subjective image quality. RESULTS: CBCT resulted in significantly lower CTDIvol (up to 14%) and DLP (up to 33%) compared to MSCT. Multi-slice CT image quality was considered superior to CBCT scans. However, all scans were considered to be of sufficient quality for diagnosis. CONCLUSION: The O-arm device allowed for an immediate postoperative CBCT examination in the operating theater using the same anesthesia induction. Radiation exposure was lower in CBCT compared to MSCT scans, thus further encouraging the use of O-arms. Cone-beam CT imaging with an O-arm is a feasible method for postoperative craniosynostosis imaging, yielding less anesthesia to patients, lower health costs and the possibility to immediately evaluate results of the surgical operation.

Comparing image quality of five breast tomosynthesis systems based on radiologists' reviews of phantom data.

BACKGROUND: Previous studies have shown differences in technical image quality between digital breast tomosynthesis (DBT) systems. However, quantitative image quality measurements may not necessarily fully reflect the clinical performance of DBT. PURPOSE: To study the subjective image quality of five DBT systems manufactured by Fujifilm, GE, Hologic, Planmed, and Siemens using phantom images. MATERIAL AND METHODS: A TOR MAM test object with polymethyl methacrylate plates was imaged on five DBT systems from different vendors. Three DBT acquisitions were performed at mean glandular doses of 1.0 mGy, 2.0 mGy, and 3.5 mGy while maintaining a constant phantom set-up. Eight DBT acquisitions with different test plate positions and phantom set-up thicknesses were performed at clinically applied dose levels. Additionally, three conventional two-dimensional mammogram images were acquired with different phantom thicknesses. Six radiologists ranked the systems based on the visibilities of the targets seen in the phantom images. RESULTS: In the DBT acquisitions performed at comparable dose levels, one system differed significantly from all other systems in microcalcification scores. When using site-specific DBT protocols, significant differences were found between the devices for microcalcification, filament, and low-contrast targets. A strong correlation was observed between the reviewer scores and radiation doses in DBT acquisitions, whereas no such correlation was observed in the 2D acquisitions. CONCLUSION: In DBT acquisitions, dose level was found to be a major factor explaining image quality differences between the systems, regardless of other acquisition parameters. Most DBT systems performed equally well at similar dose levels.

Detecting Distal Radius Fractures Using a Segmentation-Based Deep Learning Model.

Deep learning algorithms can be used to classify medical images. In distal radius fracture treatment, fracture detection and radiographic assessment of fracture displacement are critical steps. The aim of this study was to use pixel-level annotations of fractures to develop a deep learning model for precise distal radius fracture detection. We randomly divided 3785 consecutive emergency wrist radiograph examinations from six hospitals to a training set (3399 examinations) and test set (386 examinations). The training set was used to develop the deep learning model and the test set to assess its validity. The consensus of three hand surgeons was used as the gold standard for the test set. The area under the ROC curve was 0.97 (CI 0.95-0.98) and 0.95 (CI 0.92-0.98) for examinations without a cast. Fractures were identified with higher accuracy in the postero-anterior radiographs than in the lateral radiographs. Our deep learning model performed well in our multi-hospital and multi-radiograph system manufacturer settings. Thus, segmentation-based deep learning models may provide additional benefit. Further research is needed with algorithm comparison and external validation.

EFFECT OF PELVIC LEAD SHIELD ON ORGAN DOSES IN POSTERO-ANTERIOR CHEST RADIOGRAPHY.

Lead shields are commonly used in X-ray imaging to protect radiosensitive organs and to minimise patient's radiation dose. However, they might also complicate or interfere with the examination, and even decrease the diagnostic value if they are positioned incorrectly. In this study, the radiation dose effect of waist half-apron lead shield was examined via Monte Carlo simulations of postero-anterior (PA) chest radiography examinations using a female anthropomorphic phantom. Relevant organs for dose determination were lungs, breasts, liver, kidneys and uterus. The organ dose reductions varied depending on shield position and organ but were negligible for properly positioned shields. The shield that had the largest effective dose reduction (9%) was partly positioned inside the field of view, which should not be done in practice. Dose reduction was practically 0% for properly positioned shields. Therefore, the use of lead shield in the pelvic region during chest PA examinations should be discontinued.

Convolutional neural network -based phantom image scoring for mammography quality control.

BACKGROUND: Visual evaluation of phantom images is an important, but time-consuming part of mammography quality control (QC). Consistent scoring of phantom images over the device's lifetime is highly desirable. Recently, convolutional neural networks (CNNs) have been applied to a wide range of image classification problems, performing with a high accuracy. The purpose of this study was to automate mammography QC phantom scoring task by training CNN models to mimic a human reviewer. METHODS: Eight CNN variations consisting of three to ten convolutional layers were trained for detecting targets (fibres, microcalcifications and masses) in American College of Radiology (ACR) accreditation phantom images and the results were compared with human scoring. Regular and artificially degraded/improved QC phantom images from eight mammography devices were visually evaluated by one reviewer. These images were used in training the CNN models. A separate test set consisted of daily QC images from the eight devices and separately acquired images with varying dose levels. These were scored by four reviewers and considered the ground truth for CNN performance testing. RESULTS: Although hyper-parameter search space was limited, an optimal network depth after which additional layers resulted in decreased accuracy was identified. The highest scoring accuracy (95%) was achieved with the CNN consisting of six convolutional layers. The highest deviation between the CNN and the reviewers was found at lowest dose levels. No significant difference emerged between the visual reviews and CNN results except in case of smallest masses. CONCLUSION: A CNN-based automatic mammography QC phantom scoring system can score phantom images in a good agreement with human reviewers, and can therefore be of benefit in mammography QC.

Marker-controlled watershed with deep edge emphasis and optimized H-minima transform for automatic segmentation of densely cultivated 3D cell nuclei.

BACKGROUND: The segmentation of 3D cell nuclei is essential in many tasks, such as targeted molecular radiotherapies (MRT) for metastatic tumours, toxicity screening, and the observation of proliferating cells. In recent years, one popular method for automatic segmentation of nuclei has been deep learning enhanced marker-controlled watershed transform. In this method, convolutional neural networks (CNNs) have been used to create nuclei masks and markers, and the watershed algorithm for the instance segmentation. We studied whether this method could be improved for the segmentation of densely cultivated 3D nuclei via developing multiple system configurations in which we studied the effect of edge emphasizing CNNs, and optimized H-minima transform for mask and marker generation, respectively. RESULTS: The dataset used for training and evaluation consisted of twelve in vitro cultivated densely packed 3D human carcinoma cell spheroids imaged using a confocal microscope. With this dataset, the evaluation was performed using a cross-validation scheme. In addition, four independent datasets were used for evaluation. The datasets were resampled near isotropic for our experiments. The baseline deep learning enhanced marker-controlled watershed obtained an average of 0.69 Panoptic Quality (PQ) and 0.66 Aggregated Jaccard Index (AJI) over the twelve spheroids. Using a system configuration, which was otherwise the same but used 3D-based edge emphasizing CNNs and optimized H-minima transform, the scores increased to 0.76 and 0.77, respectively. When using the independent datasets for evaluation, the best performing system configuration was shown to outperform or equal the baseline and a set of well-known cell segmentation approaches. CONCLUSIONS: The use of edge emphasizing U-Nets and optimized H-minima transform can improve the marker-controlled watershed transform for segmentation of densely cultivated 3D cell nuclei. A novel dataset of twelve spheroids was introduced to the public.

The impact of vertical off-centering on image noise and breast dose in chest CT with organ-based tube current modulation: A phantom study.

PURPOSE: To determine the effects of patient vertical off-centering when using organ-based tube current modulation (OBTCM) in chest computed tomography (CT) with focus on breast dose. MATERIALS AND METHODS: An anthropomorphic adult female phantom with two different breast attachment sizes was scanned on GE Revolution EVO and Siemens Definition Edge CT systems using clinical chest CT protocols and anterior-to-posterior scouts. Scans with and without OBTCM were performed at different table heights (GE: centered, ±6 cm, and ± 3 cm; Siemens: centered, -6 cm, and ± 3 cm). The dose effects were studied with metal-oxidesemiconductor field-effect transistor dosimeters with complementary Monte Carlo simulations to determine full dose maps. Changes in image noise were studied using standard deviations of subtraction images from repeated acquisitions without dosimeters. RESULTS: Patient off-centering affected both the behavior of the normal tube current modulation as well as the extent of the OBTCM. Generally, both OBTCM techniques provided a substantial decrease in the breast doses (up to 30% local decrease). Lateral breast regions may, however, in some cases receive higher doses when OBTCM is enabled. This effect becomes more prominent when the patient is centered too low in the CT gantry. Changes in noise roughly followed the expected inverse of the change in dose. CONCLUSIONS: Patient off-centering was shown to affect the outcome of OBTCM in chest CT examination, and on some occasions, resulting in higher exposure. The use of modern dose optimization tools such as OBTCM emphasizes the importance of proper centering when preparing patients to CT scans.

Characterizing geometric distortions of 3D sequences in clinical head MRI.

OBJECTIVE: Phantoms are often used to estimate the geometric accuracy in magnetic resonance imaging (MRI). However, the distortions may differ between anatomical and phantom images. This study aimed to investigate the applicability of a phantom-based and a test-subject-based method in evaluating geometric distortion present in clinical head-imaging sequences. MATERIALS AND METHODS: We imaged a 3D-printed phantom and test subjects with two MRI scanners using two clinical head-imaging 3D sequences with varying patient-table positions and receiver bandwidths. The geometric distortions were evaluated through nonrigid registrations: the displaced acquisitions were compared against the ideal isocenter positioning, and the varied bandwidth volumes against the volume with the highest bandwidth. The phantom acquisitions were also registered to a computed tomography scan. RESULTS: Geometric distortion magnitudes increased with larger table displacements and were in good agreement between the phantom and test-subject acquisitions. The effect of increased distortions with decreasing receiver bandwidth was more prominent for test-subject acquisitions. CONCLUSION: Presented results emphasize the sensitivity of the geometric accuracy to positioning and imaging parameters. Phantom limitations may become an issue with some sequence types, encouraging the use of anatomical images for evaluating the geometric accuracy.

Automatic head computed tomography image noise quantification with deep learning.

PURPOSE: Computed tomography (CT) image noise is usually determined by standard deviation (SD) of pixel values from uniform image regions. This study investigates how deep learning (DL) could be applied in head CT image noise estimation. METHODS: Two approaches were investigated for noise image estimation of a single acquisition image: direct noise image estimation using supervised DnCNN convolutional neural network (CNN) architecture, and subtraction of a denoised image estimated with denoising UNet-CNN experimented with supervised and unsupervised noise2noise training approaches. Noise was assessed with local SD maps using 3D- and 2D-CNN architectures. Anthropomorphic phantom CT image dataset (N = 9 scans, 3 repetitions) was used for DL-model comparisons. Mean square error (MSE) and mean absolute percentage errors (MAPE) of SD values were determined using the SD values of subtraction images as ground truth. Open-source clinical head CT low-dose dataset (Ntrain = 37, Ntest = 10 subjects) were used to demonstrate DL applicability in noise estimation from manually labeled uniform regions and in automated noise and contrast assessment. RESULTS: The direct SD estimation using 3D-CNN was the most accurate assessment method when comparing in phantom dataset (MAPE = 15.5%, MSE = 6.3HU). Unsupervised noise2noise approach provided only slightly inferior results (MAPE = 20.2%, MSE = 13.7HU). 2DCNN and unsupervised UNet models provided the smallest MSE on clinical labeled uniform regions. CONCLUSIONS: DL-based clinical image assessment is feasible and provides acceptable accuracy as compared to true image noise. Noise2noise approach may be feasible in clinical use where no ground truth data is available. Noise estimation combined with tissue segmentation may enable more comprehensive image quality characterization.

Automatic CT Angiography Lesion Segmentation Compared to CT Perfusion in Ischemic Stroke Detection: a Feasibility Study.

In stroke imaging, CT angiography (CTA) is used for detecting arterial occlusions. These images could also provide information on the extent of ischemia. The study aim was to develop and evaluate a convolutional neural network (CNN)-based algorithm for detecting and segmenting acute ischemic lesions from CTA images of patients with suspected middle cerebral artery stroke. These results were compared to volumes reported by widely used CT perfusion-based RAPID software (IschemaView). A 42-layer-deep CNN was trained on 50 CTA volumes with manually delineated targets. The lower bound for predicted lesion size to reliably discern stroke from false positives was estimated. The severity of false positives and false negatives was reviewed visually to assess the clinical applicability and to further guide the method development. The CNN model corresponded to the manual segmentations with voxel-wise sensitivity 0.54 (95% confidence interval: 0.44-0.63), precision 0.69 (0.60-0.76), and Sørensen-Dice coefficient 0.61 (0.52-0.67). Stroke/nonstroke differentiation accuracy 0.88 (0.81-0.94) was achieved when only considering the predicted lesion size (i.e., regardless of location). By visual estimation, 46% of cases showed some false findings, such as CNN highlighting chronic periventricular white matter changes or beam hardening artifacts, but only in 9% the errors were severe, translating to 0.91 accuracy. The CNN model had a moderately strong correlation to RAPID-reported Tmax > 10 s volumes (Pearson's r = 0.76 (0.58-0.86)). The results suggest that detecting anterior circulation ischemic strokes from CTA using a CNN-based algorithm can be feasible when accompanied with physiological knowledge to rule out false positives.

Computed tomography angiography-based deep learning method for treatment selection and infarct volume prediction in anterior cerebral circulation large vessel occlusion.

BACKGROUND: Computed tomography perfusion (CTP) is the mainstay to determine possible eligibility for endovascular thrombectomy (EVT), but there is still a need for alternative methods in patient triage. PURPOSE: To study the ability of a computed tomography angiography (CTA)-based convolutional neural network (CNN) method in predicting final infarct volume in patients with large vessel occlusion successfully treated with endovascular therapy. MATERIALS AND METHODS: The accuracy of the CTA source image-based CNN in final infarct volume prediction was evaluated against follow-up CT or MR imaging in 89 patients with anterior circulation ischemic stroke successfully treated with EVT as defined by Thrombolysis in Cerebral Infarction category 2b or 3 using Pearson correlation coefficients and intraclass correlation coefficients. Convolutional neural network performance was also compared to a commercially available CTP-based software (RAPID, iSchemaView). RESULTS: A correlation with final infarct volumes was found for both CNN and CTP-RAPID in patients presenting 6-24 h from symptom onset or last known well, with r = 0.67 (p < 0.001) and r = 0.82 (p < 0.001), respectively. Correlations with final infarct volumes in the early time window (0-6 h) were r = 0.43 (p = 0.002) for the CNN and r = 0.58 (p < 0.001) for CTP-RAPID. Compared to CTP-RAPID predictions, CNN estimated eligibility for thrombectomy according to ischemic core size in the late time window with a sensitivity of 0.38 and specificity of 0.89. CONCLUSION: A CTA-based CNN method had moderate correlation with final infarct volumes in the late time window in patients successfully treated with EVT.

Performance of a 3D convolutional neural network in the detection of hypoperfusion at CT pulmonary angiography in patients with chronic pulmonary embolism: a feasibility study.

BACKGROUND: Chronic pulmonary embolism (CPE) is a life-threatening disease easily misdiagnosed on computed tomography. We investigated a three-dimensional convolutional neural network (CNN) algorithm for detecting hypoperfusion in CPE from computed tomography pulmonary angiography (CTPA). METHODS: Preoperative CTPA of 25 patients with CPE and 25 without pulmonary embolism were selected. We applied a 48%-12%-40% training-validation-testing split (12 positive and 12 negative CTPA volumes for training, 3 positives and 3 negatives for validation, 10 positives and 10 negatives for testing). The median number of axial images per CTPA was 335 (min-max, 111-570). Expert manual segmentations were used as training and testing targets. The CNN output was compared to a method in which a Hounsfield unit (HU) threshold was used to detect hypoperfusion. Receiver operating characteristic area under the curve (AUC) and Matthew correlation coefficient (MCC) were calculated with their 95% confidence interval (CI). RESULTS: The predicted segmentations of CNN showed AUC 0.87 (95% CI 0.82-0.91), those of HU-threshold method 0.79 (95% CI 0.74-0.84). The optimal global threshold values were CNN output probability ≥ 0.37 and ≤ -850 HU. Using these values, MCC was 0.46 (95% CI 0.29-0.59) for CNN and 0.35 (95% CI 0.18-0.48) for HU-threshold method (average difference in MCC in the bootstrap samples 0.11 (95% CI 0.05-0.16). A high CNN prediction probability was a strong predictor of CPE. CONCLUSIONS: We proposed a deep learning method for detecting hypoperfusion in CPE from CTPA. This model may help evaluating disease extent and supporting treatment planning.

Evaluation of a CTA-based convolutional neural network for infarct volume prediction in anterior cerebral circulation ischaemic stroke.

BACKGROUND: Computed tomography angiography (CTA) imaging is needed in current guideline-based stroke diagnosis, and infarct core size is one factor in guiding treatment decisions. We studied the efficacy of a convolutional neural network (CNN) in final infarct volume prediction from CTA and compared the results to a CT perfusion (CTP)-based commercially available software (RAPID, iSchemaView). METHODS: We retrospectively selected 83 consecutive stroke cases treated with thrombolytic therapy or receiving supportive care that presented to Helsinki University Hospital between January 2018 and July 2019. We compared CNN-derived ischaemic lesion volumes to final infarct volumes that were manually segmented from follow-up CT and to CTP-RAPID ischaemic core volumes. RESULTS: An overall correlation of r = 0.83 was found between CNN outputs and final infarct volumes. The strongest correlation was found in a subgroup of patients that presented more than 9 h of symptom onset (r = 0.90). A good correlation was found between the CNN outputs and CTP-RAPID ischaemic core volumes (r = 0.89) and the CNN was able to classify patients for thrombolytic therapy or supportive care with a 1.00 sensitivity and 0.94 specificity. CONCLUSIONS: A CTA-based CNN software can provide good infarct core volume estimates as observed in follow-up imaging studies. CNN-derived infarct volumes had a good correlation to CTP-RAPID ischaemic core volumes.

Automating chest radiograph imaging quality control.

PURPOSE: To automate diagnostic chest radiograph imaging quality control (lung inclusion at all four edges, patient rotation, and correct inspiration) using convolutional neural network models. METHODS: The data comprised of 2589 postero-anterior chest radiographs imaged in a standing position, which were divided into train, validation, and test sets. We increased the number of images for the inclusion by cropping appropriate images, and for the inclusion and the rotation by flipping the images horizontally. The image histograms were equalized, and the images were resized to a 512 × 512 resolution. We trained six convolutional neural networks models to detect the image quality features using manual image annotations as training targets. Additionally, we studied the inter-observer variability of the image annotation. RESULTS: The convolutional neural networks' areas under the receiver operating characteristic curve were >0.88 for the inclusions, and >0.70 and >0.79 for the rotation and the inspiration, respectively. The inter-observer agreement between two human annotators for the assessed image-quality features were: 92%, 90%, 82%, and 88% for the inclusion at patient's left, patient's right, cranial, and caudal edges, and 78% and 89% for the rotation and inspiration, respectively. Higher inter-observer agreement was related to a smaller variance in the network confidence. CONCLUSIONS: The developed models provide automated tools for the quality control in a radiological department. Additionally, the convolutional neural networks could be used to obtain immediate feedback of the chest radiograph image quality, which could serve as an educational instrument.

Inter- and intra-scanner variations in four magnetic resonance imaging image quality parameters.

Purpose: In addition to less frequent and more comprehensive tests, quality assurance (QA) protocol for a magnetic resonance imaging (MRI) scanner may include cursory daily or weekly phantom checks to verify equipment constancy. With an automatic image analysis workflow, the daily QA images can be further used to study scanner baseline performance and both long- and short-term variations in image quality. With known baselines and variation profiles, automatic error detection can be employed. Approach: Four image quality parameters were followed for 17 MRI scanners over six months: signal-to-noise ratio (SNR), image intensity uniformity, ghosting artifact, and geometrical distortions. Baselines and normal variations were determined. An automatic detection of abnormal QA images was compared with image deviations visually detected by human observers. Results: There were significant inter-scanner differences in the QA parameters. In some cases, the results exceeded commonly accepted tolerances. Scanner field strengths, or a unit being stationary versus mobile, did not have a clear relationship with the QA results. Conclusions: The variations and baseline levels of image QA parameters can differ significantly between MRI scanners. Scanner specific error thresholds based on parameter means and standard deviations are a viable option for detecting abnormal QA images.

Measuring geometric accuracy in magnetic resonance imaging with 3D-printed phantom and nonrigid image registration.

OBJECTIVE: We aimed to develop a vendor-neutral and interaction-free quality assurance protocol for measuring geometric accuracy of head and brain magnetic resonance (MR) images. We investigated the usability of nonrigid image registration in the analysis and looked for the optimal registration parameters. MATERIALS AND METHODS: We constructed a 3D-printed phantom and imaged it with 12 MR scanners using clinical sequences. We registered a geometric-ground-truth computed tomography (CT) acquisition to the MR images using an open-source nonrigid-registration-toolbox with varying parameters. We applied the transforms to a set of control points in the CT image and compared their locations to the corresponding visually verified reference points in the MR images. RESULTS: With optimized registration parameters, the mean difference (and standard deviation) of control point locations when compared to the reference method was (0.17 ± 0.02) mm for the 12 studied scanners. The maximum displacements varied from 0.50 to 1.35 mm or 0.89 to 2.30 mm, with vendors' distortion correction on or off, respectively. DISCUSSION: Using nonrigid CT-MR registration can provide a robust and relatively test-object-agnostic method for estimating the intra- and inter-scanner variations of the geometric distortions.