Computed Tomography Slices Research Articles

Artificial intelligence for detection of intracranial haemorrhage on head computed tomography scans: diagnostic accuracy in Hong Kong.

The use of artificial intelligence (AI) to identify acute intracranial haemorrhage (ICH) on computed tomography (CT) scans may facilitate initial imaging interpretation in the accident and emergency department. However, AI model construction requires a large amount of annotated data for training, and validation with real-world data has been limited. We developed an algorithm using an open-access dataset of CT slices, then assessed its utility in clinical practice by validating its performance on CT scans from our institution. Using a publicly available international dataset of >750 000 expert-labelled CT slices, we developed an AI model which determines ICH probability for each CT scan and nominates five potential ICH-positive CT slices for review. We validated the model using retrospective data from 1372 non-contrast head CT scans (84 [6.1%] with ICH) collected at our institution. The model achieved an area under the curve of 0.842 (95% confidence interval=0.791-0.894; P<0.001) for scan-based detection of ICH. A pre-specified probability threshold of ≥50% for the presence of ICH yielded 78.6% accuracy, 73% sensitivity, 79% specificity, 18.6% positive predictive value, and 97.8% negative predictive value. There were 62 true-positive scans and 22 false-negative scans, which could be reduced to six false-negative scans by manual review of model-nominated CT slices. Our model exhibited good accuracy in the CT scan-based detection of ICH, considering the low prevalence of ICH in Hong Kong. Model refinement to allow direct localisation of ICH will facilitate the use of AI solutions in clinical practice.

Consensus guide on CT-based prediction of stopping-power ratio using a Hounsfield look-up table for proton therapy

Background and purposeStudies have shown large variations in stopping-power ratio (SPR) prediction from computed tomography (CT) across European proton centres. To standardise this process, a step-by-step guide on specifying a Hounsfield look-up table (HLUT) is presented here. Materials and methodsThe HLUT specification process is divided into six steps: Phantom setup, CT acquisition, CT number extraction, SPR determination, HLUT specification, and HLUT validation. Appropriate CT phantoms have a head- and body-sized part, with tissue-equivalent inserts in regard to X-ray and proton interactions. CT numbers are extracted from a region-of-interest covering the inner 70% of each insert in-plane and several axial CT slices in scan direction. For optimal HLUT specification, the SPR of phantom inserts is measured in a proton beam and the SPR of tabulated human tissues is computed stoichiometrically at 100 MeV. Including both phantom inserts and tabulated human tissues increases HLUT stability. Piecewise linear regressions are performed between CT numbers and SPRs for four tissue groups (lung, adipose, soft tissue, and bone) and then connected with straight lines. Finally, a thorough but simple validation is performed. ResultsThe best practices and individual challenges are explained comprehensively for each step. A well-defined strategy for specifying the connection points between the individual line segments of the HLUT is presented. The guide was tested exemplarily on three CT scanners from different vendors, proving its feasibility. ConclusionThe presented step-by-step guide for CT-based HLUT specification with recommendations and examples can contribute to reduce inter-centre variations in SPR prediction.

Tensor-RT-Based Transfer Learning Model for Lung Cancer Classification.

Cancer is a leading cause of death across the globe, in which lung cancer constitutes the maximum mortality rate. Early diagnosis through computed tomography scan imaging helps to identify the stages of lung cancer. Several deep learning-based classification methods have been employed for developing automatic systems for the diagnosis and detection of computed tomography scan lung slices. However, the diagnosis based on nodule detection is a challenging task as it requires manual annotation of nodule regions. Also, these computer-aided systems have yet not achieved the desired performance in real-time lung cancer classification. In the present paper, a high-speed real-time transfer learning-based framework is proposed for the classification of computed tomography lung cancer slices into benign and malignant. The proposed framework comprises of three modules: (i) pre-processing and segmentation of lung images using K-means clustering based on cosine distance and morphological operations; (ii) tuning and regularization of the proposed model named as weighted VGG deep network (WVDN); (iii) model inference in Nvidia tensor-RT during post-processing for the deployment in real-time applications. In this study, two pre-trained CNN models were experimented and compared with the proposed model. All the models have been trained on 19,419 computed tomography scan lung slices, which were obtained from the publicly available Lung Image Database Consortium and Image Database Resource Initiative dataset. The proposed model achieved the best classification metric, an accuracy of 0.932, precision, recall, an F1 score of 0.93, and Cohen's kappa score of 0.85. A statistical evaluation has also been performed on the classification parameters and achieved a p-value <0.0001 for the proposed model. The quantitative and statistical results validate the improved performance of the proposed model as compared to state-of-the-art methods. The proposed framework is based on complete computed tomography slices rather than the marked annotations and may help in improving clinical diagnosis.

Reference values for low muscle mass and myosteatosis using tomographic muscle measurements in living kidney donors

Low muscle mass and myosteatosis are associated with poor clinical outcomes. Computed tomography (CT) imaging is an objective method for muscle mass and quality assessment; however consensus on cut-off values is lacking. This study assessed age-, sex-, and body mass index (BMI)-specific reference values of skeletal muscle parameters and correlated muscle mass with 24-h urinary creatinine excretion (24-h UCE). In total, 960 healthy subjects were included in this study. Muscle mass and quality were determined using axial CT slices at the vertebral level L3. The muscle area was indexed for height (skeletal muscle index [SMI]). The mean age was 53 ± 11 years, and 50% were male. The SMI reference values for low muscle mass in males were 38.8 cm2/m2 (20–29 years), 39.2 (30–39 years), 39.9 (40–49 years), 39.0 (50–59 years), 37.0 (60–69 years), and 36.8 (70–79 years). For females, these reference values were 37.5 cm2/m2 (20–29 years), 35.5 (30–39 years), 32.8 (40–49 years), 33.2 (50–59 years), 31.2 (60–69 years), and 31.5 (70–79 years). 24-h UCE and SMI were significantly correlated (r = 0.54, p < 0.001) without bias between the two methods of assessing muscle mass. This study provides age-, sex-, and BMI-specific reference values for skeletal muscle parameters that will support clinical decision making.

Automatic measurement of slice thickness in CT images of a Siemens phantom

This study aims to develop a program in Python language for automatic measurement of slice thickness in computed tomography (CT) images of a Siemens phantom with different values of slice thickness, field of view (FOV), and pitch. A Siemens phantom was scanned using a Siemens 64-slice Somatom Perspective CT scanner with various slice thicknesses (i.e. 2, 4, 6, 8, and 10 mm), FOVs (i.e. 220, 260, and 300 mm), and pitch (i.e. 0.7, 0.9, and 1). Automatic measurement of slice thickness was performed by segmenting the ramp insert in the image and detecting angles of the ramp insert using the Hough transform. The resulting angles were subsequently used to rotate the image. Profiles of pixel along the ramp insert were made from the rotated images, and the slice thickness was calculated by determining the full-width at half maximum (FWHM) of the profiles. The product of the FWHM in pixels and the pixel size was corrected by the tangent of the ramp insert (i.e., 23°) to obtain the measured slice thickness. The results of the automatic measurements were compared with manual measurements carried out using a MicroDicom Viewer. The differences between the automatic and manual measurements at all slice thicknesses were less than 0.30 mm. The automatic and manual measurements had high linear correlations. For variations of the FOV and pitch, the differences between the automatic and manual measurement were less than 0.16 mm. The automatic and manual measurements were significantly different (p-value < 0.05) for slice thickness variation. In addition, the automatic and manual measurements were not significantly different (p-value > 0.05) for variations of FOV and pitch.

Longitudinal Variability Analysis on Low-dose Abdominal CT with Deep Learning-based Segmentation.

Metabolic health is increasingly implicated as a risk factor across conditions from cardiology to neurology, and efficiency assessment of body composition is critical to quantitatively characterizing these relationships. 2D low dose single slice computed tomography (CT) provides a high resolution, quantitative tissue map, albeit with a limited field of view. Although numerous potential analyses have been proposed in quantifying image context, there has been no comprehensive study for low-dose single slice CT longitudinal variability with automated segmentation. We studied a total of 1816 slices from 1469 subjects of Baltimore Longitudinal Study on Aging (BLSA) abdominal dataset using supervised deep learning-based segmentation and unsupervised clustering method. 300 out of 1469 subjects that have two year gap in their first two scans were pick out to evaluate longitudinal variability with measurements including intraclass correlation coefficient (ICC) and coefficient of variation (CV) in terms of tissues/organs size and mean intensity. We showed that our segmentation methods are stable in longitudinal settings with Dice ranged from 0.821 to 0.962 for thirteen target abdominal tissues structures. We observed high variability in most organ with ICC<0.5, low variability in the area of muscle, abdominal wall, fat and body mask with average ICC≥0.8. We found that the variability in organ is highly related to the cross-sectional position of the 2D slice. Our efforts pave quantitative exploration and quality control to reduce uncertainties in longitudinal analysis.

Diagnostic accuracy for coronary artery disease by computed tomography angiography as compared to conventional invasive angiography

Background: Emerging as a major cause of death and disability in the developing countries is coronary artery disease (CAD). Traditional coronary angiography has traditionally been the preferred diagnostic procedure for CAD. Non-invasive evaluation of coronary artery disease (CAD) using computed tomography (CT) coronary angiography shows promise. This observational study's goal is to assess the relative efficacy of noninvasive multi-section computed tomography (MSCT) coronary angiography vs invasive selective coronary angiography in detecting coronary stenosis.Method and patient: This observational study carried out in the Azadi teaching Hospital/Kirkuk between November 2021 to March 2022. Patients who suspected to have coronary artery disease underwent both CTCA and invasive coronary angiography in different centers in our country comparing and analyzing their results.Result: In total, 92 participants were enrolled in the trial. Patients had a mean age of 57.207.3 years, and men made up the vast majority of those who appeared (67.3 percent). Chest pain was the most common presenting complaint (94 percent). Patients exhibited a prevalence of 84% for HTN, 55% for hyperlipidaemia, 58% for diabetes, and 4% for ischemic heart disease. In addition, 50% of the patients are either current or former smokers. Half of the patients received 64-slice CT scans, nearly half had 128-slice scans, and 3.2% got 256-slice scans. CCTA's sensitivity was 100%, and its specificity was 66.7%. Positive predictive value for CCTA was 98.8 percent (meaning it correctly identified positive cases in 98.8 percent of patients) and negative predictive value was 100 percent, for a total diagnostic accuracy of 98.9 percent (which mean CCTA can exclude all negative cases and did not miss any case). When looking at individual arteries, the LAD artery was shown to be the most accurate, while the LCX artery was the least accurate (95.6 percent and 78.2 percent respectively). In terms of sensitivity, LAD performed at a 99.1% rate, while LMS performed at a 0% rate (33.3 percent). RCA had the most specificity (95.2 percent), whereas LAD had the lowest (88.8 percent) (40 percent). In order to determine what factors, influence CCTA positivity, we performed a binary logistic regression analysis for positive CCTA (negative as reference) with all variables from the studies. The results showed that IHD, lesions in the LAD artery, lesions in the LCX artery, and CT slice thickness (128 and 256 slices) all have a negative effect on CCTA (p0.05).Conclusions: When compared to traditional angiography, multi-slice computed tomography has demonstrated superior sensitivity and specificity for the identification of coronary artery disease.

Role of Stem Cells in Transport Distraction Osteogenesis for Reconstruction of Mandibular Defects after Segmental Mandibulectomy

Purpose: to restore the continuity and functions of the mandible after segmental mandibulectomy for mandibular defect reconstruction by using a technique of transport distraction osteogenesis with stem cells injection. Subjects and methods: Clinical study was conducted on eight patients suffering from mandibular deficiency after tumor resection. All steps of manufacturing of the transport distractor, tumor resection and segmental mandibulectomy of transport disc were guided by Stereolithographic (STL) model. Stem cells sample was prepared to be injected along the soft callus at the end of the activation period. Patients were followed up postoperative clinically every week and radiographically by Panoramic x-ray and computed tomography (CT) at immediate postoperatively, then by panoramic x-ray monthly along the periods of distraction and finally by multi slice CT at the end of the consolidation to evaluate the time consuming for treatment. Results: seven cases out of eight cases recorded excellent distraction regenerate and excellent soft tissue coverage and one case out of these eight cases in the control group failed at the first week of activation. Multi slice CT showed a significantly higher value in the mean of the regenerated bone density at the end of the consolidation (1163.76±111.38) in the study group than the control group (747.67±268.6). Conclusion: From our study, stem cells injection in the distraction regenerate for reconstruction of mandibular defect cases achieved short consolidation period and high regenerated bone density.

Role of Stem Cells in Transport Distraction Osteogenesis for Reconstruction of Mandibular Defects after Segmental Mandibulectomy

Purpose: to restore the continuity and functions of the mandible after segmental mandibulectomy for mandibular defect reconstruction by using a technique of transport distraction osteogenesis with stem cells injection. Subjects and methods: Clinical study was conducted on eight patients suffering from mandibular deficiency after tumor resection. All steps of manufacturing of the transport distractor, tumor resection and segmental mandibulectomy of transport disc were guided by Stereolithographic (STL) model. Stem cells sample was prepared to be injected along the soft callus at the end of the activation period. Patients were followed up postoperative clinically every week and radiographically by Panoramic x-ray and computed tomography (CT) at immediate postoperatively, then by panoramic x-ray monthly along the periods of distraction and finally by multi slice CT at the end of the consolidation to evaluate the time consuming for treatment. Results: seven cases out of eight cases recorded excellent distraction regenerate and excellent soft tissue coverage and one case out of these eight cases in the control group failed at the first week of activation. Multi slice CT showed a significantly higher value in the mean of the regenerated bone density at the end of the consolidation (1163.76±111.38) in the study group than the control group (747.67±268.6). Conclusion: From our study, stem cells injection in the distraction regenerate for reconstruction of mandibular defect cases achieved short consolidation period and high regenerated bone density.

Role of ultra-low dose computed tomography (CT) in COVID-19 suspected patients: A cross sectional study

Background: This study aims to evaluate the accuracy of ultra-low-dose CT in detecting COVID 19 related chest findings in COVID 19 suspected patients. Materials and Methods: 106 COVID 19 infection suspected patients who are referred for HRCT chest screening, were evaluated by CT using 128 slice CT scanner (Wipro GE Optima 660). Ultra low dose CT with kv 120 & mA 10 was used. The Ground glass opacities and consolidations will be assessed in every lobe and the severity and grading will be done. CT severity score - in this both lungs are divided into five lobes, and each lobe was assessed individually. Each lobe could be awarded a CT score from 0 to 5, depending on the percentage of the involved lobe. All the values of each lobe is added and graded as mild (0-8), moderate (8-15) and severe (15-25). Then we will correlate the ultra low dose CT findings with RT-PCR results. Results: Among the study population with RT-PCR positive, 67 (85.9%) of them ground glass opacity was present and 11 (14.1%) of them absent. The difference in proportion of Ground glass opacity between RT-PCR was statistically significant (p value

Will improved neuroradiology facilities debunk the reported rarity of intracranial aneurysms in Sub-Saharan Africa?

Intracranial aneurysms (IAN) are rare in the Sub-Saharan Africa unlike other parts of the world. The debate is whether the low frequency might be apparent because of the scarcity of advanced neuroimaging services, or real. This study investigated if improved imaging facilities would debunk the rarity of IAN in our subregion. This is a retrospective cohort study of prospectively recorded data of patients with subarachnoid hemorrhage (SAH) and IAN managed over 19 years (2003-2021), at the study center with a catchment population of over 47 million. The center witnessed progressive improvements in neuroimaging facilities: 2-Slice, 8-slice, and 64-slice computed tomography (CT) and 0.35T, 1.5T magnetic resonance imaging (MRI) during the period. There were 241 cases of SAH, but only 166 aneurysms were confirmed in 158 patients. Between 2003 and 2008, only 27 IAN patients (4.5 IAN/year) were diagnosed. After introduction of CT angiography/magnetic resonance angiography MRA using 8-slice CT/0.35T magnetic resonance imaging (MRI), between 2009 and 2014, the frequency of IAN increased to 8/year. Between 2015 and 2018 after installation of a 64-slice CT in 2014, the IAN remained the same (8/year). MRI 1.5T was added in 2018, the frequency doubled to 17 cases/year. The females were more (67.7%), the mean age was 46.3 years, but peak incidence was the sixth decade. Internal carotid artery aneurysms including posterior communicating artery were the most common (43%) followed by ACA with anterior communicating artery (24%) and middle cerebral artery (20%). Multiple aneurysms were seen in ten patients. Improved neuroimaging between 2003 and 2021 did not debunk the rarity of IAN in our region.

Simulation of X-ray projections on GPU: Benchmarking gVirtualXray with clinically realistic phantoms

Background and objectives: This study provides a quantitative comparison of images created using gVirtualXray (gVXR) to both Monte Carlo (MC) and real images of clinically realistic phantoms. gVirtualXray is an open-source framework that relies on the Beer-Lambert law to simulate X-ray images in realtime on a graphics processor unit (GPU) using triangular meshes.Methods: Images are generated with gVirtualXray and compared with a corresponding ground truth image of an anthropomorphic phantom: (i) an X-ray projection generated using a Monte Carlo simulation code, (ii) real digitally reconstructed radiographs (DRRs), (iii) computed tomography (CT) slices, and (iv) a real radiograph acquired with a clinical X-ray imaging system. When real images are involved, the simulations are used in an image registration framework so that the two images are aligned.Results: The mean absolute percentage error (MAPE) between the images simulated with gVirtualXray and MC is 3.12%, the zero-mean normalised cross-correlation (ZNCC) is 99.96% and the structural similarity index (SSIM) is 0.99. The run-time is 10 days for MC and 23 ms with gVirtualXray. Images simulated using surface models segmented from a CT scan of the Lungman chest phantom were similar to (i) DRRs computed from the CT volume and (ii) an actual digital radiograph. CT slices reconstructed from images simulated with gVirtualXray were comparable to the corresponding slices of the original CT volume.Conclusions: When scattering can be ignored, accurate images that would take days using MC can be generated in milliseconds with gVirtualXray. This speed of execution enables the use of repetitive simulations with varying parameters, e.g. to generate training data for a deep-learning algorithm, and to minimise the objective function of an optimisation problem in image registration. The use of surface models enables the combination of X-ray simulation with real-time soft-tissue deformation and character animation, which can be deployed in virtual reality applications.

Feasibility of Simple Evaluation of Coronary Artery Calcium Using a Number of Chest Computed Tomography Slices: A Multicenter Study

Objective: We aimed to evaluate the visual measurements of coronary artery calcium (CAC) on nonelectrocardiogram (ECG)-gated chest computed tomography (CT) using a simple scoring method that involves counting the number of CT slices containing CAC. Materials and Methods: We analyzed 163 participants who underwent both coronary and chest CT examinations at six centers within 3 months. Agatston scores were calculated on standard ECG-gated scans and classified as none (0), mild (1-99), moderate (100-400), or severe (>400). Next, chest CT images were reconstructed to standard 5.0 mm axial slices. Then, CAC on chest CT scans was measured using two methods: the Weston score (sum of the assigned score of each vessel, range: 0-12) and number of slices showing CAC (Ca-slice#). Results: When the Weston score and Ca-slice# were divided into four levels according to the optimal divisional levels corresponding to the Agatston score classes, good agreements with the 4-grade Agatston score were observed (kappa value=0.610 and 0.794, respectively). The sensitivity and specificity of Ca-slice# ≥9 to identify severe Agatston scores of >400 were 86% and 96%, respectively. Conclusion: The Ca-slice#, a simple scoring method using chest CT scans, was in good agreement with the ECG-gated Agatston score.

Classification of metastatic hepatic carcinoma and hepatocellular carcinoma lesions using contrast-enhanced CT based on EI-CNNet.

For hepatocellular carcinoma and metastatic hepatic carcinoma, imaging is one of the main diagnostic methods. In clinical practice, diagnosis mainly relied on experienced imaging physicians, which was inefficient and cannot met the demand for rapid and accurate diagnosis. Therefore, how to efficiently and accurately classify the two types of liver cancer based on imaging is an urgent problem to be solved at present. The purpose of this study was to use the deep learning classification model to help radiologists classify the single metastatic hepatic carcinoma and hepatocellular carcinoma based on the enhanced features of enhanced CT (Computer Tomography) portal phase images of the liver site. In this retrospective study, 52 patients with metastatic hepatic carcinoma and 50 patients with hepatocellular carcinoma were among the patients who underwent preoperative enhanced CT examinations from 2017-2020. A total of 565 CT slices from these patients were used to train and validate the classification network (EI-CNNet, training/validation: 452/113). First, the EI block was used to extract edge information from CT slices to enrich fine-grained information and classify them. Then, ROC (Receiver Operating Characteristic) curve was used to evaluate the performance, accuracy, and recall of the EI-CNNet. Finally, the classification results of EI-CNNet were compared with popular classification models. By utilizing 80% data for model training and 20% data for model validation, the average accuracy of this experiment was 98.2% ± 0.62 (mean ± standard deviation (SD)), the recall rate was 97.23%±2.77, the precision rate was 98.02%±2.07, the network parameters were 11.83 MB, and the validation time was 9.83s/sample. The classification accuracy was improved by 20.98% compared to the base CNN network and the validation time was10.38s/sample. Compared with other classification networks, the InceptionV3 network showed improved classification results, but the number of parameters was increased and the validation time was 33s/sample, and the classification accuracy was improved by 6.51% using this method. EI-CNNet demonstrated promised diagnostic performance and has potential to reduce the workload of radiologists and may help distinguish whether the tumor is primary or metastatic in time; otherwise, it may be missed or misjudged. This article is protected by copyright. All rights reserved.

Clinical commissioning of the first point-of-care spectral photon-counting CT for the upper extremities.

Acceptance testing and quality assurance (QA) of computed tomography (CT) scans are of great importance to ensure the appropriate performance of the systems. However, current standards and guidelines do not include a dedicated QA program for spectral photon-counting CT (SPCCT), nor adapted tolerancelevels. To evaluate the technical performance, in terms of image quality and radiation dose, of the first point-of-care SPCCT for the upper extremities (MARS Extremity 5X120, MARS Bioimaging Ltd., Christchurch, New Zealand) and to establish a comprehensive QAprogram. The specific dimensions of the scanner with a 125mm diameter gantry and a small voxel size of 0.1×0.1×0.1mm3 require the use of suitable phantoms and evaluation techniques. Indicators such as CT number accuracy, image noise, uniformity, and slice thickness were assessed to characterize the image quality. The in-plane and longitudinal spatial resolutions were evaluated by means of the modulation transfer function (MTF). Noise power spectra (NPS) were calculated to further evaluate the image noise. Material identification capabilities were assessed using clinically relevant high-Z materials (iodine, gold, gadolinium, and calcium). A 100-mm diameter CTDI-like phantom was used to measure the dose indices. A complete radiation survey was carried out to measure the radiation exposure at different points around thescanner. The proposed QA program is based on international and local recommendations as well as practical experience. It includes standardised CT tests and SPCCT-specific methods. Additional methodologies to further assess the system performance are also presented. Tolerance levels are discussed and revised when appropriate. Both in-plane and longitudinal high spatial resolutions were evidenced by the MTF measurements with 1.8lp· mm-1 and 5.0lp· mm-1 at 10%, respectively. The calculated effective slice thickness ranged between 0.15and 0.16mm for the five energy bins and for a reconstructed voxel size of 0.1×0.1×0.1mm3 . Reference values of the linear attenuation coefficient of water have been calculated and used to assess the CT number uniformity of water. Evaluation of the CT number accuracy and stability of various clinically relevant materials showed excellent spectral correlation and linearity between HU values and concentrations (r2 > 0.99). The NPS showed less noise correlation between slices than within transverse slice, as well as a systematic increase at low spatial frequencies. The volume CT dose index (CTDI ) for a custom-made 100mm diameter phantom was 9.32mGy. Radiation measurements around the scanner showed that it is completely shielded except for the access port, and that no additional protective measures are necessary for thepatient. A routine QA framework for SPCCT systems has been proposed. Image quality and radiation dose were assessed using newly designed phantoms, relevant metrics, and automated algorithms. Baseline values were established and tolerance levels discussed for the MARS SPCCT scanner based on collected data and internationalrecommendations.

Deep Neural Networks-Based Age Estimation of Cadavers Using CT Imaging of Vertebrae.

Although age estimation upon death is important in the identification of unknown cadavers for forensic scientists, to the best of our knowledge, no study has examined the utility of deep neural network (DNN) models for age estimation among cadavers. We performed a postmortem computed tomography (CT) examination of 1000 and 500 male and female cadavers, respectively. These CT slices were converted into 3-dimensional images, and only the thoracolumbar region was extracted. Eighty percent of them were categorized as training datasets and the others as test datasets for both sexes. We fine-tuned the ResNet152 models using the training datasets. We conducted 4-fold cross-validation, and the mean absolute error (MAE) of the test datasets was calculated using the ensemble learning of four ResNet152 models. Consequently, the MAE of the male and female models was 7.25 and 7.16, respectively. Our study shows that DNN models can be useful tools in the field of forensic medicine.

FasterRib: A deep learning algorithm to automate identification and characterization of rib fractures on chest computed tomography scans.

Characterizing and enumerating rib fractures are critical to informing clinical decisions, yet in-depth characterization is rarely performed because of the manual burden of annotating these injuries on computed tomography (CT) scans. We hypothesized that our deep learning model, FasterRib , could predict the location and percentage displacement of rib fractures using chest CT scans. The development and internal validation cohort comprised more than 4,700 annotated rib fractures from 500 chest CT scans within the public RibFrac. We trained a convolutional neural network to predict bounding boxes around each fracture per CT slice. Adapting an existing rib segmentation model, FasterRib outputs the three-dimensional locations of each fracture (rib number and laterality). A deterministic formula analyzed cortical contact between bone segments to compute percentage displacements. We externally validated our model on our institution's data set. FasterRib predicted precise rib fracture locations with 0.95 sensitivity, 0.90 precision, 0.92 f1 score, with an average of 1.3 false-positive fractures per scan. On external validation, FasterRib achieved 0.97 sensitivity, 0.96 precision, and 0.97 f1 score, and 2.24 false-positive fractures per scan. Our publicly available algorithm automatically outputs the location and percent displacement of each predicted rib fracture for multiple input CT scans. We built a deep learning algorithm that automates rib fracture detection and characterization using chest CT scans. FasterRib achieved the highest recall and the second highest precision among known algorithms in literature. Our open source code could facilitate FasterRib's adaptation for similar computer vision tasks and further improvements via large-scale external validation. Diagnostic Tests/Criteria; Level III.

Implementation of Standard Deviation Map (SDM) for an Automation of Spatial Resolution Measurements on Computed Tomography Images of ACR CT Accreditation Phantom

This study is to develop an automated method to determine the spatial resolution of computed tomography (CT) images on line-pair objects of the American College of Radiology (ACR) CT accreditation phantom. The ACR phantom was scanned using a GE Healthcare 128-slice CT scanner with seven different reconstruction filters of E1, E2, E3, LU, S1, S2, S3. The automated method involved building a standard deviation map (SDM), segmenting the line-pair objects within SDM images, determining the region of interest (ROI) within line pair object, and determining a resolvable line-pair object with dynamic threshold which is dependent on image noise. The scanning parameters were fixed at 120 kVp and 160 mAs. The results of automated method were compared with those from manual measurements performed by five human observers. The automatic method produced spatial resolution results of 0.7, 0.7, 0.7, 0.7, 0.6, 0.6, and 0.6 lp/mm for filters E1, E2, E3, LU, S1, S2, and S3, respectively, while the manual measurements yielded results of 0.6, 0.6, 0.6, 0.7, 0.6, 0.5, and 0.5 lp/mm. The differences between the manual and automatic measurement results were small, with a maximum difference of 0.1 lp/mm. Hence, the automatic measurement of spatial resolution on line-pair objects of the ACR CT accreditation phantom is a feasible and reliable method.

Assessment of Utility of Computed Tomography Pulmonary Angiography in Pulmonary Embolism

Background: Among all the available diagnostic tools, computed tomography (CT) is the preferred diagnostic tool for clinically suspected pulmonary embolism (PE) cases. Further, with the development of multi–detector row CT (MDCT) technology, previous limitations of CT have been overcome effectively. For patients with clinical suspicion of PE, MDCT pulmonary angiography has become the first line of modality for the diagnosis. Aims and Objectives: To evaluate various common observations of CT pulmonary angiography (CTPA) in suspected cases of PE that could accurately determine the choice of management. Materials and Methods: Fifty patients with clinically suspected PE referred to the department of Radiodiagnosis for CTPA from various departments were studied in the cross-sectional study. CT images were obtained with a SIEMENS 64-slice multi-detector CT scanner (somatom definition AS). Images were reconstructed into 3D, axial, coronal, sagittal maximum intensity projection sections, and volume rendering techniques. The axial as well as reformatted coronal and sagittal images were evaluated. Results: The incidence of PE was 52%. The incidence of PE was highest in patients older than 60 years (30.8%) and males (30.3%; p=0.662). The most common symptoms evoking a suspicion of pulmonary embolism were dyspnea (69.2%), followed by chest pain (15.4%), cough (11.5%), and hemoptysis (3.8%). Out of 26 patients who had PE, 61.5% had right ventricular enlargement. DVT (23.1%) was the most common past history. The majority of the patients had thrombus in the right main pulmonary artery (21.1 %), followed by the left main pulmonary artery (19.3 %). Conclusion: In clinically suspected cases of PE, MDCT pulmonary angiography was a useful diagnostic tool. CTPA is the modality of choice for its rapid, minimally invasive, and well-tolerated nature.

Coronary computed tomography angiography assessment of relationship between right coronary artery-aorta angle and the development of coronary artery disease.

Whilst a correlation has been established between wide left main coronary artery bifurcation [left anterior descending-left circumflex (LAD-LCx)] angle (>80°) and the development of coronary artery disease (CAD), this retrospective, causal-comparative pilot study aimed to explore whether a relationship exists between right coronary artery (RCA)-aorta angle and CAD. Thirty normal cases were identified via radiology reports and selected as the control group with coronary computed tomography angiography (CCTA) scans performed on a 320-slice computed tomography (CT) scanner. Thirty CAD cases were selected with invasive coronary angiography performed to confirm the degree of stenosis, and CCTA performed on dual source and 320-slice CT scanners. An independent sample t-test was used to compare the differences in coronary angles between the normal and CAD group, and analysis of variance (ANOVA) was used to assess for significant differences between coronary angles in normal and CAD subgroups. Coronary angle measurements were conducted by two independent assessors with high intraclass correlation (r=0.971-0.998, P<0.001). RCA-aorta angle measurements were significantly larger in the normal group [87.47°, 95% confidence interval (CI): 79.31° to 95.78°] compared to the CAD group (76.82°, 95% CI: 67.82° to 85.61°, P=0.05). No significant difference was found between RCA-aorta angle and degree of coronary stenosis (P=0.75). This study suggests a relationship between narrow RCA-aorta angle and CAD.