64-slice CT Research Articles

Comparative assessment of two imaging modalities and their application in the evaluation of pancreatic/peripancreatic changes due to acute pancreatitis

Aim: Comparison of two imaging modalities and their application in the evaluation of pancreatic/peripancreatic changes due to acute pancreatitis.Materials and Methods: The present comparative study was performed on 100 patients suspected of acute pancreatitis confirmed by elevated serum amylase and serum lipase levels, between November 2015 to May 2017 in the Department of Radiodiagnosis in Victoria Hospital and Bowring and Lady Curzon Hospital attached to Bangalore Medical College and Research Institute, Bengaluru. All patients underwent a real-time ultrasound scan of the abdomen using a curvilinear transducer of 2 to 6 MHz of Philips IU22 and triple-phase contrast-enhanced computed tomography of the abdomen by Philips INGENUITY 128 slice CT in Victoria hospital or Siemens SOMATOM 6 slice CT present in Bowring hospital.Results: Most patients with acute pancreatitis (~30%) belonged to the age group 41-50 years while 28% each belonged to the age group 31-41 and 41-50 years respectively. Amongst males (n=88), 72% (n=64) were cases of acute pancreatitis and 28% (n=24) were the cases of acute on pancreatitis. On the other hand, of 12 females, 58% (n=07) were the cases of acute pancreatitis and 42% (n=05) were the cases of acute on chronic pancreatitis. Both ultrasonography (US) and MDCT were able to delineate normal-sized pancreas, increased size of head and body of pancreas and atrophic pancreas. While the US was able to identify only 17 cases, CT was able to detect 55 with a sensitivity of ~31%, which was statistically significant (p

Optimizing Region of Interest Size and Placement for Opportunistic CT

Opportunistic CT is increasingly used clinically, however, no guidance exists regarding optimal region of interest (ROI) size or placement to obtain Hounsfield Unit (HU) data. As ROI size and placement alter HU, standardization is desirable. This studies purposes were to: 1.) Evaluate various L1 and L4 ROI sizes and locations to obtain reproducible HU values and; 2.) Compare HU values from axial and sagittal images. On lumbar spine CT, 2 physicians independently placed varying size circular ROIs (100, 200,300 mm2 and maximum) at L1 and L4 on axial and sagittal images. Subsequent analyses placed a 200 mm2 ROI at L1 and L4 in anterior, mid and posterior aspects and also cranial, central and caudal and left/right locations. Intra and inter-observer reliability was determined using intraclass correlation coefficients. ROI size comparison was performed by ANOVA and between axial and sagittal by t-test. Limits of agreement was assessed by Bland-Altman. The study included 30 spine clinic patients; 21F/9M mean (SD) age 59.2 (14.4) yrs. Vertebral body centroid identification, i.e., CT slice location used for HU measurement, was nearly identical between readers with ICC >0.99 for both projections. Intra- and inter-observer reliability was excellent for all sizes on both projections with ICC >0.95; Table 1. HU did not differ between the 4 ROI sizes on either projection for L1 and L4; all p >0.95; Table 1. Correlations between ROI sizes were excellent; r >0.95. Axial HU was higher than sagittal; p < 0.001 for all ROI sizes at L1 and L4; Table 1. Anterior, mid or posterior placement did not affect axial or sagittal HU; p >0.5 at L1 or L4; Table 2. However, moving away from the vertebral body centroid, either cranial/caudal or left/right, altered HU; Table 3. Limits of agreement between cranial/caudal and left/right location were large (from 10-90 HU). Non-radiologists can reliably identify the vertebral body centroid to facilitate ROI placement on axial and sagittal CT images and measure HU. While ROI size did not affect HU, 200 mm2 is recommended as larger ROI sizes may extend to the vertebral cortex in small patients which changes HU. Axial results were ∼6-10 HU higher than sagittal. Anterior, mid or posterior ROI location did not affect HU but image plane (left/right or cranial/caudal) alters HU likely reflecting anisotropic vertebral microarchitecture. For clinical use, we recommend default ROI placement be a 200 mm2 circular ROI placed at the vertebral body centroid.

Variations of Anterior Cerebral Artery (ACA) on CT Angiography (CTA)

Introduction: The variations in anterior cerebral artery (ACA) has high clinical significance for surgeons&#x0D; Aim: This study aims to assess the variations of ACA with the help of CT angiography (CTA)&#x0D; Methodology: This retrospective study was conducted from Nov 2019 to Feb 2020 at Lahore General Hospital, Lahore. In total, 150 patients went under detailed analysis. The demographic data such as age and gender were noted for each patient. The CTA was performed with the help of 128-slice CT scanner. The reports were assessed by an experienced radiologist. Anatomic variations were noted and presented as frequencies and percentages.&#x0D; Results: The study included 92 (61.33%) were females and 58 (38.66%) were males with mean age 50.2 ± 3.7 years. Azygos anterior cerebral artery was present in 14 patients (9%), hypoplasia of anterior cerebral artery A1 segment in 43 (35.66%) patients,11 (7.33%) patients showed bi-hemispheric anterior cerebral artery, right anterior cerebral artery dominance in 2 (1.33%) patients, 8 (4.33 %) were detected with congenital absence of A1 segment. Anterior cerebral artery trifurcation had prevalence of 6(3.2%).&#x0D; Conclusion: It can be concluded that anterior cerebral artery variations have low prevalence and CT angiography is a gold method for detecting them.

Estimation of Covid-19 lungs damage based on computer tomography images analysis

Modern treatment is based on reproducible quantitative analysis of available data. The Covid-19 pandemic did accelerate development and research in several multidisciplinary areas. One of them is the use of software tools for faster and reproducible patient data evaluation. A CT scan can be invaluable for a search of details, but it is not always easy to see the big picture in 3D data. Even in the visual analysis of CT slice by slice can inter and intra variability makes a big difference. We present an ImageJ tool developed together with the radiology center of Faculty hospital Královské Vinohrady for CT evaluation of patients with COVID-19. The tool was developed to help estimate the percentage of lungs affected by the infection. The patients can be divided into five groups based on percentage score and proper treatment can be applied

Spinal Canal Measurements at the Level of Lower Three Lumbar Vertebrae by 128-Slice CT Scanner in Bangladeshi Population

Background &amp; objective : The knowledge of normal diameter of lumbar spinal canal is very important for diagnosing lumbar spinal canal stenosis and also for performing spinal surgeries at the lumbar level by Neurosurgeons and Orthopedicians. However, it varies widely among ethnic groups and between sexes in the same ethnic group. The present study was conducted to describe the morphometry of lumbar vertebrae at the level of L3-L5.&#x0D; Methods: This descriptive cross-sectional study was conducted on 302 patients (aged 20-60 years) at Ibrahim Cardiac Hospital &amp; Research Institute between July 2019 to December, 2020 who underwent diagnostic CT scans for abdominal or genitourinary complaints without any known vertebral column pathology or complaints of low back pain or other abnormalities attributable to lumbar spine. Patients having sciatic pain with or without pain in the back, having past history of back surgery and patients with osteophytes or developmental anomalies, trauma or vertebral fracture and known case of lordosis, scoliosis or kyphosis or other abnormalities in lumbar vertebrae were also excluded. A CT scan was performed using 128-slice multi-detector CT with unenhanced CT images from the level of diaphragm to pubic symphysis so that the area from D12 to S1 vertebra be covered (field of view). Sections 3-mm thick with reconstruction up to 1 mm were analyzed and different parameters of the spinal canal and vertebral body were measured in sagittal and transverse sections. The images were reconstructed in true axial, coronal, and sagittal planes. The canal-body ratio (CBR) was determined by dividing the sagittal diameter of the spinal canal by the sagittal diameter of the vertebral body.&#x0D; Result: The findings of the present study demonstrate that nearly two-thirds (65.5%) of the subjects were early middle-aged or middle-aged with mean age of the subjects being 43 years. A male predominance was observed in the series with male to female ratio being 4:1. The study revealed that all the dimensions of the canal and the body in transverse and sagittal sections were observed to increase from level L3 to L5, while the canal body ratio was decreased slightly from L3-L5. The males have significantly wider vertebral body at all levels (L3-L5) of lumbar vertebrae compared to their female counterparts. Vertebral body diameter in sagittal section at L3 was significantly greater in males than those in females; however, they were not different between sexes at L4 and L5 levels. Canal-body ratio (CBR) was fairly comparable between males and females at L3 and L4, but it was greater in males than that in females at L5. No significant difference was observed between L3 &amp; L4 and L4 &amp; L5 in the same individuals in terms of CBR.&#x0D; Conclusion: The study concluded that there are significant variations in some of the lumbar spinal canal dimensions and vertebral body measurements (including CBR) from L3-L5 between males and females. Although there was no significant difference between L3 and L4 in the same individuals in terms of CBR, there was reasonable difference between L4 and L5 lumbar vertebrae. The findings are of much significance in the investigation of vertebral column pathology in the context of our population.&#x0D; Ibrahim Card Med J 2021; 11 (1): 8-13

Generation of synthetic training data for SEEG electrodes segmentation.

Stereoelectroencephalography (SEEG) is a minimally invasive surgical procedure, used to locate epileptogenic zones. An accurate identification of the metallic contacts recording the SEEG signal is crucial to ensure effectiveness of the upcoming treatment. However, due to the presence of metal, post-operative CT scans contain strong streak artefacts that interfere with deep learning segmentation algorithms and require a lot of training data to distinguish from actual contacts. We propose a method to generate synthetic data and use them to train a neural network to precisely locate SEEG electrode contacts. Random electrodes were generated following manufacturer's specifications and dimensions and placed in acceptable regions inside metal-free CT images. Metal artefacts were simulated in the generated data set using radon transform, beam hardening, and filtered back projection. A UNet neural network was trained for the contacts segmentation task using various training set-ups combining real data, basic augmented data, and synthetic data. The results were compared. We reported a higher accuracy when including synthetic data during the network training, while training only on real and basic augmented data more often led to misclassified artefacts or missed contacts. The network segments post-operative CT slices in less than 2 s using 4 GeForce RTX2080 Ti GPUs and in under a minute using a standard PC with GeForce GTX1060. Using synthetic data to train the network significantly improves contact detection and segmentation accuracy.

Revisiting segmentation of lung tumors from CT images

Lung cancer is a leading cause of death throughout the world. Because the prompt diagnosis of tumors allows oncologists to discern their nature, type, and mode of treatment, tumor detection and segmentation from CT scan images is a crucial field of study. This paper investigates lung tumor segmentation via a two-dimensional Discrete Wavelet Transform (DWT) on the LOTUS dataset (31,247 training, and 4458 testing samples) and a Deeply Supervised MultiResUNet model. Coupling the DWT, which is used to achieve a more meticulous textural analysis while integrating information from neighboring CT slices, with the deep supervision of the model architecture results in an improved dice coefficient of 0.8472. A key characteristic of our approach is its avoidance of 3D kernels (despite being used for a 3D segmentation task), thereby making it quite lightweight.

Deep learning representations to support COVID-19 diagnosis on CT slices.

Introduction: The coronavirus disease 2019 (COVID-19) has become a significant public health problem worldwide. In this context, CT-scan automatic analysis has emerged as a COVID-19 complementary diagnosis tool allowing for radiological finding characterization, patient categorization, and disease follow-up. However, this analysis depends on the radiologist’s expertise, which may result in subjective evaluations.Objective: To explore deep learning representations, trained from thoracic CT-slices, to automatically distinguish COVID-19 disease from control samples.Materials and methods: Two datasets were used: SARS-CoV-2 CT Scan (Set-1) and FOSCAL clinic’s dataset (Set-2). The deep representations took advantage of supervised learning models previously trained on the natural image domain, which were adjusted following a transfer learning scheme. The deep classification was carried out: (a) via an end-to-end deep learning approach and (b) via random forest and support vector machine classifiers by feeding the deep representation embedding vectors into these classifiers.Results: The end-to-end classification achieved an average accuracy of 92.33% (89.70% precision) for Set-1 and 96.99% (96.62% precision) for Set-2. The deep feature embedding with a support vector machine achieved an average accuracy of 91.40% (95.77% precision) and 96.00% (94.74% precision) for Set-1 and Set-2, respectively.Conclusion: Deep representations have achieved outstanding performance in the identification of COVID-19 cases on CT scans demonstrating good characterization of the COVID-19 radiological patterns. These representations could potentially support the COVID-19 diagnosis in clinical settings.

Role of Multi-Detector Computed Tomography Indices in Predicting Extracorporeal Shockwave Lithotripsy Outcome in Patients With Nephrolithiasis

BackgroundNephrolithiasis is one of the most common renal pathologies and is routinely encountered in daily practice. Non-contrast computed tomography (NCCT) is the gold standard diagnostic imaging modality for urolithiasis. The role of HU (Hounsfield units) in calculus as a predictor of extracorporeal shock wave lithotripsy (ESWL) has been studied in the past. This study aims to evaluate the role of HU value and various other NCCT indices in predicting the outcome of ESWL.Material and methodsThis was a prospective observational study that included 45 patients suffering from nephrolithiasis who underwent NCCT-KUB (kidney, ureter, and bladder) followed by ESWL. The NCCT indices were evaluated and correlated with the outcome of ESWL. NCCT-KUB was performed using multidetector SIEMENS® SOMATOM EMOTION 16-slice CT scanner (SIEMENS, Munich, Germany).ResultsIn our study, the HU value turned out to be a statistically significant predictor of ESWL success (p <0.05), and the renal pelvis also proved to be a good prognostic indicator for ESWL success. The cut-off value of <1179 HU favored a successful outcome of ESWL, while if >1179 HU, ESWL is likely to fail. Hence, the successful outcome of ESWL is inversely proportional to the HU value. Renal pelvic calculi (n=14) showed a 100% success rate, which was better than all other calculus locations (p<0.05). However, the rest of the indices did not show any statistical significance.ConclusionMulti-detector NCCT-KUB indices can help in the selection of patients with a good prognosis for ESWL, which will prevent the patient from undergoing undesired invasive procedures.

Ensemble Learning With Attention-Based Multiple Instance Pooling for Classification of SPT

Tuberculosis is one of the most common infections in the human population, while reactivation of the MTB bacteria can lead to secondary pulmonary tuberculosis (SPT). SPT is a significant health risk for both immunocompromised individuals and the population of regional hotspots. This paper proposes a novel model for ensemble learning for the efficient identification of SPT in CT imaging. We reformulate the ensemble process as multiple instance learning and utilizes an attention-based pooling layer to ensemble features extracted from a range of base networks. The best performing model consists of 4 base models and an ensemble network with Hopfield-pooling. We abbreviate this model as Hopfield-pooling ensemble (HFPE). For the classification of CT slices between healthy controls and SPT patients, HFPE achieved an accuracy of 96.00±2.55%, specificity of 97.00±2.45%, precision of 96.99±2.46%, sensitivity of 95.00±4.47, and F1-score of 95.92±2.63%, achieving comparable or higher performances to a number of state-of-art approaches.

Temporal Bone Cholesteatoma: Typical Findings and Evaluation of Diagnostic Utility on High Resolution Computed Tomography.

BackgroundPre-operative assessment of middle ear cholesteatoma is a must for assessing the disease's location, extent, and complication, and high-resolution computed tomography (HRCT) is the modality of choice. Therefore, this study aims to assess the common signs of cholesteatoma on HRCT and its diagnostic ability. MethodsFifty patients with suspected cholesteatoma were considered for the study, which was carried out on an Ingenuity Core 128 slice CT scanner (Philips, Amsterdam, Netherlands). The bilateral temporal bones of 50 patients were assessed for soft tissue density and associated findings. The number of temporal bones with soft tissue density was then correlated with intra-operative and histopathological examinations (HPE). ResultsOut of 100 temporal bones, 63 were diseased, and 37 were normal temporal bones. Epitympanum/Prussak's space was the most involved site with soft tissue density seen in 60/63 (95.2%) diseased temporal bones, followed by aditus ad antrum and mesotympanum, which was seen in 51/63 (80.9%) diseased temporal bones. The majority of the soft tissue lesions were non-dependent, accounting for 43/63 (68.2%) of the diseased temporal bones. Bony erosions were seen in 54/63 (85.7%) and bony expansion in 35/63 (55.5%) of the diseased temporal bones. HRCT showed a sensitivity of 100%, specificity of 88.1%, a positive predictive value (PPV) of 92.1%, a negative predictive value (NPV) of 100%, and accuracy of 95% for detection of cholesteatoma. ConclusionHRCT of the temporal bone precisely demonstrates cholesteatoma's location, extent, and bony changes. Therefore, it has exceptional sensitivity, high specificity, and accuracy in diagnosing cholesteatoma.

Transfer learning for data-efficient abdominal muscle segmentation with convolutional neural networks.

BackgroundSkeletal muscle segmentation is an important procedure for assessing sarcopenia, an emerging imaging biomarker of patient frailty. Data annotation remains the bottleneck for training deep learning auto‐segmentation models.PurposeThere is a need to define methodologies for applying models to different domains (e.g., anatomical regions or imaging modalities) without dramatically increasing data annotation.MethodsTo address this problem, we empirically evaluate the generalizability of various source tasks for transfer learning: natural image classification, natural image segmentation, unsupervised image reconstruction, and self‐supervised jigsaw solving. Axial CT slices at L3 were extracted from PET‐CT scans for 204 oesophago‐gastric cancer patients and the skeletal muscle manually delineated by an expert. Features were transferred and segmentation models trained on subsets (n=5,10,25,50,75,100,125) of the manually annotated training set. Four‐fold cross‐validation was performed to evaluate model generalizability. Human‐level performance was established by performing an inter‐observer study consisting of ten trained radiographers.ResultsWe find that accurate segmentation models can be trained on a fraction of the data required by current approaches. The Dice similarity coefficient and root mean square distance‐to‐agreement were calculated for each prediction and used to assess model performance. Models pre‐trained on a segmentation task and fine‐tuned on 10 images produce delineations that are comparable to those from trained observers and extract reliable measures of muscle health.ConclusionsAppropriate transfer learning can generate convolutional neural networks for abdominal muscle segmentation that achieve human‐level performance while decreasing the required data by an order of magnitude, compared to previous methods (n=160→10). This work enables the development of future models for assessing skeletal muscle at other anatomical sites where large annotated data sets are scarce and clinical needs are yet to be addressed.

Application of Esmolol to Control Heart Rate and Artificial Intelligence Reconstruction Analysis for Rapid Zero-Cooperation Emergency Coronary CTA.

Objective To investigate a rapid and effective method for the examination of coronary CTA in emergency patients requiring coronary CTA examination, who have faster heart rate (≧80 bpm) or cannot cooperate with the examination due to the inability of breath holding at poor physical conditions. Methods Before coronary CTA examination, with the ECG monitoring, intravenous injection of esmolol was given to achieve rapid heart rate reduction. Without the patient's cooperation, coronary CTA examination was then performed in a quick and effective manner using the 640-slice high-speed CT. The diagnosis report was obtained through the subsequent reconstruction analysis using artificial intelligence software. Conclusion Esmolol injection can rapidly reduce the heart rate of normal people during exercise and at rest, and the steady blood concentration can be reached in 2 minutes. The half-life is about 5 minutes, with short duration and fewer side effects on patients. The diagnostic rate of coronary artery segment using (excellent + good) CTA image of the patients with esmolol and artificial intelligence analysis in the experimental group was 95.4%, while the diagnostic rate was 91.1% in the control group, and there was no significant statistical difference between the two groups (P > 0.05). Esmolol injection can rapidly reduce heart rate in patients with high heart rate, without holding breath or long-term preoperative preparation; the combination with the analysis of subsequent artificial intelligence reconstruction is a new method for rapid and effective coronary CTA examination in all patients.

Evaluating Intervertebral Foramen Measurements for Patients with Anterior Cervical Disc Surgery in Oblique CT

Objective: It is particularly difficult to determine the level of radiculopathy pain. Because radicular pain usually doesn't fit the dermatomal map. unlike the foramen lumbar vertebrae, where the cervical nerve root is exerted, it has a slope of approximately 45 degrees to the vertebrae in the axical plane and 10-15 degrees in the horizontal plane. Due to this structure of cervical foremen, the evaluation of conventional ct results in incorrect or incomplete diagnosis. In our study, we compared the change in cervical foramen patients with anterior cervical discectomy and cage (fusion) to oblique ct reconstructive and preop and postop. Methods: CT images were acquired in a 64 slice CT system (60–80 mAs, 120 kVp, FC86 reconstruction core and 2.0 mm slice thickness at 2.0 mm intervals). Oblique sagittal images were created with OsiriX Lite v. 12.0.1 version and measurements were performed. For each level, the area has been measured in addition to the short and long axle length of the neural foramens from the pre-op and post-op oblique sagittal images.Results: In our study, 98 cervical intervertebral foramen oblique CT de preop-postop, long axle, short axle and field measurements were evaluated with 43 patients from 16/01/2019 to 03/02/2020, 3 patients with two levels of anterior ceral discectomy + interbody cage. Right side, long axle, short axle, and area respectively; 13,37%, 5,99%, 16,87% increase. Left side, long axle, short axle, and area respectively, increased by 14,27%, 7,11%, 23,20%Conclusion: The cage, which is placed on the intervertebral level, will increase the area of the foramen, increase the space of the foramen, and allow the nerve to decompression. This is why it is necessary to insert the material with the highest height of the intervertebral disk level that can be placed during surgery.

The application research of AI image recognition and processing technology in the early diagnosis of the COVID-19

BackgroundThis study intends to establish a combined prediction model that integrates the clinical symptoms,the lung lesion volume, and the radiomics features of patients with COVID-19, resulting in a new model to predict the severity of COVID-19.MethodsThe clinical data of 386 patients with COVID-19 at several hospitals, as well as images of certain patients during their hospitalization, were collected retrospectively to create a database of patients with COVID-19 pneumonia. The contour of lungs and lesion locations may be retrieved from CT scans using a CT-image-based quantitative discrimination and trend analysis method for COVID-19 and the Mask R-CNN deep neural network model to create 3D data of lung lesions. The quantitative COVID-19 factors were then determined, on which the diagnosis of the development of the patients' symptoms could be established. Then, using an artificial neural network, a prediction model of the severity of COVID-19 was constructed by combining characteristic imaging features on CT slices with clinical factors. ANN neural network was used for training, and tenfold cross-validation was used to verify the prediction model. The diagnostic performance of this model is verified by the receiver operating characteristic (ROC) curve.ResultsCT radiomics features extraction and analysis based on a deep neural network can detect COVID-19 patients with an 86% sensitivity and an 85% specificity. According to the ROC curve, the constructed severity prediction model indicates that the AUC of patients with severe COVID-19 is 0.761, with sensitivity and specificity of 79.1% and 73.1%, respectively.ConclusionsThe combined prediction model for severe COVID-19 pneumonia, which is based on deep learning and integrates clinical aspects, pulmonary lesion volume, and radiomics features of patients, has a remarkable differential ability for predicting the course of disease in COVID-19 patients. This may assist in the early prevention of severe COVID-19 symptoms.

Micro-deformation and fracture evolution of in-situ coal affected by temperature, confining pressure, and differential stress

The actual micro-deformation and fracture-evolution of in-situ coal seam are investigated in this paper through simulation experiment from three aspects: temperature, confining pressure, and differential stress. Firstly, the heatable triaxial-compressor is employed to reconstruct the evolutionary process of the coal seam. Then, the CT slices of the compressed samples in 3D, matching with the optical micrographs, are applied to analyze the coal micro-deformation. The effect of each factor on coal fracture evolution is discussed through CT models from the three-axis direction. Finally, this paper proposes the brittle & ductile deformation law of coal and the evolutionary pattern of tectonically deformed coals in the shallow crust. The main results are presented as follows: (1) The three factors have an individual mechanism on coal micro-deformation. Brittle deformation is independent of temperature and closely related to differential stress. Ductile deformation is divided into low-temperature and high-temperature. (2) Most fractures extend along the orientation similar to differential stress and always develop densely in the stress-concentrated region. These three factors in different stages have various effects on coal fracture evolution. (3) At low temperature, the evolutionary order of tectonically deformed coals is cataclastic-, schistose-, mortar-, granulitic-, scaly-, wrinkle-, and mylonitic coal; Ductile deformed coal evolves from parts of brittle deformed coal. These conclusions are helpful to understand the process of micro-deformation and fracture-evolution of in-situ coal.

Accidental diagnosis of significant coronary artery disease by cardiac CT scan to rule out left atrial appendage thrombus in patients with atrial fibrillation during COVID19 pandemic

Abstract Funding Acknowledgements Type of funding sources: None. Introduction The exclusion of left atrial appendage (LAA) thrombus prior to urgent electrical cardioversion (DCCV) for atrial fibrillation (AF) is traditionally performed by transesophageal echocardiography (TEE). During the COVID19 pandemic, TEE was replaced by cardiac CT (CCT) due to its high aerosol generating property, which in addition to assessing the anatomy of the heart, can also be used to evaluate the coronary arteries at the same time. Methods In our retrospective study, we evaluated the cardiac CT scans of patients undergoing DCCV in our institution between January 1, 2020 and February 28, 2021 for coronary artery disease (CAD). The scans were performed by a GE Revolution 256-slice CT scanner. Results CCT scans were performed in 32 patients to rule out LAA thrombus (24 male; 8 female; age: 61.8 ± 11.2 years; BMI: 29.2 ± 4.4; heart rate: 79.2 ± 24.4 1/min; CHA2DS2VASc score: 2.4 ± 1.5). The mean radiation exposure of the scans was DLP: 356.3 ± 130.1 mGy-cm; effective dose: 5.0 ±1.8 mSv, Ca-score: 361.4 ± 883.0. In 4 equivocal cases TEE was also performed with negative results. No CAD was confirmed in 7 cases, mild CAD in 14 patients. CCT was not diagnostic only in 4 cases. Significant (moderate or severe) CAD was detected in 7 cases, invasive coronary angiography (ICA) was also performed in 5 cases. In 2 cases significant one-vessel disease, in 2 other cases borderline (FFR: 0.81 and 0.84) stenosis and in 1 case only mild CAD was diagnosed by ICA. Conclusions CCT scans performed by our 256-slice CT scanner for AF can identify patients, who require further invasive or invasive and functional coronary artery assessment with satisfactory accuracy. In terms of coronary artery disease, the non-diagnostic scan rate was low, despite the presence of arrhythmia, suboptimal heart rate and higher than usual Ca-score.

Comparison of segmentation methods to improve throughput in annotating AI-observer for detecting colorectal cancer.

142 Background: Colorectal cancer (CRC) is the second leading cause of cancer-related deaths, and its outcome can be improved with better detection of incidental early CRC on routine CT of the abdomen and pelvis (CTAP). AI-second observer (AI) has the potential as shown in our companion abstract. The bottleneck in training AI is the time required for radiologists to segment the CRC. We compared two techniques for accelerating the segmentation process: 1) Sparse annotation (annotating some of the CT slice containing CRC instead of every slice); 2) Allowing AI to perform initial segmentation followed by human adjustment. Methods: 2D U-Net convolutional neural network (CNN) containing 31 million trainable parameters was trained with 58 CRC CT images from Banner MD Anderson (AZ) and MD Anderson Cancer Center (TX) (51 used for training and 7 for validation) and 59 normal CT scans from Banner MD Anderson Cancer Center. Twenty of the 25 CRC cases from public domain data (The Cancer Genome Atlas) were used to evaluate the performance of the models. The CRC was segmented using ITK-SNAP open-source software (v. 3.8). For the first objective, 3 separate models were trained (fully annotated CRC, every other slice, and every third slice). The AI-annotation on the TCGA dataset was analyzed by the percentage of correct detection of CRC, the number of false positives, and the Dice similarity coefficient (DSC). If parts of the CRC were flagged by AI, then it was considered correct. A detection was considered false positive if the marked lesion did not overlap with CRC; contiguous false positives across different slices of CT image were considered a single false positive. DSC measures the quality of the segmentation by measuring the overlap between the ground-truth and AI detected lesion. For the second objective, the time required to adjust the AI-produced annotation was compared to the time required for annotating the entire CRC without AI assistance. The AI-models were trained using ensemble learning (see our companion abstract for details of the techniques). Results: Our results showed that skipping slices of tumor in training did not alter the accuracy, false positives, or DSC classification of the model. When adjusting the AI-observer segmentation, there was a trend toward decreasing the time required to adjust the annotation compared to full manual segmentation, but the difference was not statistically significant (Table; p=0.121). Conclusions: Our results show that both skipping slices of tumor as well as starting with AI-produced annotation can potentially decrease the effort required to produce high-quality ground truth without compromising the performance of AI. These techniques can help improve the throughput to obtain a large volume of cases to train AI for detecting CRC.[Table: see text]

Prevalence and risk factors of retro-styloid lymph node metastasis in oropharyngeal carcinoma

Background Supporting data defining the selection criteria of level VIIb for inclusion in the target volume in radiotherapy (RT) planning are insufficient. We evaluated the prevalence of level VIIb retro-styloid lymph node metastasis (RSLNM) and associated risk factors in patients with oropharyngeal carcinoma (OPC). Materials and methods We retrospectively reviewed pre-treatment [18F]-fluoro-2-deoxy-d-glucose–positron emission tomography/computed tomography (CT) along with contrast-enhanced thin slice CT and magnetic resonance (MR) images of 137 patients pathologically confirmed as having OPC who underwent RT. The location of lymph nodes (LNs) was confirmed on the planning CT images. Fisher’s exact test and logistic regression analyses were made to determine the risk factors of RSLNM. Results RSLNM was confirmed in 18 (13%) patients. All RSLNMs were located within level VIIb on the planning CT images. No patients exhibited LNM in contralateral level VIIb. Furthermore, no patients with negative or single ipsilateral cervical LNM had RSLNM. Fisher’s exact test revealed that smoking status (p=.027), multiple ipsilateral cervical LNM (p=.045) and LN ≥15 mm in the upper limit of ipsilateral level II (p<.001) were significantly associated with RSLNM. Logistic regression analyses revealed that the presence of LNs ≥15 mm in upper limit of ipsilateral level II was significantly associated with RSLNM (odds ratio: 977.297; 95% confidence interval: 57.629–16573.308; p<.001). Conclusions RSLNM is relatively common in patients with OPC with a prevalence rate of approximately 10%. The prevalence of RSLNM in patients with negative or single ipsilateral cervical LNM and contralateral RSLNM is extremely low; therefore, level VIIb can be excluded from the target volume in such patients. LN ≥15 mm in the upper limit of ipsilateral level II is a risk factor for RSLNM. Ipsilateral level VIIb should be included in the target volume for patients with this risk factor. KEY MESSAGE Retro-styloid lymph node metastasis (RSLNM) prevalence is ∼10% in oropharyngeal carcinoma. Lymph node ≥15 mm in ipsilateral level II upper limit is a risk factor for RSLNM.

Effect of Body Mass Index in Coronary CT Angiography Performed on a 256-Slice Multi-Detector CT Scanner.

We aimed to investigate the effect of a patient’s body mass index (BMI) on radiation dose and image quality in prospectively ECG-triggered coronary CT angiography (CCTA) performed on a 256-slice multi-detector CT scanner. In total, 87 consecutive patients receiving CCTA examinations acquired with tube current modulation (TCM) and iterative reconstruction (IR) were enrolled in this study. The dose report recorded from the CT scanner console was used to derive the effective dose for patients. Subjective image quality scoring and objective noise measurements were conducted to quantify the impact of BMI on the image quality of CCTA. Because of the TCM technique, we expected tube current and radiation dose to increase as BMI increased. However, using TCM did not always guarantee sufficient radiation exposure to achieve consistent image quality for overweight or obese patients since the maximum X-ray tube output in milliamperes and kilovoltage peak was reached. The impact of photon starvation noise on image quality was not significant until BMI ≥ 27 kg/m2; this result could be due to IR’s noise reduction capability. Our results also suggest that using TCM with a noise index of 25 HU can reduce radiation dose without compromising image quality compared to images obtained based on the manufacturer’s default settings.