scholarly journals Consecutive Serial Non-Contrast CT Scan-Based Deep Learning Model Facilitates the Prediction of Tumor Invasiveness of Ground-Glass Nodules

2021 ◽  
Vol 11 ◽  
Author(s):  
Yao Xu ◽  
Yu Li ◽  
Hongkun Yin ◽  
Wen Tang ◽  
Guohua Fan
Keyword(s):  
Ct Images ◽  
Image Features ◽  
Combined Model ◽  
Baseline Model ◽  
Discrimination Ability ◽  
Tumor Invasiveness ◽  
Testing Dataset ◽  
Noncontrast Ct ◽  
Independent Testing Dataset

IntroductionTumors are continuously evolving biological systems which can be monitored by medical imaging. Previous studies only focus on single timepoint images, whether the performance could be further improved by using serial noncontrast CT imaging obtained during nodule follow-up management remains unclear. In this study, we evaluated DL model for predicting tumor invasiveness of GGNs through analyzing time series CT imagesMethodsA total of 168 pathologically confirmed GGN cases (48 noninvasive lesions and 120 invasive lesions) were retrospectively collected and randomly assigned to the development dataset (n = 123) and independent testing dataset (n = 45). All patients underwent consecutive noncontrast CT examinations, and the baseline CT and 3-month follow-up CT images were collected. The gross region of interest (ROI) patches containing only tumor region and the full ROI patches including both tumor and peritumor regions were cropped from CT images. A baseline model was built on the image features and demographic features. Four DL models were proposed: two single-DL model using gross ROI (model 1) or full ROI patches (model 3) from baseline CT images, and two serial-DL models using gross ROI (model 2) or full ROI patches (model 4) from consecutive CT images (baseline scan and 3-month follow-up scan). In addition, a combined model integrating serial full ROI patches and clinical information was also constructed. The performance of these predictive models was assessed with respect to discrimination and clinical usefulness.ResultsThe area under the curve (AUC) of the baseline model, models 1, 2, 3, and 4 were 0.562 [(95% confidence interval (C)], 0.406~0.710), 0.693 (95% CI, 0.538–0.822), 0.787 (95% CI, 0.639–0.895), 0.727 (95% CI, 0.573–0.849), and 0.811 (95% CI, 0.667–0.912) in the independent testing dataset, respectively. The results indicated that the peritumor region had potential to contribute to tumor invasiveness prediction, and the model performance was further improved by integrating imaging scans at multiple timepoints. Furthermore, the combined model showed best discrimination ability, with AUC, sensitivity, specificity, and accuracy achieving 0.831 (95% CI, 0.690–0.926), 86.7%, 73.3%, and 82.2%, respectively.ConclusionThe DL model integrating full ROIs from serial CT images shows improved predictive performance in differentiating noninvasive from invasive GGNs than the model using only baseline CT images, which could benefit the clinical management of GGNs.

COVID-19 pneumonia diagnosis using a simple 2D deep learning framework with a single chest CT image (Preprint)

2020 ◽  
Author(s):  
Jinseok Lee
Keyword(s):  
Deep Learning ◽  
Diagnostic Performance ◽  
Ct Images ◽  
Chest Ct ◽  
University Hospital ◽  
Detection Accuracy ◽  
Ct Image ◽  
Test Dataset ◽  
Learning Framework ◽  
Testing Dataset

BACKGROUND The coronavirus disease (COVID-19) has explosively spread worldwide since the beginning of 2020. According to a multinational consensus statement from the Fleischner Society, computed tomography (CT) can be used as a relevant screening tool owing to its higher sensitivity for detecting early pneumonic changes. However, physicians are extremely busy fighting COVID-19 in this era of worldwide crisis. Thus, it is crucial to accelerate the development of an artificial intelligence (AI) diagnostic tool to support physicians. OBJECTIVE We aimed to quickly develop an AI technique to diagnose COVID-19 pneumonia and differentiate it from non-COVID pneumonia and non-pneumonia diseases on CT. METHODS A simple 2D deep learning framework, named fast-track COVID-19 classification network (FCONet), was developed to diagnose COVID-19 pneumonia based on a single chest CT image. FCONet was developed by transfer learning, using one of the four state-of-art pre-trained deep learning models (VGG16, ResNet50, InceptionV3, or Xception) as a backbone. For training and testing of FCONet, we collected 3,993 chest CT images of patients with COVID-19 pneumonia, other pneumonia, and non-pneumonia diseases from Wonkwang University Hospital, Chonnam National University Hospital, and the Italian Society of Medical and Interventional Radiology public database. These CT images were split into a training and a testing set at a ratio of 8:2. For the test dataset, the diagnostic performance to diagnose COVID-19 pneumonia was compared among the four pre-trained FCONet models. In addition, we tested the FCONet models on an additional external testing dataset extracted from the embedded low-quality chest CT images of COVID-19 pneumonia in recently published papers. RESULTS Of the four pre-trained models of FCONet, the ResNet50 showed excellent diagnostic performance (sensitivity 99.58%, specificity 100%, and accuracy 99.87%) and outperformed the other three pre-trained models in testing dataset. In additional external test dataset using low-quality CT images, the detection accuracy of the ResNet50 model was the highest (96.97%), followed by Xception, InceptionV3, and VGG16 (90.71%, 89.38%, and 87.12%, respectively). CONCLUSIONS The FCONet, a simple 2D deep learning framework based on a single chest CT image, provides excellent diagnostic performance in detecting COVID-19 pneumonia. Based on our testing dataset, the ResNet50-based FCONet might be the best model, as it outperformed other FCONet models based on VGG16, Xception, and InceptionV3.

scholarly journals Using Ultrasound and Inflammation to Improve Prediction of Ischemic Stroke: A Secondary Analysis of the Multi-Ethnic Study of Atherosclerosis

2021 ◽  
pp. 37-43
Author(s):  
Hediyeh Baradaran ◽  
Alen Delic ◽  
Ka-Ho Wong ◽  
Nazanin Sheibani ◽  
Matthew Alexander ◽  
...  
Keyword(s):  
Ischemic Stroke ◽  
Goodness Of Fit ◽  
Primary Outcome ◽  
Stroke Risk ◽  
Secondary Analysis ◽  
Predictor Variables ◽  
Baseline Model ◽  
Fit Statistics ◽  
Ischemic Stroke Risk

Introduction: Current ischemic stroke risk prediction is primarily based on clinical factors, rather than imaging or laboratory markers. We examined the relationship between baseline ultrasound and inflammation measurements and subsequent primary ischemic stroke risk. Methods: In this secondary analysis of the Multi-Ethnic Study of Atherosclerosis (MESA), the primary outcome is the incident ischemic stroke during follow-up. The predictor variables are 9 carotid ultrasound-derived measurements and 6 serum inflammation measurements from the baseline study visit. We fit Cox regression models to the outcome of ischemic stroke. The baseline model included patient age, hypertension, diabetes, total cholesterol, smoking, and systolic blood pressure. Goodness-of-fit statistics were assessed to compare the baseline model to a model with ultrasound and inflammation predictor variables that remained significant when added to the baseline model. Results: We included 5,918 participants. The primary outcome of ischemic stroke was seen in 105 patients with a mean follow-up time of 7.7 years. In the Cox models, we found that carotid distensibility (CD), carotid stenosis (CS), and serum interleukin-6 (IL-6) were associated with incident stroke. Adding tertiles of CD, IL-6, and categories of CS to a baseline model that included traditional clinical vascular risk factors resulted in a better model fit than traditional risk factors alone as indicated by goodness-of-fit statistics. Conclusions: In a multiethnic cohort of patients without cerebrovascular disease at baseline, we found that CD, CS, and IL-6 helped predict the occurrence of primary ischemic stroke. Future research could evaluate if these basic ultrasound and serum measurements have implications for primary prevention efforts or clinical trial inclusion criteria.

scholarly journals Robustness of radiomic features in CT images with different slice thickness, comparing liver tumour and muscle

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Lorena Escudero Sanchez ◽  
Leonardo Rundo ◽  
Andrew B. Gill ◽  
Matthew Hoare ◽  
Eva Mendes Serrao ◽  
...  
Keyword(s):  
Large Fraction ◽  
Therapy Response ◽  
Response Assessment ◽  
Ct Images ◽  
Optimal Number ◽  
Image Features ◽  
Liver Tumour ◽  
Slice Thickness ◽  
Invasive Method ◽  
Feature Values

AbstractRadiomic image features are becoming a promising non-invasive method to obtain quantitative measurements for tumour classification and therapy response assessment in oncological research. However, despite its increasingly established application, there is a need for standardisation criteria and further validation of feature robustness with respect to imaging acquisition parameters. In this paper, the robustness of radiomic features extracted from computed tomography (CT) images is evaluated for liver tumour and muscle, comparing the values of the features in images reconstructed with two different slice thicknesses of 2.0 mm and 5.0 mm. Novel approaches are presented to address the intrinsic dependencies of texture radiomic features, choosing the optimal number of grey levels and correcting for the dependency on volume. With the optimal values and corrections, feature values are compared across thicknesses to identify reproducible features. Normalisation using muscle regions is also described as an alternative approach. With either method, a large fraction of features (75–90%) was found to be highly robust (< 25% difference). The analyses were performed on a homogeneous CT dataset of 43 patients with hepatocellular carcinoma, and consistent results were obtained for both tumour and muscle tissue. Finally, recommended guidelines are included for radiomic studies using variable slice thickness.

Abstract TP10: Prediction of Thrombolytic Efficacy: Usefulness of Measurement of Enhancement in the Thrombus on Thin-reconstructed Perfusion CT

Stroke ◽  
2016 ◽  
Vol 47 (suppl_1) ◽  
Author(s):  
Dong Hoon Shin ◽  
Eung Yeop Kim
Keyword(s):  
Collateral Circulation ◽  
Motion Correction ◽  
Hounsfield Unit ◽  
Spiral Ct ◽  
Perfusion Ct ◽  
Vascular Imaging ◽  
Escape Trial ◽  
Tissue Plasminogen ◽  
Noncontrast Ct

Purpose: To directly measure enhancement in acute thrombi using thin-reconstructed perfusion CT images for prediction thrombolytic efficacy. Materials and Methods: Prior to administration of tissue plasminogen activator (tPA), noncontrast CT (NCCT), 60-second 70-kVp adaptive 4D spiral CT (CTP), and CT angiography (CTA) were prospectively obtained and reconstructed at 1-mm thickness. Length and Hounsfield unit ratio (HUr) of thrombus were measured using 1-mm NCCT. Collateral circulation was assessed on dynamic CTA that were reconstructed from 1-mm CTP images. Good collateral circulation was defined as the criteria that were used for ESCAPE trial. After spatial motion correction of 1-mm CTP images, circular regions of interest were drawn in the central portion and each end of thrombus to measure the level of HU increase from baseline on time-attenuation curves (TAC). Recanalization was assessed on follow-up vascular imaging studies that were obtained within 24 hours after tPA. Modified TICI 2b or 3 was considered successful recanalization. Thrombus length, HUr, collaterals, and minimum increase of HU on TAC (HUmin) were compared between the recanalized and non-recanalized groups. Results: Of 57 patients who received tPA therapy, 31 patients (female, 13; mean age, 66.5 years) with occlusions in ICA (n=7), M1 (n=8), M1-M2 (n=6), and M2 (n=10) were only assessed. Thrombus length ranged 3-45 mm (median, 12 mm; IQR, 7). HUr was measured from 1.03 to 1.69 (median, 1.26; IQR, 0.19). Good collaterals were noted in 27 patients. HUmin ranged 3-70 HU (median, 15; IQR, 12), and showed negative correlation with thrombus length (rho=-0.410, P=0.022), but not with HUr. HUmin was significantly higher in the recanalized group (n=19) than the non-recanalized group (mean HUmin 23.79 vs 7.83; P<0.0001) independent of thrombus location. Thrombus length, HUr, or collateral status was not significantly different between the two groups. HUmin > 13 was determined with sensitivity of 89.5%, specificity of 91.7%, and AUC of 0.961 for prediction of recanalization. Conclusion: HUmin of thrombus was significantly higher in patients with successful recanalization after tPA therapy.

scholarly journals A graph-cut approach for pulmonary artery-vein segmentation in noncontrast CT images

2019 ◽  
Vol 52 ◽  
pp. 144-159 ◽  
Author(s):  
Daniel Jimenez-Carretero ◽  
David Bermejo-Peláez ◽  
Pietro Nardelli ◽  
Patricia Fraga ◽  
Eduardo Fraile ◽  
...  
Keyword(s):  
Pulmonary Artery ◽  
Ct Images ◽  
Graph Cut ◽  
Noncontrast Ct

scholarly journals Computerized Tomography Image Features under the Reconstruction Algorithm in the Evaluation of the Effect of Ropivacaine Combined with Dexamethasone and Dexmedetomidine on Assisted Thoracoscopic Lobectomy

2021 ◽  
Vol 2021 ◽  
pp. 1-9
Author(s):  
Yan Cui ◽  
Yang Sun ◽  
Meng Xia ◽  
Dan Yao ◽  
Jun Lei
Keyword(s):  
Fourier Transform ◽  
Iterative Reconstruction ◽  
Analytical Method ◽  
Reconstruction Algorithm ◽  
Ct Images ◽  
Image Features ◽  
Anesthesia Induction ◽  
Iterative Reconstruction Algorithm ◽  
Thoracoscopic Lobectomy ◽  
The Fourier Transform

This research was aimed to study CT image features based on the backprojection filtering reconstruction algorithm and evaluate the effect of ropivacaine combined with dexamethasone and dexmedetomidine on assisted thoracoscopic lobectomy to provide reference for clinical diagnosis. A total of 110 patients undergoing laparoscopic resection were selected as the study subjects. Anesthesia induction and nerve block were performed with ropivacaine combined with dexamethasone and dexmedetomidine before surgery, and chest CT scan was performed. The backprojection image reconstruction algorithm was constructed and applied to patient CT images for reconstruction processing. The results showed that when the overlapping step size was 16 and the block size was 32 × 32, the running time of the algorithm was the shortest. The resolution and sharpness of reconstructed images were better than the Fourier transform analytical method and iterative reconstruction algorithm. The detection rates of lung nodules smaller than 6 mm and 6–30 mm (92.35% and 95.44%) were significantly higher than those of the Fourier transform analytical method and iterative reconstruction algorithm (90.98% and 87.53%; 88.32% and 90.87%) ( P < 0.05 ). After anesthesia induction and lobectomy with ropivacaine combined with dexamethasone and dexmedetomidine, the visual analogue scale (VAS) decreased with postoperative time. The VAS score decreased to a lower level (1.76 ± 0.54) after five days. In summary, ropivacaine combined with dexamethasone and dexmedetomidine had better sedation and analgesia effects in patients with thoracoscopic lobectomy. CT images based on backprojection reconstruction algorithm had a high recognition accuracy for lung lesions.

scholarly journals Maximum Diameter Measurements of Aortic Aneurysms on Axial CT Images After Endovascular Aneurysm Repair: Sufficient for Follow-up?

2010 ◽  
Vol 34 (6) ◽  
pp. 1182-1189 ◽  
Author(s):  
Stephan Baumueller ◽  
Thi Dan Linh Nguyen ◽  
Robert Paul Goetti ◽  
Mario Lachat ◽  
Burkhardt Seifert ◽  
...  
Keyword(s):  
Endovascular Aneurysm Repair ◽  
Ct Images ◽  
Aortic Aneurysms ◽  
Maximum Diameter ◽  
Axial Ct ◽  
Aneurysm Repair

Abstract T P10: Hounsfield Unit Value and Clot Length in the Acutely Occlude Vessel and Time Required to Achieve Thrombectomy, Complications and Outcome

Stroke ◽  
2014 ◽  
Vol 45 (suppl_1) ◽  
Author(s):  
Alejandro Spiotta ◽  
Jan Vargas ◽  
Harris Hawk ◽  
Raymond Turner ◽  
Imran Chaudry ◽  
...  
Keyword(s):  
Ischemic Stroke ◽  
Acute Ischemic Stroke ◽  
Functional Outcome ◽  
Hounsfield Unit ◽  
Hounsfield Units ◽  
Unit Value ◽  
Procedural Complications ◽  
Noncontrast Ct ◽  
Time Required

Introduction: Intra-arterial therapy for acute ischemic stroke (AIS) now has an established role. We investigated if Hounsfield Units (HU) quantification on noncontrast CT is associated with ease and efficacy of mechanical thrombectomy and outcomes. Methods: We retrospectively studied a prospectively maintained database of cases of acute ischemic stroke that underwent intra-arterial therapy between May 2008 and August 2012. Functional outcome was assessed by ninety-day follow up modified Rankin Scale (mRS). Patients were dichotomized base on time to recanalization. Hounsfield units were calculated on head CT. Thrombus location and length were determined on CT angiography. Simple linear regression was used to analyze the association between clot length, average HU, and other clinical variables. Results: 141 patients were included. There was no difference in clot length or average HU among patients with good recanalization achieved within an hour compared to those in which procedures extended beyond an hour. There was no relationship between clot length or density and recanalization. The thrombus length and density were not significantly different between patients with procedural complications and those without. The presence of post procedure intracranial hemorrhage was not associated with thrombus length or density. Ninety day mRS was not associated with thrombus length or density. Conclusions: We have not found any significant associations between either thrombus length or density and likelihood of recanalization, time to achieve recanalization, intraprocedural complications, postprocedural hemorrhage or functional outcome at ninety days. These results do not support a predictive value for thrombus quantification in the evaluation of AIS.

Sign in / Sign up

Export Citation Format

Share Document