scholarly journals An optimal deep learning framework for multi-type hemorrhagic lesions detection and quantification in head CT images for traumatic brain injury

2021 ◽  
Author(s):  
Aniwat Phaphuangwittayakul ◽  
Yi Guo ◽  
Fangli Ying ◽  
Ahmad Yahya Dawod ◽  
Salita Angkurawaranon ◽  
...  
Keyword(s):  
Traumatic Brain Injury ◽  
Deep Learning ◽  
Brain Injury ◽  
Ct Images ◽  
Head Ct ◽  
Learning Framework ◽  
Detection And Quantification

scholarly journals Multiclass semantic segmentation and quantification of traumatic brain injury lesions on head CT using deep learning: an algorithm development and multicentre validation study

2020 ◽  
Vol 2 (6) ◽  
pp. e314-e322 ◽  
Author(s):  
Miguel Monteiro ◽  
Virginia F J Newcombe ◽  
Francois Mathieu ◽  
Krishma Adatia ◽  
Konstantinos Kamnitsas ◽  
...  
Keyword(s):  
Traumatic Brain Injury ◽  
Deep Learning ◽  
Brain Injury ◽  
Validation Study ◽  
Semantic Segmentation ◽  
Algorithm Development ◽  
Head Ct

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 ResBCDU-Net: A Deep Learning Framework for Lung CT Image Segmentation

Sensors ◽  
2021 ◽  
Vol 21 (1) ◽  
pp. 268
Author(s):  
Yeganeh Jalali ◽  
Mansoor Fateh ◽  
Mohsen Rezvani ◽  
Vahid Abolghasemi ◽  
Mohammad Hossein Anisi
Keyword(s):  
Image Segmentation ◽  
Deep Learning ◽  
Ct Images ◽  
Medical Image Segmentation ◽  
Great Success ◽  
Morphological Operations ◽  
Ct Image ◽  
Feature Maps ◽  
Learning Framework ◽  
Lung Ct

Lung CT image segmentation is a key process in many applications such as lung cancer detection. It is considered a challenging problem due to existing similar image densities in the pulmonary structures, different types of scanners, and scanning protocols. Most of the current semi-automatic segmentation methods rely on human factors therefore it might suffer from lack of accuracy. Another shortcoming of these methods is their high false-positive rate. In recent years, several approaches, based on a deep learning framework, have been effectively applied in medical image segmentation. Among existing deep neural networks, the U-Net has provided great success in this field. In this paper, we propose a deep neural network architecture to perform an automatic lung CT image segmentation process. In the proposed method, several extensive preprocessing techniques are applied to raw CT images. Then, ground truths corresponding to these images are extracted via some morphological operations and manual reforms. Finally, all the prepared images with the corresponding ground truth are fed into a modified U-Net in which the encoder is replaced with a pre-trained ResNet-34 network (referred to as Res BCDU-Net). In the architecture, we employ BConvLSTM (Bidirectional Convolutional Long Short-term Memory)as an advanced integrator module instead of simple traditional concatenators. This is to merge the extracted feature maps of the corresponding contracting path into the previous expansion of the up-convolutional layer. Finally, a densely connected convolutional layer is utilized for the contracting path. The results of our extensive experiments on lung CT images (LIDC-IDRI database) confirm the effectiveness of the proposed method where a dice coefficient index of 97.31% is achieved.

A deep Residual U-Net algorithm for automatic detection and quantification of ascites on the abdominopelvic computed tomography acquired in the emergency department (Preprint)

2021 ◽  
Author(s):  
Hoon Ko ◽  
Jimi Huh ◽  
Kyung Won Kim ◽  
Heewon Chung ◽  
Yousun Ko ◽  
...  
Keyword(s):  
Computed Tomography ◽  
Deep Learning ◽  
Automatic Detection ◽  
High Specificity ◽  
Ct Images ◽  
Free Fluid ◽  
Detection Accuracy ◽  
Abdominopelvic Ct ◽  
Segmentation Accuracy ◽  
Detection And Quantification

BACKGROUND Detection and quantification of intraabdominal free fluid (i.e., ascites) on computed tomography (CT) are essential processes to find emergent or urgent conditions in patients. In an emergent department, automatic detection and quantification of ascites will be beneficial. OBJECTIVE We aimed to develop an artificial intelligence (AI) algorithm for the automatic detection and quantification of ascites simultaneously using a single deep learning model (DLM). METHODS 2D deep learning models (DLMs) based on a deep residual U-Net, U-Net, bi-directional U-Net, and recurrent residual U-net were developed to segment areas of ascites on an abdominopelvic CT. Based on segmentation results, the DLMs detected ascites by classifying CT images into ascites images and non-ascites images. The AI algorithms were trained using 6,337 CT images from 160 subjects (80 with ascites and 80 without ascites) and tested using 1,635 CT images from 40 subjects (20 with ascites and 20 without ascites). The performance of AI algorithms was evaluated for diagnostic accuracy of ascites detection and for segmentation accuracy of ascites areas. Of these DLMs, we proposed an AI algorithm with the best performance. RESULTS The segmentation accuracy was the highest in the deep residual U-Net with a mean intersection over union (mIoU) value of 0.87, followed by U-Net, bi-directional U-Net, and recurrent residual U-net (mIoU values 0.80, 0.77, and 0.67, respectively). The detection accuracy was the highest in the deep residual U-net (0.96), followed by U-Net, bi-directional U-net, and recurrent residual U-net (0.90, 0.88, and 0.82, respectively). The deep residual U-net also achieved high sensitivity (0.96) and high specificity (0.96). CONCLUSIONS We propose the deep residual U-net-based AI algorithm for automatic detection and quantification of ascites on abdominopelvic CT scans, which provides excellent performance.

scholarly journals Head CT overuse in children with a mild traumatic brain injury within two Canadian emergency departments

2019 ◽  
Vol 25 (1) ◽  
pp. 26-32
Author(s):  
Martin Gariepy ◽  
Jocelyn Gravel ◽  
France Légaré ◽  
Edward R Melnick ◽  
Erik P Hess ◽  
...  
Keyword(s):  
Traumatic Brain Injury ◽  
Brain Injury ◽  
Mild Traumatic Brain Injury ◽  
Emergency Departments ◽  
Retrospective Chart Review ◽  
Trauma Centre ◽  
Diagnostic Tools ◽  
General Level ◽  
Head Ct ◽  
Significant Difference

Abstract Background The validated Pediatric Emergency Care Applied Network (PECARN) rule helps determine the relevance of a head computerized tomography (CT) for children with mild traumatic brain injury (mTBI). We sought to estimate the potential overuse of head CT within two Canadian emergency departments (EDs). Methods We conducted a retrospective chart review of children seen in 2016 in a paediatric Level I (site 1) and a general Level II (site 2) trauma centre. We reviewed charts to determine the appropriateness of head CT use according to the PECARN rule in a random subset of children presenting with head trauma. Simple descriptive statistics were applied. Results One thousand five hundred and forty-six eligible patients younger than 17 years consulted during the study period. Of the 203 randomly selected cases per setting, 16 (7.9%) and 24 (12%), respectively from sites 1 and 2 had a head CT performed. Based on the PECARN rule, we estimated the overuse for the younger group (<2 years) to be below 3% for both hospitals without significant difference between them. For the older group (≥2 years), the overuse rate was higher at site 2 (9.3%, 95% confidence interval [CI]: 4.8 to 17% versus 1.2%, 95% CI: 0.2 to 6.5%, P=0.03). Conclusion Both EDs demonstrated overuse rates below 10% although it was higher for the older group at site 2. Such low rates can potentially be explained by the university affiliation of both hospitals and by two Canadian organizations working to raise awareness among physicians about the overuse of diagnostic tools and dangers inherent to radiation.

scholarly journals Prognostic performance of computerized tomography scoring systems in civilian penetrating traumatic brain injury: an observational study

2019 ◽  
Vol 161 (12) ◽  
pp. 2467-2478 ◽  
Author(s):  
Matias Lindfors ◽  
Caroline Lindblad ◽  
David W. Nelson ◽  
Bo-Michael Bellander ◽  
Jari Siironen ◽  
...  
Keyword(s):  
Traumatic Brain Injury ◽  
Brain Injury ◽  
Computerized Tomography ◽  
Outcome Prediction ◽  
Unfavorable Outcome ◽  
Receiver Operating Curve ◽  
Excellent Performance ◽  
Head Ct ◽  
Severity Scores ◽  
Penetrating Traumatic Brain Injury

Abstract Background The prognosis of penetrating traumatic brain injury (pTBI) is poor yet highly variable. Current computerized tomography (CT) severity scores are commonly not used for pTBI prognostication but may provide important clinical information in these cohorts. Methods All consecutive pTBI patients from two large neurotrauma databases (Helsinki 1999–2015, Stockholm 2005–2014) were included. Outcome measures were 6-month mortality and unfavorable outcome (Glasgow Outcome Scale 1–3). Admission head CT scans were assessed according to the following: Marshall CT classification, Rotterdam CT score, Stockholm CT score, and Helsinki CT score. The discrimination (area under the receiver operating curve, AUC) and explanatory variance (pseudo-R2) of the CT scores were assessed individually and in addition to a base model including age, motor response, and pupil responsiveness. Results Altogether, 75 patients were included. Overall 6-month mortality and unfavorable outcome were 45% and 61% for all patients, and 31% and 51% for actively treated patients. The CT scores’ AUCs and pseudo-R2s varied between 0.77–0.90 and 0.35–0.60 for mortality prediction and between 0.85–0.89 and 0.50–0.57 for unfavorable outcome prediction. The base model showed excellent performance for mortality (AUC 0.94, pseudo-R2 0.71) and unfavorable outcome (AUC 0.89, pseudo-R2 0.53) prediction. None of the CT scores increased the base model’s AUC (p > 0.05) yet increased its pseudo-R2 (0.09–0.15) for unfavorable outcome prediction. Conclusion Existing head CT scores demonstrate good-to-excellent performance in 6-month outcome prediction in pTBI patients. However, they do not add independent information to known outcome predictors, indicating that a unique score capturing the intracranial severity in pTBI may be warranted.

Sign in / Sign up

Export Citation Format

Share Document