GANs for medical image analysis

Kazeminia, Salome; Baur, Christoph; Kuijper, Arjan; Ginneken, Bram van; Navab, Nassir; Albarqouni, Shadi; Mukhopadhyay, Anirban

doi:10.1016/j.artmed.2020.101938

Cited by 313 publications

(200 citation statements)

References 84 publications

Supporting

Mentioning

170

Contrasting

Unclassified

Order By: Relevance

“…Although some methods, such as motion correction or the bootstrap strategy, used in this study may be helpful to speed up the labeling process, developers of cardiac deep learning applications still spend significant hours to establish labeled training sets 61 . Recent development of synthetic data labeling can learn a generative model from a small training set and synthesize more training data by sampling the learned probability distribution 62 . This strategy is promising although also requires expert supervision to verify the generated labels are correct.…”

Section: Discussionmentioning

confidence: 99%

“…61 Recent development of synthetic data labeling can learn a generative model from a small training set and synthesize more training data by sampling the learned probability distribution. 62 This strategy is promising although also requires expert supervision to verify the generated labels are correct. Another related method to mitigate data labeling workload is through transfer learning.…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

Automated detection of left ventricle in arterial input function images for inline perfusion mapping using deep learning: A study of 15,000 patients

Xue

Tseng

Knott

et al. 2020

Magnetic Resonance in Med

View full text Add to dashboard Cite

Purpose Quantification of myocardial perfusion has the potential to improve the detection of regional and global flow reduction. Significant effort has been made to automate the workflow, where one essential step is the arterial input function (AIF) extraction. Failure to accurately identify the left ventricle (LV) prevents AIF estimation required for quantification, therefore high detection accuracy is required. This study presents a robust LV detection method using the convolutional neural network (CNN). Methods CNN models were trained by assembling 25,027 scans (N = 12,984 patients) from three hospitals, seven scanners. Performance was evaluated using a hold‐out test set of 5721 scans (N = 2805 patients). Model inputs were a time series of AIF images (2D+T). Two variations were investigated: (1) two classes (2CS) for background and foreground (LV mask), and (2) three classes (3CS) for background, LV, and RV. The final model was deployed on MRI scanners using the Gadgetron reconstruction software framework. Results Model loading on the MRI scanner took ~340 ms and applying the model took ~180 ms. The 3CS model successfully detected the LV in 99.98% of all test cases (1 failure out of 5721). The mean Dice ratio for 3CS was 0.87 ± 0.08 with 92.0% of all cases having Dice >0.75. The 2CS model gave a lower Dice ratio of 0.82 ± 0.22 (P < 1e−5). There was no significant difference in foot‐time, peak‐time, first‐pass duration, peak value, and area‐under‐curve (P > .2) comparing automatically extracted AIF signals with signals from manually drawn contours. Conclusions A CNN‐based solution to detect the LV blood pool from the arterial input function image series was developed, validated, and deployed. A high LV detection accuracy of 99.98% was achieved.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

Automated detection of left ventricle in arterial input function images for inline perfusion mapping using deep learning: A study of 15,000 patients

Xue

Tseng

Knott

et al. 2020

Magnetic Resonance in Med

View full text Add to dashboard Cite

show abstract

“…Automated image analysis (Kazeminia et al, 2020) and object detection/recognition (Liu et al, 2019) have become popular as well, and there are more and more services that the users can read their images into and the services will recognize the scene, parts of the scene and details on the images. Likewise, speech-to-text services (Myers, 2020) enable voice-based input of search queries when using search engines, as spoken input gets automatically converted to text (the search string).…”

Section: Examples Of Aimentioning

confidence: 99%

Quo Vadis, AI?

Devedžić

2021

Manage: J. Sustainable Bus. Manage, Solutions Emerging Econ.

View full text Add to dashboard Cite

Research Question: This article reflects on currently popular topics and future development and application of AI. Motivation: AI has become a key technology in many areas – from business and services, to science engineering and engineering, natural language processing and news, publishing and writing, human resource management and education, medicine and healthcare, transportation and aviation, government, military and agriculture. Idea: As the field advances and expands at an unprecedented pace, it becomes more and more difficult to distinguish between AI and other fields. Data: The article is based on a qualitative analysis of relevant literature and direct insights into the field of AI during the last 35 years. Tools: The article uses direct observation and content analysis to derive conclusions. Findings: Just like any other technology, AI has both its bright and its dark sides. There is a number of expectations, hopes and promises of AI that still remain to be fulfilled, yet new avenues for further development open day after day. Contribution: The article attempts to put AI developments in different contexts and shed different lights on AI applications in general, AI applications in business, in particular, and current limitations, trends and challenges in the field.

show abstract

“…There are different types of DCGA N used in the medical field. Kazemin ia et al [2] exp lained about the different types of GANs and DCGAN as shown in solves the following problems in medical image analysis like synthesis, segmentation, reconstruction, detection, denoising, registration and classification and improves overall image resolution. The major difference between the basic GA N [5] and DCGAN is that a GAN is supervised learning, not stable, used only for generating small data sets and also cannot detect overfitting, so to overcome the limitations of the basic GAN architecture and to improve the resolution of synthesized images, Radford et al [6] (2015) proposed Deep Convolutional GAN (DCGAN) a deep learning-based generative model which is unsupervised learning and does not require any labels.…”

Section: Deep Convolutional Generative Adversarial Neural Networmentioning

confidence: 99%

A Survey on Deep Convolutional Generative Adversarial Neural Network (DCGAN) for Detection of Covid-19 using Chest X-ray/CT-Scan

Bushra

Shobana²

2020

2020 3rd International Conference on Intelligent Sustainable Systems (ICISS)

View full text Add to dashboard Cite

Artificial Neural Network is a domain where an abundant amount of data is needed to train the neural network model to predict the output precisely. But in a real scenario, there are many application areas like astronomy, clinical dataset, rare species identification, financial datasets, and satellite images where data are scarce the network is unable to predict unusual classes of events. The environment in which we live is uncertain, unpredictable and in the future, we may face more challenges. To overcome this, we have the most predominant image and video analysis technique called Generative Adversarial Neural Network (GAN) which is very useful for any sparse datasets especially in pandemic situations like COVID-19 and also for any future events for which we don't have enough datasets to train the model. This paper aims to overview the details about the systems recently developed to diagnose novel COVID-19 with the help of X-ray and CT-scan images collected from different infected persons using one of the most important branches of Deep learning technique known as DCGAN(Deep Convolutional Generative Adversarial Neural Network) for detecting various new emerging diseases in the medical field for which dataset is very less and DCGAN acts as an efficient tool for generating synthetic data as there is scarcity in the availability of data for diagnosing coronavirus. İt also describes the datasets taken from various publicly available sources so that a large number of image datasets can be trained and tested in various state-of-art deep network architectures to check the performance and efficiency of each model in terms of accuracy, complexity, and time of execution. These results help radiologists to quickly identify the disease without any further delay in spreading and infecting others.

show abstract

GANs for medical image analysis

Cited by 313 publications

References 84 publications

Automated detection of left ventricle in arterial input function images for inline perfusion mapping using deep learning: A study of 15,000 patients

Automated detection of left ventricle in arterial input function images for inline perfusion mapping using deep learning: A study of 15,000 patients

Quo Vadis, AI?

A Survey on Deep Convolutional Generative Adversarial Neural Network (DCGAN) for Detection of Covid-19 using Chest X-ray/CT-Scan

Contact Info

Product

Resources

About