Cats or CAT scans: Transfer learning from natural or medical image source data sets?

Cheplygina, Veronika

doi:10.1016/j.cobme.2018.12.005

Cited by 49 publications

(35 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Transfer learning is another technique that can help overcome the limitation of small labelled data sets. It has found particular application for transferring pretrained CNNs for image recognition tasks or for NIR spectroscopy calibration transfer across spectrometers [75,76]. A model trained on another system, for example a laboratory or pilot scale model system, can be used to aid in the prediction of the state of the target system.…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

Monitoring Mixing Processes Using Ultrasonic Sensors and Machine Learning

Bowler

Bakalis

Watson

2020

Sensors

View full text Add to dashboard Cite

Mixing is one of the most common processes across food, chemical, and pharmaceutical manufacturing. Real-time, in-line sensors are required for monitoring, and subsequently optimising, essential processes such as mixing. Ultrasonic sensors are low-cost, real-time, in-line, and applicable to characterise opaque systems. In this study, a non-invasive, reflection-mode ultrasonic measurement technique was used to monitor two model mixing systems. The two systems studied were honey-water blending and flour-water batter mixing. Classification machine learning models were developed to predict if materials were mixed or not mixed. Regression machine learning models were developed to predict the time remaining until mixing completion. Artificial neural networks, support vector machines, long short-term memory neural networks, and convolutional neural networks were tested, along with different methods for engineering features from ultrasonic waveforms in both the time and frequency domain. Comparisons between using a single sensor and performing multisensor data fusion between two sensors were made. Classification accuracies of up to 96.3% for honey-water blending and 92.5% for flour-water batter mixing were achieved, along with R2 values for the regression models of up to 0.977 for honey-water blending and 0.968 for flour-water batter mixing. Each prediction task produced optimal performance with different algorithms and feature engineering methods, vindicating the extensive comparison between different machine learning approaches.

show abstract

Section: Discussionmentioning

confidence: 99%

“…For example, the optimised signal processing, network weights, or ML hyperparameter values from the first system can be used as initial training values for the target system. Alternatively, the outputs of the previously trained model applied to the target system may be used as inputs to a second model [75].…”

Section: Discussionmentioning

confidence: 99%

Monitoring Mixing Processes Using Ultrasonic Sensors and Machine Learning

Bowler

Bakalis

Watson

2020

Sensors

View full text Add to dashboard Cite

show abstract

“…We made our basis earlier that using the networks pre-trained on ImageNet might not provide us better feature representations that are supported by the research community. Veronika cheplygina [ 44 ] conducted a study to provide a stance on whether the use of ImageNet pre-training is useful in medical imaging studies or not. The conclusion of the said study was “it depends” suggesting that if the volume of the data is small then it’s better to use a pre-trained network rather than initializing the weights randomly, however, if the volume of data is enough then the network should be trained on medical images from scratch.…”

Section: Proposed Methodsmentioning

confidence: 99%

Triage of potential COVID-19 patients from chest X-ray images using hierarchical convolutional networks

Dev

Khowaja

Bist³

et al. 2021

Neural Comput & Applic

View full text Add to dashboard Cite

The current COVID-19 pandemic has motivated the researchers to use artificial intelligence techniques for a potential alternative to reverse transcription-polymerase chain reaction due to the limited scale of testing. The chest X-ray (CXR) is one of the alternatives to achieve fast diagnosis, but the unavailability of large-scale annotated data makes the clinical implementation of machine learning-based COVID detection difficult. Another issue is the usage of ImageNet pre-trained networks which does not extract reliable feature representations from medical images. In this paper, we propose the use of hierarchical convolutional network (HCN) architecture to naturally augment the data along with diversified features. The HCN uses the first convolution layer from COVIDNet followed by the convolutional layers from well-known pre-trained networks to extract the features. The use of the convolution layer from COVIDNet ensures the extraction of representations relevant to the CXR modality. We also propose the use of ECOC for encoding multiclass problems to binary classification for improving the recognition performance. Experimental results show that HCN architecture is capable of achieving better results in comparison with the existing studies. The proposed method can accurately triage potential COVID-19 patients through CXR images for sharing the testing load and increasing the testing capacity.

show abstract

“…The second group consists of studies wherein some of the early layers of the pre-trained model on large-scale natural image dataset were frozen and their weights kept unchanged while the final layers were finetuned [29]. This practice is based on the fact that the early layer features are more generic (e.g., edges), whereas the later-layer features are more specific to a particular task or dataset [17].…”

Section: Transfer Learningmentioning

confidence: 99%

Deep Learning Approaches for Detecting COVID-19 From Chest X-Ray Images: A Survey

et al. 2021

View full text Add to dashboard Cite

Chest X-ray (CXR) imaging is a standard and crucial examination method used for suspected cases of coronavirus disease (COVID-19). In profoundly affected or limited resource areas, CXR imaging is preferable owing to its availability, low cost, and rapid results. However, given the rapidly spreading nature of COVID-19, such tests could limit the efficiency of pandemic control and prevention. In response to this issue, artificial intelligence methods such as deep learning are promising options for automatic diagnosis because they have achieved state-of-the-art performance in the analysis of visual information and a wide range of medical images. This paper reviews and critically assesses the preprint and published reports between March and May 2020 for the diagnosis of COVID-19 via CXR images using convolutional neural networks and other deep learning architectures. Despite the encouraging results, there is an urgent need for public, comprehensive, and diverse datasets. Further investigations in terms of explainable and justifiable decisions are also required for more robust, transparent, and accurate predictions. INDEX TERMS Chest x-ray, coronavirus, COVID-19, deep learning, radiological imaging.

show abstract

Cats or CAT scans: Transfer learning from natural or medical image source data sets?

Cited by 49 publications

References 33 publications

Monitoring Mixing Processes Using Ultrasonic Sensors and Machine Learning

Monitoring Mixing Processes Using Ultrasonic Sensors and Machine Learning

Triage of potential COVID-19 patients from chest X-ray images using hierarchical convolutional networks

Deep Learning Approaches for Detecting COVID-19 From Chest X-Ray Images: A Survey

Contact Info

Product

Resources

About