Application of Dynamic Time Warping to Determine the Shear Wave Velocity from the Down-Hole Test

Duda-Mróz, Natalia; Koperska, Wioletta; Stefaniak, Paweł; Anufriiev, Sergii; Stachowiak, Maria; Stefanek, Paweł

doi:10.3390/app13179736

Cited by 1 publication

(2 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…N. Duda-Mróz [18] proposed a steganographic technique based on dynamic time warping (DTW) to embed secret messages within the ECG signal. The authors used DTW to find similar segments in the ECG signal and modified the amplitude of these segments to carry the hidden information.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Robust and Efficient ECG Classification with Privacy Preservation using LSTM+CNN and Adaptive Least Signal Bit

Sakr,

Ahmed,

Rezk

2023

IJCI. International Journal of Computers and Information

View full text Add to dashboard Cite

Maintaining patient data privacy is a critical concern in the medical domain. However, safeguarding crucial information during data transmission presents significant challenges for healthcare professionals and organizations. Additionally, the accurate classification of electrocardiogram (ECG) signals is hindered by the need for supplementary information to ensure precise examinations and accurate diagnoses. This paper introduces a novel hybrid framework that addresses both ECG classification challenges and privacy preservation. The proposed framework consists of two main phases. In the first phase, a combination of Long Short-Term Memory (LSTM) and Convolutional Neural Network (CNN) with an attention gate is employed for ECG classification. This phase enhances the accuracy of ECG signal classification by incorporating deep learning techniques. The second phase of the framework utilizes adaptive least signal bit with neutrosophic logic to conceal critical medical data during transmission. Neutrosophic sets, which represent data as degrees of truth, falsehood, and indeterminacy, are leveraged and passed through an embedding layer. In the sender part, important data is converted into three degrees and embedded within the ECG signal using true and false degrees. An intermediate set acts as a shared dynamic key between the sender and receiver. The receiver can reconstruct the important data using either the shared dynamic key or the intermediate set. The experiments done prove the efficiency of different parts of the framework; the augmentation model archives 14.67 and 37.89 for the inception score and freshet inception distance, respectively. The hybrid approach also achieves 99.01 and 99.12 for dice score and accuracy, respectively.

show abstract

Section: Related Workmentioning

confidence: 99%

“…Secondly, data augmentation mitigates the problem of limited labeled data, allowing the model to learn from a larger and more diverse dataset, thus reducing overfitting and improving model performance. Lastly, data augmentation can increase the robustness of the model by exposing it to a broader range of ECG signal variations [17,18].…”

Section: Introductionmentioning

confidence: 99%