A Unified Framework for Quality Indexing and Classification of Seismocardiogram Signals
Executive Summary
This study introduces a unified framework for quality indexing and classification of seismocardiogram (SCG) signals using dynamic-time feature matching (DTFM), a novel method for signal distance estimation. The framework defines a Signal Quality Index (SQI) based on the inverse distance between SCG signals and reference templates, and employs ensembled quadratic discriminant analysis (QDA) for classification tasks, including sensor misplacement detection. The method achieved an F1 score of 0.83 for misplacement detection and demonstrated improved signal stratification compared to traditional dynamic time warping (DTW). These findings address key limitations in SCG processing, paving the way for automated and robust clinical applications.
Answer Machine Insights
Q: What is the main advantage of DTFM over DTW in SCG signal processing?
DTFM prioritizes feature correspondence over distance minimization, leading to more accurate alignment of SCG signal features.
DTFM aligns corresponding peaks in the warped signal, reducing the likelihood of peak misalignment compared to DTW.
Q: How does the ensembled QDA classifier perform in detecting SCG sensor misplacement?
The ensembled QDA classifier achieved an F1 score of 0.83 for detecting SCG sensor misplacement using held-out validation.
For binary classification, DTFM-based classifiers achieved an F1 score of 0.83; these results are comparable to the F1 score of 0.82 achieved in prior work.
Key Results
DTFM-based SQI achieved significant stratification of SCG signals across activity levels, outperforming DTW.
Ensembled QDA classifier achieved an F1 score of 0.83 for detecting SCG sensor misplacement on held-out subjects.
Visual Evidence
![Fig. 5. (a) Stratification of SCG signals using the DTFM-based SQI for a single subject. The columns from left to right show SCG intervals taken from rest, recovery, and exercise respectively. The top row corresponds to the top 2% of signals in each category based on their SQI; the middle rows correspond to the 50th percentile; and the bottom row corresponds to the bottom 2%. The correlation between scores assigned by manual annotation and the DTFM-based SQI are shown for the (b) rest, (c) recovery, (d) squatting, and (e) walking activity levels. The best-fit line for these scores is shown (black, dotted), with the corresponding R2 of the fit overlaid. For visual purposes, SQI scores were normalized to the range [0, 1].](/_next/image?url=%2Fscience-hub%2Fimages%2F10.1109_jbhi.2019.2931348%2F10.1109_jbhi.2019.2931348_p9_i1.png&w=3840&q=75)
Fig. 5. (a) Stratification of SCG signals using the DTFM-based SQI for a single subject. The columns from left to right show SCG intervals taken from rest, recovery, and exercise respectively. The top row corresponds to the top 2% of signals in each category based on their SQI; the middle rows correspond to the 50th percentile; and the bottom row corresponds to the bottom 2%. The correlation between scores assigned by manual annotation and the DTFM-based SQI are shown for the (b) rest, (c) recovery, (d) squatting, and (e) walking activity levels. The best-fit line for these scores is shown (black, dotted), with the corresponding R2 of the fit overlaid. For visual purposes, SQI scores were normalized to the range [0, 1].
Clinical Snapshot
Evidence Rating
Relevance
high Priority