Vascular medicine (London, England) · 2022
Machine learning-based classification of arterial spectral waveforms for the diagnosis of peripheral artery disease in the context of diabetes: A proof-of-concept study.
СтатьяПоддержка исследования: вне правительства СШАНазвание и аннотация: Перевод пока не подготовлен
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Normahani Pasha, Sounderajah Viknesh, Mandic Danilo, Jaffer Usman
Аннотация
BACKGROUND: Point-of-care duplex ultrasound has emerged as a promising test for the diagnosis of peripheral artery disease (PAD). However, the interpretation of morphologically diverse Doppler arterial spectral waveforms is challenging and associated with wide inter-observer variation. The aim of this study is to evaluate the utility of machine learning techniques for the diagnosis of PAD from Doppler arterial spectral waveforms sampled at the level of the ankle in patients with diabetes.
METHODS: In two centres, 590 Doppler arterial spectral waveform images (PAD 369, no-PAD 221) from 305 patients were prospectively collected. Doppler arterial spectral waveform signals were reconstructed. Blinded full lower-limb reference duplex ultrasound results were used to label waveform according to PAD status (i.e., PAD, no-PAD). Statistical metrics and multiscale wavelet variance were extracted as discriminatory features. A long short-term memory (LSTM) network was used for the classification of raw signals, and logistic regression (LR) and support vector machines (SVM) were used for classification of extracted features. Signals and feature vectors were randomly divided into training (80%) and testing (20%) sets.
RESULTS: The highest overall accuracy was achieved using a logistic regression model with a combination of statistical and multiscale wavelet variance features, with 88% accuracy, 92% sensitivity, and 82% specificity. The area under the receiver operating characteristics curve (AUC) was 0.93.
CONCLUSION: We have constructed a machine learning algorithm with high discriminatory ability for the diagnosis of PAD using Doppler arterial spectral waveforms sampled at the ankle vessels.
METHODS: In two centres, 590 Doppler arterial spectral waveform images (PAD 369, no-PAD 221) from 305 patients were prospectively collected. Doppler arterial spectral waveform signals were reconstructed. Blinded full lower-limb reference duplex ultrasound results were used to label waveform according to PAD status (i.e., PAD, no-PAD). Statistical metrics and multiscale wavelet variance were extracted as discriminatory features. A long short-term memory (LSTM) network was used for the classification of raw signals, and logistic regression (LR) and support vector machines (SVM) were used for classification of extracted features. Signals and feature vectors were randomly divided into training (80%) and testing (20%) sets.
RESULTS: The highest overall accuracy was achieved using a logistic regression model with a combination of statistical and multiscale wavelet variance features, with 88% accuracy, 92% sensitivity, and 82% specificity. The area under the receiver operating characteristics curve (AUC) was 0.93.
CONCLUSION: We have constructed a machine learning algorithm with high discriminatory ability for the diagnosis of PAD using Doppler arterial spectral waveforms sampled at the ankle vessels.