AI-enabled workflow for automated classification and analysis of feto-placental Doppler images DOI Creative Commons
Ainhoa M. Aguado, Guillermo Jiménez-Pérez, Devyani Chowdhury

et al.

Frontiers in Digital Health, Journal Year: 2024, Volume and Issue: 6

Published: Oct. 16, 2024

Introduction Extraction of Doppler-based measurements from feto-placental Doppler images is crucial in identifying vulnerable new-borns prenatally. However, this process time-consuming, operator dependent, and prone to errors. Methods To address this, our study introduces an artificial intelligence (AI) enabled workflow for automating four sites (i.e., Umbilical Artery (UA), Middle Cerebral (MCA), Aortic Isthmus (AoI) Left Ventricular Inflow Outflow (LVIO)), involving classification waveform delineation tasks. Derived data a low- middle-income country, approach's versatility was tested validated using dataset high-income showcasing its potential standardized accurate analysis across varied healthcare settings. Results The views approached through three distinct blocks: (i) velocity amplitude-based model with accuracy 94%, (ii) two Convolutional Neural Networks (CNN) accuracies 89.2% 67.3%, (iii) view- dataset-dependent confidence models detect misclassifications higher than 85%. extraction indices utilized Doppler-view dependent CNNs coupled post-processing techniques. yielded mean absolute percentage error 6.1 ± 4.9% ( n = 682), 1.8 1.5% 1,480), 4.7 4.0% 717), 3.5 3.1% 1,318) the magnitude location systolic peak LVIO, UA, AoI MCA views, respectively. Conclusions developed proved be highly classifying extracting essential images. integration AI-enabled holds significant promise reducing manual workload enhancing efficiency image analysis, even non-trained readers.

Language: Английский

AI-enabled workflow for automated classification and analysis of feto-placental Doppler images DOI Creative Commons
Ainhoa M. Aguado, Guillermo Jiménez-Pérez, Devyani Chowdhury

et al.

Frontiers in Digital Health, Journal Year: 2024, Volume and Issue: 6

Published: Oct. 16, 2024

Introduction Extraction of Doppler-based measurements from feto-placental Doppler images is crucial in identifying vulnerable new-borns prenatally. However, this process time-consuming, operator dependent, and prone to errors. Methods To address this, our study introduces an artificial intelligence (AI) enabled workflow for automating four sites (i.e., Umbilical Artery (UA), Middle Cerebral (MCA), Aortic Isthmus (AoI) Left Ventricular Inflow Outflow (LVIO)), involving classification waveform delineation tasks. Derived data a low- middle-income country, approach's versatility was tested validated using dataset high-income showcasing its potential standardized accurate analysis across varied healthcare settings. Results The views approached through three distinct blocks: (i) velocity amplitude-based model with accuracy 94%, (ii) two Convolutional Neural Networks (CNN) accuracies 89.2% 67.3%, (iii) view- dataset-dependent confidence models detect misclassifications higher than 85%. extraction indices utilized Doppler-view dependent CNNs coupled post-processing techniques. yielded mean absolute percentage error 6.1 ± 4.9% ( n = 682), 1.8 1.5% 1,480), 4.7 4.0% 717), 3.5 3.1% 1,318) the magnitude location systolic peak LVIO, UA, AoI MCA views, respectively. Conclusions developed proved be highly classifying extracting essential images. integration AI-enabled holds significant promise reducing manual workload enhancing efficiency image analysis, even non-trained readers.

Language: Английский

Citations

1