An End-to-End Framework for Neonatal Asphyxia Detection Through Video Analysis

Identifying asphyxia using computer vision in real-world settings poses challenges due to varying video quality, diverse lighting conditions, and subtle color changes in the newborn’s skin. This study presents an end-to-end framework for automated neonatal asphyxia detection using time se...

Full description

Saved in:
Bibliographic Details
Main Authors: Kurnianingsih, Sou Nobukawa, Melyana Nurul Widyawati, Cipta Pramana, Nurseno Bayu Aji, Afandi Nur Aziz Thohari, Dwiana Hendrawati, Eri Sato-Shimokawara, Naoyuki Kubota
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/10975012/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850203779046047744
author Kurnianingsih
Sou Nobukawa
Melyana Nurul Widyawati
Cipta Pramana
Nurseno Bayu Aji
Afandi Nur Aziz Thohari
Dwiana Hendrawati
Eri Sato-Shimokawara
Naoyuki Kubota
author_facet Kurnianingsih
Sou Nobukawa
Melyana Nurul Widyawati
Cipta Pramana
Nurseno Bayu Aji
Afandi Nur Aziz Thohari
Dwiana Hendrawati
Eri Sato-Shimokawara
Naoyuki Kubota
author_sort Kurnianingsih
collection DOAJ
description Identifying asphyxia using computer vision in real-world settings poses challenges due to varying video quality, diverse lighting conditions, and subtle color changes in the newborn’s skin. This study presents an end-to-end framework for automated neonatal asphyxia detection using time series video analysis and makes three key contributions. First, the proposed framework integrates YOLOv8-based instance segmentation with advanced feature extraction across multiple color spaces and texture analysis to detect neonatal asphyxia through the multi-modal analysis of skin features in video streams. Second, we introduce a new quality-aware temporal analysis framework that includes adaptive quality assessment for evaluating frames in real time, multi-stage feature stability tracking across temporal windows, hysteresis-based decision logic for ensuring temporal consistency, and LightGBM classification with comprehensive feature engineering to assess severity. Third, we provide a curated time series video dataset of 12,973 frames from 45 neonates, of which some were healthy, and some had asphyxia of varying severity. The findings show that the YOLOv8-based instance segmentation achieved a mean average precision (mAP@0.5) of 0.925 for accurate skin region isolation, and the LightGBM classifier outperformed traditional models with an accuracy of 0.998 and an F1-score of 0.998. The system maintains real-time processing at 30 FPS for normal and mild asphyxia cases with a minor reduction to 20 FPS in more challenging scenarios and exhibits robust temporal stability across severity levels, with consistency scores above 0.90. This framework has the potential to enhance neonatal care through continuous monitoring and timely intervention.
format Article
id doaj-art-fcf00e5ff1e84363a12c14aebff54021
institution OA Journals
issn 2169-3536
language English
publishDate 2025-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-fcf00e5ff1e84363a12c14aebff540212025-08-20T02:11:25ZengIEEEIEEE Access2169-35362025-01-0113746897470410.1109/ACCESS.2025.356370210975012An End-to-End Framework for Neonatal Asphyxia Detection Through Video Analysis Kurnianingsih0https://orcid.org/0000-0001-7339-7449Sou Nobukawa1https://orcid.org/0000-0001-7003-6912Melyana Nurul Widyawati2https://orcid.org/0000-0003-2039-2892Cipta Pramana3https://orcid.org/0000-0001-8991-0147Nurseno Bayu Aji4https://orcid.org/0009-0005-7486-3524Afandi Nur Aziz Thohari5https://orcid.org/0000-0003-3468-1994Dwiana Hendrawati6https://orcid.org/0009-0008-1834-6209Eri Sato-Shimokawara7https://orcid.org/0000-0002-4189-9724Naoyuki Kubota8https://orcid.org/0000-0001-8829-037XDepartment of Electrical Engineering, Politeknik Negeri Semarang, Semarang, Central Java, IndonesiaDepartment of Computer Science, Chiba Institute of Technology, Narashino, Chiba, JapanDepartment of Postgraduate Program, Poltekkes Kemenkes Semarang, Semarang, Central Java, IndonesiaDepartment of Obstetrics and Gynecology, KRMT Wongsonegoro Hospital, Semarang, Central Java, IndonesiaDepartment of Electrical Engineering, Politeknik Negeri Semarang, Semarang, Central Java, IndonesiaDepartment of Electrical Engineering, Politeknik Negeri Semarang, Semarang, Central Java, IndonesiaDepartment of Mechanical Engineering, Politeknik Negeri Semarang, Semarang, Central Java, IndonesiaGraduate School of Systems Design, Tokyo Metropolitan University, Hino, Tokyo, JapanGraduate School of Systems Design, Tokyo Metropolitan University, Hino, Tokyo, JapanIdentifying asphyxia using computer vision in real-world settings poses challenges due to varying video quality, diverse lighting conditions, and subtle color changes in the newborn’s skin. This study presents an end-to-end framework for automated neonatal asphyxia detection using time series video analysis and makes three key contributions. First, the proposed framework integrates YOLOv8-based instance segmentation with advanced feature extraction across multiple color spaces and texture analysis to detect neonatal asphyxia through the multi-modal analysis of skin features in video streams. Second, we introduce a new quality-aware temporal analysis framework that includes adaptive quality assessment for evaluating frames in real time, multi-stage feature stability tracking across temporal windows, hysteresis-based decision logic for ensuring temporal consistency, and LightGBM classification with comprehensive feature engineering to assess severity. Third, we provide a curated time series video dataset of 12,973 frames from 45 neonates, of which some were healthy, and some had asphyxia of varying severity. The findings show that the YOLOv8-based instance segmentation achieved a mean average precision (mAP@0.5) of 0.925 for accurate skin region isolation, and the LightGBM classifier outperformed traditional models with an accuracy of 0.998 and an F1-score of 0.998. The system maintains real-time processing at 30 FPS for normal and mild asphyxia cases with a minor reduction to 20 FPS in more challenging scenarios and exhibits robust temporal stability across severity levels, with consistency scores above 0.90. This framework has the potential to enhance neonatal care through continuous monitoring and timely intervention.https://ieeexplore.ieee.org/document/10975012/Neonatal asphyxiavideo analysisinstance segmentationdeep learningmedical imaging
spellingShingle Kurnianingsih
Sou Nobukawa
Melyana Nurul Widyawati
Cipta Pramana
Nurseno Bayu Aji
Afandi Nur Aziz Thohari
Dwiana Hendrawati
Eri Sato-Shimokawara
Naoyuki Kubota
An End-to-End Framework for Neonatal Asphyxia Detection Through Video Analysis
IEEE Access
Neonatal asphyxia
video analysis
instance segmentation
deep learning
medical imaging
title An End-to-End Framework for Neonatal Asphyxia Detection Through Video Analysis
title_full An End-to-End Framework for Neonatal Asphyxia Detection Through Video Analysis
title_fullStr An End-to-End Framework for Neonatal Asphyxia Detection Through Video Analysis
title_full_unstemmed An End-to-End Framework for Neonatal Asphyxia Detection Through Video Analysis
title_short An End-to-End Framework for Neonatal Asphyxia Detection Through Video Analysis
title_sort end to end framework for neonatal asphyxia detection through video analysis
topic Neonatal asphyxia
video analysis
instance segmentation
deep learning
medical imaging
url https://ieeexplore.ieee.org/document/10975012/
work_keys_str_mv AT kurnianingsih anendtoendframeworkforneonatalasphyxiadetectionthroughvideoanalysis
AT sounobukawa anendtoendframeworkforneonatalasphyxiadetectionthroughvideoanalysis
AT melyananurulwidyawati anendtoendframeworkforneonatalasphyxiadetectionthroughvideoanalysis
AT ciptapramana anendtoendframeworkforneonatalasphyxiadetectionthroughvideoanalysis
AT nursenobayuaji anendtoendframeworkforneonatalasphyxiadetectionthroughvideoanalysis
AT afandinurazizthohari anendtoendframeworkforneonatalasphyxiadetectionthroughvideoanalysis
AT dwianahendrawati anendtoendframeworkforneonatalasphyxiadetectionthroughvideoanalysis
AT erisatoshimokawara anendtoendframeworkforneonatalasphyxiadetectionthroughvideoanalysis
AT naoyukikubota anendtoendframeworkforneonatalasphyxiadetectionthroughvideoanalysis
AT kurnianingsih endtoendframeworkforneonatalasphyxiadetectionthroughvideoanalysis
AT sounobukawa endtoendframeworkforneonatalasphyxiadetectionthroughvideoanalysis
AT melyananurulwidyawati endtoendframeworkforneonatalasphyxiadetectionthroughvideoanalysis
AT ciptapramana endtoendframeworkforneonatalasphyxiadetectionthroughvideoanalysis
AT nursenobayuaji endtoendframeworkforneonatalasphyxiadetectionthroughvideoanalysis
AT afandinurazizthohari endtoendframeworkforneonatalasphyxiadetectionthroughvideoanalysis
AT dwianahendrawati endtoendframeworkforneonatalasphyxiadetectionthroughvideoanalysis
AT erisatoshimokawara endtoendframeworkforneonatalasphyxiadetectionthroughvideoanalysis
AT naoyukikubota endtoendframeworkforneonatalasphyxiadetectionthroughvideoanalysis