Skip to content

Home

Jump to bottom
utsb-fmm edited this page Jul 21, 2022 · 14 revisions

Introduction

Dataset/challenge

Toolbox

Leaderboard

The WMFB method

FS detection project

Clone this wiki locally

FHRMA : Fetal Heart Rate Morphological Analysis toolbox and challenge

Content

Related papers and citation rules

This toolbox is related to several papers. Please cite those papers if you use any of the data or source code of this toolbox. [4] must be cited if you use the toolbox. [1] must be cited if you use the morphological analysis (baseline, Acceleration, deceleration) [3] must be cited if you use the morphological analysis dataset. [5] must be cited if you use the WMFB method (current best) for morphological analysis. [6] must be cited if you use the false signal detection, method, interface and/or dataset.

[1] Houzé de l’Aulnoit, A., Boudet, S., Demailly, R., Delgranche, A., Peyrodie, L., Beuscart, R., Houzé de l’Aulnoit,D. - Automated fetal heart rate analysis for baseline determination and acceleration/deceleration detection: A comparison of 11 methods versus expert consensus. Biomedical Signal Processing and Control 49:113 -123,2019, DOI:10.1016/j.bspc.2018.10.002 Download on Researchgate

[2] Houzé de l'Aulnoit, Agathe, Boudet, Samuel, Demailly, Romain, Peyrodie, Laurent, Beuscart, Regis, Houzé de l'Aulnoit, Denis - Baseline fetal heart rate analysis: eleven automatic methods versus expert consensus. Engineering in Medicine and Biology Society (EMBC), 2016 IEEE 38th Annual International Conference of the pp. 3576--3581,2016, DOI:10.1109/EMBC.2016.7591501 Download on researchgate

[3] Boudet, S., Houzé de l’Aulnoit, A., Demailly, R., Delgranche, A., Peyrodie, L., Beuscart, R., Houzé de l’Aulnoit,D. - Fetal heart rate signal dataset for training morphological analysis methods and evaluating them against an expert consensus. Preprints pp. Submitted to data in brief,2019, DOI:10.20944/preprints201907.0039.v1 Download on researchgate

[4] Boudet, S., Houzé de l’Aulnoit, A., Demailly, R., Delgranche, A., Peyrodie, L., Beuscart, R., Houzé de l’Aulnoit,D. - A fetal heart rate morphological analysis toolbox for MATLAB. SoftwareX. 2020 Jan 1;11:100428. DOI:10.1016/j.softx.2020.100428 Download on researchgate

[5] Boudet, S., Houzé de l’Aulnoit, A., Demailly, R., Peyrodie, L., Beuscart, R., Houzé de l’Aulnoit,D. - Fetal heart rate baseline computation with a weighted median filter. Computers in biology and medicine. 2019 Nov 1;114:103468. DOI:10.1016/j.compbiomed.2019.103468 Download on researchgate

[6] Boudet, S., Houzé de l’Aulnoit, A., Demailly, R., Peyrodie, L., Houzé de l’Aulnoit,D. - Use of deep learning to detect the maternal heart rate and false signals on fetal heart rate recordings. preprints 2022. DOI:10.20944/preprints202207.0131.v1 Download on researchgate

Introduction

The FHR analyse is an important diagnostic element of fetal acidosis during the first step of delivery. There is a very important variability of inter-expert and intra-expert interpretation due to criteria sometines fuzzy and miss of pratician formation. For this reason, numerous researcher have worked on methods to automatically analyse the FHR signals and predict the risk of acidosis.

The analysis of FHR rest on the determiniation of elementary parameters :

  • baseline which is the mean level of FHR in stable periods
  • variabilty which correspond to the size of variation in stable periods
  • presence of accelerations which correspond to temporary increase of FHR
  • presence of decelarations which correspond to temporary decrease FHR
  • presence of sinusoidal pattern

Determination of baseline, accelerations and decelerations is refered as morphological analysis. There is a cyclic defintion on this problem : to locate the baseline it is required to remove accidents (accelerations and decelerations) and to detect accidents it is required to know the baseline to detect periods outside the normal variability of signal.

A most problematic case would be for example a signal which alternate between 170 bpm and 140 bpm. If the baseline is considered at 170 bpm, there would be a tachycardia and the period at 140 bpm would then be decelerations; the rhythms would then be non reassuring. If the baseline is considered at 140 bpm, the periods at 170 bpm will then be accelerations and the rhythm is then perfectly healthy.

Determining the baseline is then the first step for automated analyse of FHR signal, if the baseline is perfectly defined the remaining of other parmeters can be determine relatively easily. For now, the shared ressources (interface, dataset) concerns only the reproduction of expert morphological analysis. There is not yet clinical output criteria. For this, you can have a look to CTG-UHB dataset.

You can have a look to our presentation at the Signal Processing and Monitoring in Labour Workshop in 2019 https://youtu.be/gjpBs4utlbM

FHR can captures sometimes false signals (FS) wich can be either Maternal reart rate (MHR) interferences, or harmonics (double or half) of the FHR or MHR. This toolbox includes a method, an interfacte and a labelled dataset for the problem of FS detection which should be carried out as a preprocessing.

Contact

samuel.boudet[at]univ-catholille.fr