Table 1.

Neurosonographic studies related to artificial intelligence.

Reference, Year	Country	GA (wks)	Study Size (n) *	Data Source	Type of Method	Purpose /Target	Task	Description of AI	Clinical Value ***
Rizzo et al., 2016 [34]	I	21 (mean)	120	3D	n. s.	SFHP (axial) biometry	automated recognition of axial planes from 3D volumes	5D CNS software	++
Rizzo et al., 2016 [35] **	I	18–24	183	3D	n. s.	SFHP (axial/ sagittal/coronal) biometry	evaluation of efficacy in reconstructing CNS planes in healthy and abnormal fetuses	5D CNS+ software	+++
Ambroise-Grandjean et al., 2018 [36]	F	17–30	30	3D	n. s.	SFHP (axial) biometry (TT, TC)	automated identification of axial from 3DUS and measurement BPD and HC	SmartPlanes CNS	++
Welp et al., 2020 [30] **	D	15–36	1110	3D	n. s.	SFHP (axial/ sagittal/coronal) biometry	validating of a volumetric approach for the detailed assessment of the fetal brain	5D CNS+ software	+++
Pluym et al., 2021 [37]	USA	18–22	143	3D	n. s.	SFHP (axial) biometry	evaluation of accuracy of automated 3DUS for fetal intracranial measurements	SonoCNS software	++
Welp et al., 2022 [29] **	D	16–35	91	3D	n. s.	SFHP/anomalies biometry	evaluation of accuracy and reliability of a volumetric approach in abnormal CNSs	5D CNS+ software	+++
Gembicki et al., 2023 [28] **	D	18–36	129	3D	n. s.	SFHP (axial/ sagittal/coronal) biometry	evaluation of accuracy and efficacy of AI-assisted biometric measurements of the fetal CNS	5D CNS+ software, SonoCNS software	++
Han et al., 2024 [38]	CHN	18–42	642	2D	DL	Biometry (incl. HC, BPD, FOD, CER, CM, Vp)	automated measurement and quality assessment of nine biometric parameters	CUPID software	++
Yaqub et al., 2012 [39]	UK	19–24	30	3D	ML	multi-structure detection	localization of four local brain structures in 3D US images	Random Forest Classifier	++
Cuingnet et al., 2013 [40]	UK	19–24	78 volumes	3D	ML	SFHP	fully automatic method to detect and align fetal heads in 3DUS	Random Forest Classifier, Template deformation	++
Sofka et al., 2014 [41]	CZ	16–35	2089 volumes	3D	ML	SFHP	automatic detection and measurement of structures in CNS volumes	Integrated Detection Network (IDN)/FNN	+
Namburete et al., 2015 [42]	UK	18–34	187	3D	ML	sulcation/gyration	GA prediction	Regression Forest Classifier	++
Yaqub et al., 2015 [43]	UK	19–24	40	3D	ML	SFHP	extraction and categorization of unlabeled fetal US images	Random Forest Classifier	+
Baumgartner et al., 2016 [44]	UK	18–22	201	2D	DL	SFHP (TT, TC)	retrieval of standard planes, creation of saliency maps to extract bounding boxes of CNS anatomy	CNN	+++
Sridar et al., 2016 [45]	IND	18–20	85	2D	DL	structure detection	image classification and structure localization in US images	CNN	+
Yaqub et al., 2017 [46]	UK	19–24	40	3D	DL	SFHP, CNS anomalies	localization of CNS, structure detection, pattern learning	Random Forest Classifier	+
Qu et al., 2017 [47]	CHN	16–34	155	2D	DL	SFHP	automated recognition of six standard CNS planes	CNN, Domain Transfer Learning	++
Namburete et al., 2018 [25]	UK	18–34	739 images	2D/3D	DL	structure detection	3D brain localization, structural segmentation and alignment	multi-task CNN	++
Huang et al., 2018 [48]	CHN	20–29	285	3D	DL	multi-structure detection	detection of CNS structures in 3DUS and measurements of CER/CM	VP-Net	++
Huang et al., 2018 [49]	UK	20–30	339 images	2D	DL	structure detection (CC/CP)	standardize intracranial anatomy and measurements	Region descriptor, Boosting classifier	++
van den Heuvel et al., 2018 [50]	NL	10–40	1334 images	2D	ML	biometry (HC)	automated measurement of fetal head circumference	Random Forest Classifier Hough transform	+
Dou et al., 2019 [51]	CHN	19–31	430 volumes	3D	ML	SFHP/structure detection	automated localization of fetal brain standard planes in 3DUS	Reinforcement learning	++
Sahli et al., 2019 [52]	TUN	n/a	86	2D	ML	SFHP	automated extraction of biometric measurements and classification of normal/abnormal	SVM Classifier	++
Alansary et al., 2019 [53]	UK	n/a	72	3D	ML/DL	SFHP/structure detection	localization of target landmarks in medical scans	Reinforcement learning deep Q-Net	+
Lin et al., 2019 [54]	CHN	14–28	1771 images	2D	DL	SFHP/structure detection	automated localization of six landmarks and quality assessments	MF R-CNN	+
Bastiaansen et al., 2020 [55]	NL	1st trimester	30	2D/3D	DL	SFHP (TT)	fully automated spatial alignment and segmentation of embryonic brains in 3D US	CNN	+
Xu et al., 2020 [56]	CHN	2nd/3rd trimester	3000 images	2D	DL	SFHP	simulation of realistic 3rd- from 2nd-trimester images	Cycle-GAN	++
Ramos et al., 2020 [57]	MEX	n/a	78 images	2D	DL	SFHP biometry (TC) GA prediction	detection and localization of cerebellum in US images, biometry for GA prediction	YOLO	+
Maraci et al., 2020 [58]	UK	2nd trim	8736 images	2D	DL	biometry (TC) GA prediction	estimation of GA through automatic detection and measurement of the TCD	CNN	+
Chen et al., 2020 [59]	CHN	n/a	2900 images	2D	DL	SFHP biometry (TV)	demonstrate the superior performance of DL pipeline over manual measurements	Mask R-CNN ResNet50	+
Xie et al., 2020 [60]	CHN	18–32	92,748	2D	DL	SFHP (TV, TC) CNS anomalies	image classification as normal or abnormal, segmentation of craniocerebral regions	U-Net VGG-Net	++
Xie et al., 2020 [61]	CHN	22–26	12,780	2D	DL	SFHP, CNS anomalies	binary image classification as normal or abnormal in standard axial planes	CNN	++
Zeng et al., 2021 [62]	CHN	n/a	1354 images	2D	DL	biometry	image segmentation for automatic HC biometry	DAG V-Net	+
Burgos Artizzu et al., 2021 [63]	ESP	16–42	12,400 images (6041 CNS)	2D	DL/ML	SFHP	evaluation of the maturity of current DL classifications tested in a real clinical environment	19 different CNNs MC Boosting algorithm HOG classifier	++
Gofer et al., 2021 [64]	IL	12–14	80 images	2D	ML	SFHP/structure detection (CP)	classification of 1st trimester CNS US images and earlier diagnosis of fetal brain abnormalities	Statistical Region Merging Trainable Weka Segmentation	+
Skelton et al., 2021 [65]	UK	20–32	48	2D/3D	DL	SFHP	assessment of image quality of CNS planes automatically extracted from 3D volumes	Iterative Transformation Network (ITN)	++
Fiorentino et al., 2021 [66]	I	10–40	1334 images	2D	DL	biometry (HC)	head localization and centering	multi-task CNN	++
Yeung et al., 2021 [67]	UK	18–22	65 volumes	2D/3D	DL	SFHP/structure detection	mapping 2D US images into 3D space with minimal annotation	CNN
Montero et al., 2021 [68]	ESP	18–40	8747 images	2D	DL	SFHP	generation of synthetic US images via GANs and improvement of SFHP classification	Style-GAN	++
Moccia et al., 2021 [69]	I	10–40	1334 images	2D	DL	biometry (HC)	fully automated method for HC delineation	Mask-R2CNN	+
Wyburd et al., 2021 [70]	UK	19–30	811 images	3D	DL	structure detection/ GA prediction	automated method to predict GA by cortical development	VGG-Net ResNet-18 ResNet-10	++
Shu et al., 2022 [71]	CHN	18–26	959 images	2D	DL	SFHP (TC)	automated segmentation of the cerebellum, comparison with other algorithms	ECAU-Net	+
Hesse et al., 2022 [72]	UK	18–26	278 images	3D	DL	structure detection	automated segmentation of four CNS landmarks	CNN	+++
Di Vece et al., 2022 [73]	UK	20–25	6 volumes	2D	DL	SFHP/structure detection	estimation of the 6D pose of arbitrarily oriented US planes	ResNet-18	++
Lin et al., 2022 [74]	CHN	18–40	16,297/166	2D	DL	structure detection	detection of different patterns of CNS anomalies in standard planes	PAICS YOLOv3	+++
Sreelakshmy et al., 2022 [75] ‡	IND	18–20	740 images	2D	DL	biometry (TC)	cerebellum segmentation from fetal brain images	ResU-Net	-
Yu et al., 2022 [56]	CHN	n/a	3200 images	2D/3D	DL	SFHP	automated generation of coronal and sagittal SPs from axial planes derived from 3DVol	RL-Net	++
Alzubaidi et al., 2022 [76]	QTAR	18–40	551	2D	DL	biometry (HC)	GA and EFW prediction based on fetal head images	CNN, Ensemble Transfer Learning	++
Coronado-Gutiérrez et al., 2023 [77]	ESP	18–24	12,400 images	2D	DL	SFHP, multi-structure delineation	automated measurement of brain structures	DeepLab CNNs	++
Ghabri et al., 2023 [20]	TN	n/a	896	2D	DL	SFHP	classify fetal planes/accurate fetal organ classification	CNN: DenseNet169	++
Lin et al., 2023 [78]	CHN	n/a	558 (709 (images/videos)	2D	DL	SFHP	improved detection efficacy of fetal intracranial malformations	PAICS YOLO	+++
Rauf et al., 2023 [79]	PK	n.s.	n.s.	2D	DL	SFHP	Bayesian optimization for the classification of brain and common maternal fetal ultrasound planes	Bottleneck residual CNN	+
Alzubaidi et al., 2023 [80]	QTAR	18–40	3832 images	2D	DL	SFHP	evaluation of a large-scale annotation dataset for head biometry in US images	multi-task CNN	+
Alzubaidi et al., 2024 [81]	QTAR	18–40	3832 images (20,692 images)	2D	DL	biometry	advanced segmentation techniques for head biometrics in US imagery	FetSAM Prompt-based Learning	+
Di Vece et al., 2024 [82]	UK	20–25	6 volumes	2D/3D	DL	SFHP (TV)	detection and segmentation of the brain; plane pose regression; measurement of proximity to target SP	ResNet-18	++
Yeung et al., 2024 [83]	UK	19–21	128,256 images	2D	DL	SFHP	reconstruction of brain volumes from freehand 2D US sequences	PlaneInVol ImplicitVol	++
Dubey et al., 2024 [84]	IND	10–40	1334 images	2D	DL	biometry (HC)	automated head segmentation and HC measurement	DR-ASPnet, Robust Ellipse Fitting	++

Clinically validated (and commercially available) software in gray shaded rows. Abbreviations: 2D, two dimensional; 3D, three dimensional; BPD, biparietal diameter; CER, cerebellum; CNN, convolutional neural network; CNS, central nervous system; CP, choroid plexus; CSP, cavum septum pellucidum; DL, deep learning; FOD, fronto-occipital diameter; GA, gestational age; GAN, generative adversarial network; HC, head circumference; LV, lateral ventricles; n/a, not applicable; n.s., not specified; PAICS, prenatal ultrasound diagnosis artificial intelligence conduct system; ResNet, residual neural network; SFHP, standard fetal head plane; SVM, support vector machine; TC, transcerebellar plane; TV, transventricular plane; TT, transthalamic plane; US, ultrasound; Vp, width of the posterior horn of the lateral ventricle; YOLO, You Only Look Once algorithm; * if not otherwise specified: number of patients; ** fully automated AI-driven software update has been released; *** potential clinical impact; ‡ withdrawn article.