Diagnostic test accuracy of machine learning algorithms for the detection intracranial hemorrhage: a systematic review and meta-analysis study

Maghami, Masoud; Sattari, Shahab Aldin; Tahmasbi, Marziyeh; Panahi, Pegah; Mozafari, Javad; Shirbandi, Kiarash

doi:10.1186/s12938-023-01172-1

Table 1 Summary of findings for all studies included in the qualitative synthesis

From: Diagnostic test accuracy of machine learning algorithms for the detection intracranial hemorrhage: a systematic review and meta-analysis study

ID	Study design	ICH type		ML model type	CT-Scan equipment	Data sources	Segmentation	Sensitivity %	Specificity %	Accuracy %	AUC
Schmitt, N., et al. 2022/Germany [39]	Retrospective	ICH		CNN	64-slice multidetector, single-source scanner (Somatom Defnition AS, Siemens Healthineers)	Single/Real-time data	2D	91	89	NA	0.9
Phaphuangwittayakul, A., et al. 2022/China [36]	Retrospective	ICH		CNN	NA	Single/Benchmark	2D	95.77	96.90	96.21	NA
		EDH						95.48	96.02	95.68
		SDH						96.01	97.55	96.54
		IPH						95.83	97.13	96.41
Hopkins, B. S et al. 2022/ USA [29]	Prospective	ICH		DNN	NA	Single/Real-time data	2D	98	99	NA	0.99
Seyam, M., et al. 2022/ Switzerland [40]	Prospective	ICH		DL	256-section scanners (Somatom Force and Somatom Definition Flash, Siemens)	Single/Real-time data	2D	87.2	93.9	93	NA
Altuve, M., & Pérez, A. 2022/Venezuela [22]	Retrospective	ICH		ResNet-18	NA	Single/Real-time data	2D	95.65	96.2	95.93	NA
Tang, Z., et al. 2022/China[41]	Retrospective	ICH		CNN	NA	Single/Real-time data	2D	91.97	88.37	90.58	NA
Cortes-Ferre L, et al. 2022/ Spain [26]	Retrospective	ICH		DL	NA	Single/Benchmark	2D	91.4	94	92.7	0.978
Kau, T., et al. 2022/ Austria [30]	Retrospective	ICH		DL	NA	Single/Real-time data	2D	68.2	96.8	94	NA
Tharek A., et al. 2022/Malaysia [42]	Retrospective	ICH		CNN	NA	Single/Benchmark	2D	96.94	93.14	95	NA
Abe, D., et al. 2022/Japan [20]	Retrospective	ICH		XGBoost	NA	Single/Real-time data	2D	74	74.9	NA	0.8
Trevisi, G.et al. 2022/Italy [43]	Retrospective	ICH		RF	NA	Multiple/Real-time data	2D	77.52	86.29	83.55	0.93
Uchida, K., et al. 2022/Japan [44]	Prospective	ICH	LR	LR	NA	Multiple/Real-time data	2D	43	92	NA	0.82
			RF	LR				41	94		0.82
			XGBoost	RF				40	92		0.81
		SAH	LR	RF				27	97		0.87
			RF	XGBoost				16	98		0.85
			XGBoost	XGBoost				23	97		0.86
Alis, D.. et al. 2022/Turkey [21]	Retrospective	ICH-Binary		CNN-RNN	NA	Multiple/Real-time data	2D	96.41	95.79	96.02	0.961
		IPH						82.56	97.54	94.69	0.905
		IVH						86.84	98.31	97.35	0.925
		SAH						91.67	86.14	86.73	0.889
		SDH						88.16	90.16	89.82	0.891
		EDH						71.4	99.98	98.89	0.98
Rao, B. N. et al. 2022/ India [37]	Retrospective	ICH		VGG-16	64-slice CT scan machine (GE OPTIMA, 64 slice)	Single/Real-time data	2D	91.2	93.1	93.1	0.965
				GoogleNet (InceptionV3)				97.4	98.6	98.9	0.988
				ResNet-50				97.1	99.3	98.2	0.984
				Proposed model				97.1	99.3	98.2	0.984
				Proposed model				99.4	99.7	99.6	1
Zhou, Q., et al. 2022/ China [50]	Retrospective	EDH	ResNet-18	ResNet-18/DenseNet-121	SIEMENS/GE/TOSHIBA/Neusoft	Single/Real-time data	2D	98	88	NA	NA
		EDH	DenseNet-121					86	81
		IVH	ResNet-18					85	91
		IVH	DenseNet-121					73	85
		CPH	ResNet-18					80	91
		CPH	DenseNet-121					76	84
		SAH	ResNet-18					81	91
		SAH	DenseNet-121					81	83
		SDH	ResNet-18					93	89
		SDH	DenseNet-121					85	82
Salehinejad., H. et al. 2021/Canada [38]	Retrospective	EDH		SE-ResNeXt50-32 and SE-ResNeXt101-32 (DL)	64 row multi-detector CT scanner(Revolution, LightSpeed 64, or Optima 64, General Electric Medical Systems)	Single/Benchmark	2D	21.5	99.9	99.4	60.8
		SDH						84.3	98.5	96.5	91.4
								76.9	98.7	95.5	87.8
								93.2	98.9	97.9	96.0
		SAH						94.1	98.3	97.4	96.2
		IVH
		IPH
McLouth J., et al. 2021/USA [35]	Retrospective	Intraparenchymal, Intraventricular, Epidural/Subdural, and Subarachnoid		DL	GE Medical Systems, Philips, Siemens, Canon (Formerly Toshiba), and NMS	Multiple/Real-time data	2D	91.4	97.5	95.6	NA
Voter, Andrew., F et al. 2021/USA [45]	Retrospective	ICH		DSS (DL)	Helical GE, pitch of 0.531, 120 kV, thin axial reconstruction is 1.25-mm slices at 0.625-mm intervals	Multiple/Real-time data	2D	92.3	97.7	NA	NA
Xu J., et al. 2021/China [47]	Retrospective	ICH, EDH, and SDH		Dense U-Net (DL)	NA	Multiple/Real-time data	2D	NA	NA	NA	NA
Danilov, G., et al. 2021/Russian Federation [27]	Retrospective	EDH		ResNexT (DL)	NA	Single/Real-time data	2D	62.6	NA	82.8	0.762
		SDH						51.8	NA	81.8	0.711
		SAH						49.2	NA	82.9	0.748
		SAH						72.3	NA	95.2	0.804
		IVH						72.3	NA	95.2	0.804
		IVH						76.6	NA	0.868	0.803
		IPH						76.6	NA	0.868	0.803
XU X et al. 2021/China [48]	Retrospective	HICH		SVM	NA	Single/Real-time data	2D	90.9	84.1	85	NA
				KNN				90	82.2	83.6	NA
				LR				90	82.2	83.6	NA
				DT				90.9	84.1	85.5	NA
				RF				80	87.5	85.5	NA
				RF				93.3	92.5	92.7	NA
				XGBoost				93.3	92.5	92.7	NA
				XGBoost				92.3	88.1	89.1	NA
Wang W et al. 2021/China [46]	Retrospective	ICH		2D-CNN	Siemens/SOMATOM Definition AS CT scanner	Multiple/Benchmark	2D	95	94.4	NA	0.988
		EDH						97.4	94	NA	0.984
		IPH						96.5	95.9	NA	0.992
		IVH						97.5	97.4	NA	0.996
		SAH						94	94.2	NA	0.985
		SDH						94.6	93.2	NA	0.983
Kumaravel, P et al. 2021/India [31]	Retrospective	ICH		AlexNet	NA	Multiple/Benchmark	2D	99.35	99.07	99.21	99.96
				AlexNet-SVM				99.35	99.07	99.21	99.96
				AlexNet-SVM				99.67	99.53	99.6	99.99
				AlexNet-PCA-SVM				99.58	99.35	99.47	99.98
Ye, H., et al. 2019/ China [49]	Retrospective	ICH		CNN-RNN	NA	Multiple/Real-time data	2D	99	99	99	1
		CPH						92	83	90	0.94
		SAH						69	94	83	0.89
		IVH						84	95	91	0.93
		SDH						86	96	94	0.96
		EDH						69	98	96	0.94
Kuo, W., et al. 2019/USA [32]	Retrospective	ICH		CNN	GE, Siemens	Single/Benchmark	2D	100	90	NA	NA
Lee, H., et al. 2019/USA [33]	Retrospective/Prospective	ICH, IPH, IVH, SDH, EDH or SAH		DCNNs—VGG16, ResNet-50, Inception-v3 and Inception-ResNet-v2 (DL)	NA	Single/Real-time data	2D	ICHr: 98	ICHr: 95	NA	ICHr: 0.993
								IPHr: 92.5	IPHr: 91.8		IPHr: 0.98
								IVHr: 87	IVHr: 95.9		IVHr: 0.979
								SDHr: 87.5	SDHr: 86.9		SDHr: 0.959
								EDHr: 58.3	EDHr: 95.2		EDHr: 0.922
								SAHr: 84.1	SAHr: 88.5		SAHr: 0.96
								ICHp: 92.4	ICHp: 94.9	NA	ICHp: 0.961
								IPHp: 68.8	IPHp: 95		IPHp: 0.921
								IVHp: 83.3	IVHp: 99.5		IVHp: 0.973
								SDHp: 70.5	SDHp: 92.8		SDHp: 0.881
								EDHp: NA	EDHp: NA		EDHp: NA
								SAHp: 76.3	SAHp: 89.9		SAHp: 0.926
Arbabshirani, M. R., et al. 2018/USA [23]	Retrospective	ICH		R-CNN (DL)	17 scanners from 4 different manufacturers	Multiple/Real-time data	2D	70	87	84	0.846
Majumdar, A., et al. 2018/USA [34]	Retrospective	Epidural, Subdural, Subarachnoid, Intraparenchymal		CNN (U-Net)	NA	Single/Real-time data	2D	81	98	NA	NA
Chang, P. D., et al. 2018/USA [24]	Retrospective	ICH		Mask R-CNN + Hybrid 3D/2D CNN	NA	Single/Real-time data	2D	97.1	97.5	NA	0.984
Chang, P. D., et al. 2018/USA [24]	Prospective	ICH		Mask R-CNN + Hybrid 3D/2D CNN	NA	Single/Real-time data	2D	97.5	97.3	NA	0.972
Grewal., et al. 2018/USA[28]	Retrospective	ICH		CNN	NA	Multiple/Benchmark	2D	88.6	72.7	81.82	0.818
Chilamkurthy, S. et al. 2018/India [25]	Retrospective	ICH		CNN transfer learning (ResNet18)		Multiple/Real-time data	2D	98.07	98.73	NA
		IPH						98.09	98.83
		IVH						100	100
		SAH						93.18	99.65
		EDH						100	99.83
		SAH						100	99.83
		SAH						100	99.71

Back to article page

ISSN: 1475-925X

Contact us

Submission enquiries: journalsubmissions@springernature.com

BioMedical Engineering OnLine

Contact us