BMC Medical Imaging

Table 2 Performance of individual DLMs for body part classification: comparison among the algorithm models based on preprocessing methods and planes

From: Development and validation of an ensemble artificial intelligence model for comprehensive imaging quality check to classify body parts and contrast enhancement

Pre-processing Method	Internal validation set				External validation set
Pre-processing Method	Precision (%)	Recall (%)	Accuracy (%)	F1-score (%)	Precision (%)	Recall (%)	Accuracy (%)	F1-score (%)
AIP
Axial	100	100	100	100	94.99	94.15	94.86	94.6
Sagittal	100	100	100	100	97.52	96.85	97.09	97.2
Coronal	100	100	100	100	95.24	91.73	93.75	93.45
MIP
Axial	100	100	100	100	98.85	98.33	98.66^a	98.6
Sagittal	100	100	100	100	99.33	99.33	99.33^ab	99.3
Coronal	100	100	100	100	98.85	98.33	98.66^a	98.6
Mid-plane
Axial	100	100	100	100	96.24	96.51	96.2	96.4
Sagittal	100	100	100	100	97.7	96.32	97.32	97
Coronal	100	100	100	100	96.24	96.51	96.2	96.3

^aThe DLMs with the highest performance in each plane are selected to apply for ensemble AI model
^bThe DLM with the highest performance from all pre-processing methods is selected as the best performing individual DLM

Back to article page

ISSN: 1471-2342

Contact us

General enquiries: journalsubmissions@springernature.com