Table 4.

Various AI/ML models proposed for PDAC detection/diagnosis.

Category	Ref	Dataset	AI/ML Model	Metrics
Detection	[50]	CT	SVM	Acc = 0.922
	[51]	CEUS	ResNet-50	AUC = 0.953
	[52]	CT	CNNs	AUC = 0.986
	[53]	CT	3D TransUNet	Sens = 0.91
	[54]	EUS	EfficientNetV2-L	Sens = 0.96
	[55]	CECT	3D U-Net	Sens = 0.99
	[56]	CT	ResNet9	AUC = 0.95
Diagnosis	[57]	CECT	SVM	Acc = 0.86
	[58]	CT	VGG16-XGBoost	Acc = 0.97
	[59]	CECT	LASSO Regression	AUC = 0.75
	[60]	CT	CNNs	Acc = 0.867
	[61]	H&E Slides	Bayesian DenseNet-201	Acc = 0.856

Abbreviations used: CECT—Contrast-Enhanced Computed Tomography, SVM—Support Vector Machine, LASSO—Least Absolute Shrinkage and Selection Operation, H&E—Hematoxylin and Eosin, CEUS—Contrast-Enhanced Ultrasound, EUS—Endoscopic Ultrasound.