ColoRectalCADx: Expeditious Recognition of Colorectal Cancer with Integrated Convolutional Neural Networks and Visual Explanations Using Mixed Dataset Evidence

Table 13

Comparison of best results of individual and mixed datasets.

Dataset	Model type	Model	Training accuracy in %	Testing accuracy in %
CVC Clinic DB	Individual CNN	CNN+DWT+SVM (DenseNet-201)	95.37	99
	Integrated CNN	CNN+DWT+SVM (DaRD-22)	93.86	96
Kvasir2	Individual CNN	CNN+DWT+SVM (DenseNet-201)	80.53	88
	Integrated CNN	CNN+DWT+SVM (DaRD-22)	77.01	82
Hyper Kvasir	Individual CNN	CNN+DWT+SVM (DenseNet-201)	78.17	84
	Integrated CNN	CNN+DWT+SVM (RDEV2-22)	62.38	66
Mixed dataset (CVC Clinic DB+Kvasir2+Hyper Kvasir)	Individual CNN	CNN+LSTM (DenseNet-201)	87.1	84.7
	Integrated CNN	CNN+LSTM (ADaDR-22)	84.61	82.17