Table 1.

A brief comparison between previous prostate MRI CAD studies.

Reference	Year	Type of Approach	Features Type	Classes	Images Sequences	No. of Patients Involved	Accuracy Result
[17]	2017	Handcrafted features-based CAD	Spatial, intensity, and texture	Benign, Gleason 6, Gleason 7, Gleason 8, Gleason 9, Gleason 10	B2000, ADC, and T2W	224	SVM model achieved an AUC value of 0.86, while Random Forest achieved an AUC of 0.93
[19]	2016		Texture	Malignant or benign	T2W	45	It has a value of 0.93 AUC
[21]	2017		Texture, intensity, edge, and anatomical	Voxel-based classification	DWI, T2W, DCE, and MRSI	17	Classification performance of an average AUC of 0.836 ± 0.083 is achieved
[22]	2019		Texture	High risk patients and low risk patients	T2WI and ADC	121	Quadratic kernel based SVM is the best model with an accuracy of 0.92
[9]	2020		Texture and intensity	Benign and/or cs PCa vs. non-cs PCa	B50, b400, b800, b1400, T2WI, DCE, and ADC	206	It has an average AUC value of 0.838
[23]	2020		Shape, texture, and statistical texture	Normal vs. cancerous prostate lesion and clinically significant PCa vs. clinically insignificant PCa	ADC and T2WI	191	AUC value for normal vs. cancerous classification is 0.889, while the AUC value for clinically significant PCa vs. clinically insignificant PCa is 0.844
[13]	2019	Deep learning-based CAD		Produces a voxel probability map	T2WI	19	The model attained an AUC value of 0.995, a recall of 0.928, and an accuracy of 0.894.
[26]	2018			Produces probability maps to detect prostate cancer	T2WI, ADC, and high b-value (b1500 for cases imaged without ERC insertion, and b-2000 with ERC insertion)	186	The model attained an average AUC value of 0.94 in the peripheral zone and an average AUC value of 0.92 in transition zone.
[27]	2020			Gives a PI-RADS score to a lesion detected and segmented by a radiologist	T2WI, T1WI, ADC, and (b1500 or b2000)	687	Kappa = 0.40, sensitivity = 0.89, and specificity = 0.73.
[28]	2021			Probability that patient has prostate cancer	T2WI, b200, ADC in the first dataset, T2WI, ADC in the second dataset	249 patients in the 1st dataset and 282 patients in the 2nd dataset	AUC value for the first dataset was 0.79, and for the second dataset was 0.86.
[29]	2021			Predicting the Gleason grade group and classifying benign vs. csPCa	T1WI and T2WI	490 cases for training and 75 cases for testing from 2 different datasets	On the lesion level, AUC of 0.96 for both the first and second datasets. On the patient level, AUC of 0.87 and 0.91, for the first and second datasets, respectively.