Table 5.

Analysis of BC diagnosis based on DL and NN approaches in MRI using information from reviewed articles.

Reference	Approach	Targets for ML	Targets for BC	Advantages of the ML approach	Advantages of the BC approach	Disadvantages	Evaluation criteria
							AUC	ACC	SPE	SEN	Recall	F-Score
[90]	CNN	•Proposing an improved diagnostic approach to MRI images classification	•Enhancing the performance of the classification process	•Assisting in the accurate diagnosis of BC	•Enhanced performance across the BC classifier	•Failure to utilize clinical characteristics and effective parameters when diagnosing •Lack of evaluation of the proposed approach on other datasets	✔	✘	✔	✔	✘	✘
[91]	CNN	•Presenting an approach for achieving competitive performance in image classification	•Obtaining reliable results in the classification of images	•Assisting in the process of classification by improving its quality	•Providing more assistance in the diagnosis of BC •The proposed approach is more effective than single detection-based image classification methods	•The incapability of the proposed approach to deal with data obtained from other diagnostic methods •Failure to use powerful techniques to predict results	✔	✔	✔	✔	✘	✘
[92]	CNN	•Presenting a novel approach to accurately detecting tumor-containing axial slices	•Finding an optimal classification using tumor-containing axial slices	•Reducing the learning time of the proposed algorithm to recognize new samples from the created patterns better	•Facilitating the accurate identification of tumors •Reducing the time it takes to diagnose tumors	•Lack using of a considerable lot of samples for investigation performance proposed method •Failure to use critical clinical features in the process diagnosis	✔	✔	✔	✔	✘	✘
[93]	CNN	•Propose a novel method to evaluate the efficacy of 3D deep convolutional neural network for diagnosing BC	•Achieving high accuracy and quality in the classification of images	•Present a high accuracy for diagnosing BC	•Aid to radiologists in BC diagnosis. •Assisting in the classification of images in an optimal manner	•Ignoring the influence of other factors •Incompatibility with other clinical diagnosis methods	✔	✔	✔	✔	✘	✘
[94]	CNN	•Proposing an algorithm to evaluate the complete pathological response (pCR) to chemotherapy	•Achieving rapid and accurate detection of BC	•Improving PCR status prediction with MRI images in neoadjuvant therapy	•Aiding patients in maximizing their chances of survival •Assisting in improving the quality of treatment	•Insufficient use of important clinical parameters •A lack of implementation of the proposed approach in large datasets	✔	✘	✔	✔	✘	✘