Table 5.

Study analysis for journal papers in the screening phase.

Reference	Objective, data set, and methodology	Performance and remarks
[3]	Objective: Classification of white blood cell leukemia Data set: Acute Lymphoblastic Leukemia Image Database for Image Processing 1 and 2 Methodology: A hybrid model (CNNa and SESSAb)	Performance: Accuracy: 99.2% Sensitivity: 100% Strengths: Powerful performance using CNN Use of the salp swarm optimization method Hybrid classification method Use of transfer learning Limitations: Small limited data set insufficient to train CNNs Validation: 5-fold internal cross validation and 20% testing (external validation)
[61]	Objective: Automated identification of acute lymphoblastic leukemia Data set: Blood smear images obtained from the Department of Hematology at the University Hospital Ostrava Methodology: support vector machine/artificial neural networks	Performance: Accuracy: 98.19% Strengths: High classification accuracy Successful feature selection Limitations: Extensive preprocessing is required Lack of medical data sets Inability to generalize the results and trends for lack of comparison with other methods Validation: 10-fold cross validation repeated 10 times
[33]	Objective: Classification of chronic myeloid leukemia phases Data set: 500 pictures from Patliputra Medical College and Hospital, Dhanbad, and the blood journal repository Methodology: CNN	Performance: Accuracy: 97.8% Strengths: Use of transfer learning Limitations: Limited data set Validation: Internal validation (14 left for testing)

^aCNN: convolutional neural network.

^bSESSA: statistically enhanced salp swarm algorithm.