Research Article

CAMELYON 17 Challenge: A Comparison of Traditional Machine Learning (SVM) with the Deep Learning Method

Table 4

A summary of breast cancer image classification with the SVM method.

ReferenceDescriptorImage typeImage numberKey findings

Levman et al. [25](1) Intensity of relative signals
(2) Relative signal intensities and their derivatives in a single vector
(3) Maximum signal intensity enhancement, including enhancement time and washout time
MRI76(1) Lesions that are both benign and malignant are studied
(2) For breast image classification, a linear kernel, a polynomial kernel, and a radial basis function kernel were used in conjunction with the SVM approach
Oliveira Martins et al. [26]Ripley’s functionMammogram390(1) Classification of benign and cancerous images
(2) Accuracy, sensitivity, and specificity were attained at 94.94 percent, 92.86 percent, and 93.33 percent, respectively
Chang et al. [27]Chosen textural features:
Autocorrelation coefficient
Autocovariance coefficient
Ultrasound250(1) Images have been classed as benign or cancerous
(2) Accuracy, sensitivity, specificity, positive predictive values, and negative predictive values are 85.60 percent, 95.45 percent, 77.86 percent, 77.21 percent, and 95.61 percent, respectively
Sewak et al. [28]Area, radius, perimeter, compactness, smoothness, concavity, concave points, symmetryBiopsies569Accuracy, sensitivity, and specificity were achieved at 99.29%, 100.00%, and 98.11%, respectively
Zhang et al. [29]Chosen feature:
Information from the fractional Fourier transform
Mammogram200(1) ROI to eliminate unnecessary complexity
(2) When SVM and PCA are combined, the accuracy, sensitivity, and specificity reached are percent, %, and %, respectively
Shirazi and Rashedi [30]Grey-level cooccurrence matrixUltrasound322(1) ROI was removed to reduce the amount of superfluous complexity
(2) Feature reduction: combine SVM and the mixed gravitational search algorithm (MGSA)
(3) The acquired accuracy: was 86.00 percent; the SVM+MGSA approach achieved an accuracy: 93.10 percent