Fractional wavelet transform based diagnostic system for brain tumor detection in MR imaging

Abstract

The brain tumor detection is a highly complicated but significant task. The early detection of a brain tumor can increase the survival rate of an individual by providing proper treatment. This work proposes a computer‐aided diagnostic method for brain tumor detection using fractional wavelet transform (FrDWT) with different values of alpha (α) ranging from (0.1‐1), histogram‐based various local feature descriptors, feature selectors, and two classification methods, that is, support vector machine (SVM) as well as artificial neural network (ANN). The brain MR images dataset is taken from BraTS 2015, e‐health laboratory, and Harvard Medical School. The FrDWT and the local feature descriptors used are combined to extract features. Some of the features are selected using Eigenvector centrality and Laplacian Score techniques. The selected features are trained and classified by the two classifiers, SVM and ANN. It is a type of binary classification, so the labels provided to the classifier are named “normal” and “abnormal.” The performance is estimated using parameters like accuracy, sensitivity, precision, specificity, and F1 score. The results of FrDWT are compared with conventional discrete wavelet transform (DWT), and it is observed that FrDWT outperforms DWT at alpha (α) values lower than 0.5.

Brain Tumor Detection Based on Amended Convolution Neural Network Using MRI Images

“…Though the ensemble model combines multiple models, it may not be efficient in handling highly complex data with multiple features. The existing PCA and TK-means [28] approach and the fractional wavelet [31] achieve an accuracy of 91.5% and 92.3%which is lower than all other existing and proposed algorithms. These two methods work better when the number of datasets is limited, but their performance is affected by this limitation and it reduces their classification accuracy.…”

“…The suggested ACNN Model is described together with two other related approaches that are currently in use: 23 layers CNN [29], Ensemble model [32], Fractional Wavelet [31], and PCA and TK-means [28] by using the various performance metrics like accuracy, sensitivity, dice score, Jaccard index, recall, Positive predictive value, specificity, Hausdorff distance, precision, and F1 score. True Positive (TP), True Negative (TN), False Positive (FP), as well as False Negative (FN) classes can be utilized to compute these measurements.…”

“…This indicates that the proposed algorithm has improved the accuracy of brain tumor detection when compared to these models. The Fractional wavelet [31] and PCA and TK-means [28] models had better Dice scores and Jaccard index than the 23 layers CNN [29] and Ensemble model [32], but they had higher Hausdorff distances. This suggests that these models were better at predicting the location of the tumor, but not the boundary.…”

“…The study demonstrated encouraging outcomes; however, it should be noted that the method lacked external validation and clinical evaluation [30]. Similarly, a diagnostic system was proposed using the Fractional Wavelet Transform (FWT) for brain tumor detection with MRI, but the study's limitations included a small dataset and a lack of external validation [31]. A suggestion for a technique that uses ensemble deep learning frameworks and Class Activation Maps (CAMs) markers to diagnose brain cancers using MRI images was provided in [32].…”

“…𝑃𝑃𝑆𝑆𝑏𝑏𝑏𝑏𝑡𝑡𝑏𝑏𝑃𝑃𝑤𝑤 𝑠𝑠𝑠𝑠𝑤𝑤𝑆𝑆𝑏𝑏𝐴𝐴𝑡𝑡𝑏𝑏𝑃𝑃𝑤𝑤 𝑃𝑃𝑏𝑏𝑃𝑃𝑢𝑢𝑤𝑤 = 𝑇𝑇𝑃𝑃 𝑇𝑇𝑃𝑃+𝐹𝐹𝑃𝑃 (13) Ensemble model [32] Fractional Wavelet [31] PCA and TKmeans [ Table 4 demonstrates that the modified convolution method proposed for brain tumor detection attained an accuracy of 98.8%, surpassing the accuracy, F1 score, precision, recall, and positive predictive value of all existing algorithms. In comparison, the existing 23-layer CNN network achieves an accuracy of 96.7% which is significantly lower than the proposed algorithm.…”