Data Complexity Based Evaluation of the Model Dependence of Brain MRI Images for Classification of Brain Tumor and Alzheimer’s Disease

Abstract

The convolutional neural networks (CNN) have shown promising results for various classification problems over the past years. However, selecting various CNN architectures is still challenging as each architecture performs differently with the same dataset. This research aims to evaluate the dependence of brain MRI on various predictive models of CNN based on the complexity of the data for Brain Tumor and Alzheimer's Disease. Our proposed approach has three parts. First part is the pre-processing of the data which mainly focuses on class balancing and the estimation of data complexity. The second part uses stratified k-fold cross-validation for more reliable results. The last part corresponds to the implementation of four CNN models applying described methods. This paper compares the classification performance of rigorous experimentation on four CNN variants namely S-CNN (CNN trained from scratch), ResNet50, InceptionV3, and Xception over two brain MRI image datasets evaluated with and without the use of Principal Component Analysis (PCA). The work benchmarks CNN models by comparing the average scores of Accuracy, Precision, Recall, F1 score, and AUC score from the stratified five-fold cross-validation.

Deep Learning CNN Model-Based Anomaly Detection in 3D Brain MRI Images using Feature Distribution Similarity

“…The model extracts the features using a pre-trained CNN model and performs classification using algorithms like SVM, KNN, and softmax. A detailed analysis of various CNN models is sketched in [6], which considers the models like S-CNN (CNN trained from scratch). The method uses two brain image data sets to analyze the performance of various approaches.…”

BrainNet: A Unified Neural Network Architecture for Brain Image Classification

“…In this subsection, we delve into the literature closely related to the conducted study and the proposed network for both BT and AD classification using MRI (Majd et al (2019); Roe et al (2010); Sánchez-Valle et al ( 2017)). Kujur et al (2022) propose a stratified k-fold cross-validation method utilizing CNNs trained from scratch, ResNet-50, Inceptionv3, and Xception, to detect Alzheimer's and brain tumors simultaneously. Acquarelli et al (2022) assess the value of CNNs in diagnosing brain tumors and Alzheimer's disease, addressing challenges related to limited case numbers in datasets and resulting in interpretability in terms of relevant regions.…”

“…Notably, the test performance in our models closely aligns with the training performance, indicating the approach's efficacy. On the contrary, Kujur et al (2022) provides insights into real-world test performance, making it difficult to assess its practical utility. Furthermore, the work by Namachivayam and Puviarasan (2023) focuses on a slightly different dataset that does not consider classwise distribution within each disease category (AD, BT , Parkinson).…”

“…While encountering such diseases in daily life is relatively uncommon, addressing these challenges remains imperative. Furthermore, certain diseases within this spectrum are interconnected, as seen in the case of Alzheimer's Disease (AD) and Brain Tumor (BT ), revealing a deficiency in the current state-of-the-art (SOT A) architecture's capability to effectively identify multiple interconnected abnormalities (Kujur et al (2022)). Biologically, AD and BT exhibit distinct cellular behaviors.…”

“…Recent literature has introduced several machine 1 thanks to advancements in medical science learning models to aid clinicians and, in turn, enhance the accuracy of medical systems (Gupta et al (2020); Ranga et al (2020Ranga et al ( , 2022; Gupta et al (2021)). However, the current SOT A architecture is presently constrained in its ability to effectively identify multiple interconnected abnormalities (Kujur et al (2022)). Much ongoing research is directed towards classifying different diseases individually (Yildirim and Cinar (2020); Loddo et al (2022); Salçin et al (2019)).…”