Improved Machine Learning Method for Intracranial Tumor Detection with Accelerated Particle Swarm Optimization

Abstract

The field of image processing is distinguished by the variety of functions it offers and the wide range of applications it has in biomedical imaging. It becomes a difficult and time-consuming process for radiologists to do the manual identification and categorization of the tumour. It is a complex and time-consuming procedure conducted by radiologists or clinical professionals to remove the contaminated tumour region from magnetic resonance (MR) pictures. It is the goal of this study to improve the performance and reduce the complexity of the image segmentation process by investigating FCM predicted image segmentation procedures in order to reduce the intricacy of the process. Furthermore, relevant characteristics are collected from each segmented tissue and aligned as input to the classifiers for autonomous identification and relegation of encephalon cancers in order to increase the accuracy and quality rate of the neural network classifier. An evaluation, validation, and presentation of the experimental performance of the suggested approach have been completed. A unique APSO (accelerated particle swarm optimization) based artificial neural network model (ANNM) for the relegation of benign and malignant tumours is presented in this study effort, which allows for the automated identification and categorization of brain tumours. Using APSO training to improve the suggested ANNM model parameters would give a unique method to alleviate the stressful work of radiologists performing manual identification of encephalon cancers from MR images. The use of an APSO-based ANNM (artificial neural network model) model for automated brain tumour classification has been presented in order to demonstrate the resilience of the classification model. It has been suggested to utilise the improved enhanced fuzzy c means (IEnFCM) method for image segmentation, while the GLCM (gray level co-occurrence matrix) feature extraction approach has been employed for feature extraction from magnetic resonance imaging (MR pictures).

Performance analysis of state‐of‐the‐art CNN architectures for brain tumour detection

“…All variants of the YOLOv5 model (YOLOv5s, YOLOv5n, YOLOv5m, YOLOv5l and YOLOv5x) were tested with an accuracy of 87%, 85.2%, 89%, 90.2% and 91%, respectively, with an average accuracy of 82%-92% and a precision score of 88%. Sajad Einy et al 13 14 proposed a solution by using an improved version of the ML algorithm with accelerated particle swarm optimisation for intracranial tumour detection. The proposed methodology was used to detect brain tumour on 750 MRI images in earlier stages with the goal of reducing the time complexity of the image segmentation process, median filter was applied to remove noise and detect edges; segmentation was done by applying fuzzy c-mean clustering and features were extracted using the GLCM method and passed to ANN for classification.…”

Prospects and Challenges of Using Machine Learning for Academic Forecasting

“…Several researchers [7,10,[12][13][14][15][16][17][18][19] have highlighted the growing application and usefulness of machine learning as a tool for forecasting of events in the academic sector. Similarly, a study by Elhaj et al [13] proved the exactness of the KNN algorithm in detecting the student's learning style, which is fast becoming helpful in training at-risk students.…”