Brain Tumor Detection using CNN and PCA

Accurate identification of Brain Tumor is important in the medical field. Initially images are preprocessed to implement machine learning easy. Neural Networks are popular to extract better features from brain tumors. Three Convolution layers are considered in this work. Filters used are 8, 16, and 32 respectively in three Convolution layers. But the number of features is high. It can lead to overfitting also. Hence, Principal Component Analysis is used to reduce feature vectors. Discriminative patterns are identified by using PCA. It leads to higher accuracy. Further, the model has been trained with a streamlined architecture featuring fewer layers. Performance metrics used in this work are accuracy, precision, recall and F1-score. Experimental results indicate that the proposed method gives better accuracy.