Hybrid Machine Learning Approach of “KNN LSTM” for detection of Brain Tumour in MR images

Now a days, more existing techniques are available to found tumours on brain from MR images. But still the researchers are concentrated on to detect the brain tumour accurately with less time. So, this work proposed the hybrid machine learning algorithm of “K-Nearest Neighbour - Long Short-Term Memory” (“KNN-LSTM”) with increased ‘accuracy’, ‘sensitivity’, ‘specificity’, with less ‘time consumption’. In this work, the brain tumour is identified by predicting the MR images from huge dataset with usage of machine learning approach of “KNN-LSTM”. Brain tumour detection is difficult when to compare the abnormal and typical tissues of large capacity. In order to enable a speedy recovery following therapy, brain tumours should be identified as soon as feasible. If the brain tumour is discovered early on, patients can benefit from improved care. Due to the irregular spreading of the abrasions, brain tumours are still difficult to identify even with intensive study. Sometimes the tiny tumours look like normal, but it is challenging to find it. So, the techniques are need to increase the classification accuracy, and selecting the useful features. This paper proposes a hybrid machine learning model “KNN-LSTM” for categorizing the tumours with predicting through MR images. MRI brain imaging collection is used for our experimentation. The “KNN” algorithm usage is to extract the important characteristics from pictures once the data has been pre-processed effectively. The proposed work predicts the brain tumour with 99.15% of ‘accuracy’, 98% of ‘sensitivity’, 97.12% of ‘specificity’ and testing time consumption of 0.5 minutes.