Published Paper Details:

Tumor detection in the brain is considered a crucial effort in imaging techniques where the diagnosis made precisely and early can change the scenario in healthcare. Driving into their applications, deep learning effectiveness comes from systems where convolutional networks, under the generals of recurrent models, are well-articulated, such as long short-term memory. Thereby in the study, we came up with a hybrid model for deep learning that integrates CNN-LSTM as a base model and VGG 16 feature extractor to improve on the performance of brain tumor classification; the CNN-LSTM model is spatial and temporal dependencies extraction from MRI images while deep feature representation is facilitated by VGG16. The ensemble model improves classification accuracy by jointly preparing spatial hierarchies and sequential dependencies for data. Testing and training was accomplished from a brain tumor dataset on a very high accuracy and robust performance. Experimental results therefore show that the hybrid VGG-LSTM model is superior in brain tumor detection over conventional CNN. Thus, clinicians dealing with diagnosis could be assisted with a very easy tool by radiologists.