Computer-assisted detection of pseudo brain tumors using lstm deep neural networks on magnetic resonance spectroscopy signals

Magnetic resonance spectroscopy (MRS) is one of the non-invasive tools used in the detection of brain tumors at the present time. It has been widely preferred by physicians in recent years because it does not pose the risk of infection and death due to surgery like biopsy. MRS provides a metabolic profile about the brain. In this profile, MRS patterns of the tumors and pseudo tumors can be similar to each other. For this reason, accurate diagnosis and classification of brain tumor is of vital importance for the patient's treatment planning. In this study, using MRS data, a binary classification of actual brain tumors and normal brain tissues against pseudo brain tumors was provided with deep neural networks. In the experimental studies conducted within the scope of the study, Long Short Term Memory (LSTM) and Bi-directional Long Short Term Memory (Bi-LSTM) deep neural network architectures were used. For experimental studies in this study, MRS signals of 29 glioblastoma brain tumors, 26 normal brain tissues, 19 metastatic brain tumors, 9 diffuse astrocytoma brain tumors and 9 pseudo brain tumors in the INTERPRET (International Network for Pattern Recognition of Tumours using Magnetic Resonance) database were used. Since obtaining MRS data from a large number of tumors and pseudo tumors for training and testing of LSTM neural networks is a difficult procedural process in the real world, the number of data for the MRS dataset has been increased by data augmentation (replication) methods before the network was trained and tested. LSTM neural networks have been trained and tested with these data. During the training and testing of the LSTM neural networks, repeated k-fold cross validation method was used for each model. Neural network trainings were carried out with 5 folds and 10 repetitions for each model. As a result of this study, which proposed a method based on computer aided binary classification of MRS data, in the tests performed with the developed application, an accuracy of 93.44% with LSTM model in glioblastoma brain tumor and pseudo brain tumor classification, an accuracy of 85.56% with Bi-LSTM model in diffuse astrocytoma brain tumor and pseudo brain tumor classification, an accuracy of 88.33% with LSTM model in metastatic brain tumor and pseudo brain tumor classification and an accuracy of 99.23% with Bi-LSTM model in normal brain tissue and pseudo brain tumor classification were obtained.