![Institutional Repository [SANDBOX]](images/dias.logo.png "Institutional Repository [SANDBOX]"){kind=logo}
Institutional Repository [SANDBOX]
Technical University of Crete
Technical University of Crete
EN
|
EL
| URI: | http://purl.tuc.gr/dl/dias/749A3EC4-0658-4CBC-96C6-686B657F17CF | ||
| Year | 2018 | ||
| Type of Item | Conference Full Paper | ||
| License |
|
||
| Bibliographic Citation | E. Sergaki, G. Spiliotis, I. O. Vardiambasis, T. Kapetanakis, A. Krasoudakis, G. C. Giakos, M. Zervakis and A. Polydorou, "Application of ANN and ANFIS for detection of brain tumors in MRIs by using DWT and GLCM texture analysis," in IEEE International Conference on Imaging Systems and Techniques, 2018. doi: 10.1109/IST.2018.8577099 https://doi.org/10.1109/IST.2018.8577099 | ||
| Appears in Collections |
In this work we combine different methodologies in order to develop algorithms for Computer-Aided Diagnosis (CAD) for brain tumors from the axial plane (T2 MRI). All methods utilize texture analysis by extracting features from raw data, without post-processing, based on different techniques, such as Gray Level Co-Occurrence Matrix (GLCM), or Discrete Wavelet Transform (DWT) and different classification methods, based on ANN or ANFIS. All of our proposed methodologies are developed, validated and verified on various sub data including 65% non-healthy MRIS. The total used database consists of 202 MRIs from non-healthy patients and 18 from healthy, segmented visually by an experienced neurosurgeon. Combining different subsets of features, our best results are by using 4 GLCM features for a 4 input and two hidden layers ANN, giving sensitivity 100%, specificity 77.8% accuracy 94.3%. It is proved that the input data to train such a CAD are considered to be unbiased if the ratio between healthy/un-healthy tissue MRIs is about 35%/65%, respectively.
Copyright © DIAS 2013
