DocumentCode :
3130732
Title :
Meningioma subtype classification: A survey
Author :
Fatima, Kiran ; Majeed, Hammad
Author_Institution :
Dept. of Comput. Sci., Nat. Univ. of Comput. & Emerging Sci., Islamabad, Pakistan
fYear :
2010
fDate :
18-19 Oct. 2010
Firstpage :
55
Lastpage :
60
Abstract :
Meningioma subtype classification is a complex, real world and multi-class problem from the domain of histopathology. To solve this, various Computer Aided Techniques are employed in the past with a varying degree of success. This classification not only helps the technicians /doctors/pathologists to classify each sample correctly but also assists them in making the right decision for treatment. The main challenge in solving this problem is the inherent complexity due to high intra-class variability and low inter-class variation in the texture of tumor samples. Many attempts have been made and numerous algorithms are introduced till this day. Unfortunately, all of them have shown inconsistent performance. By inconsistent we mean that an algorithm is classifying only a subset of the subtypes correctly, on the rest it´s not much accurate. In this survey we have mentioned a few top of the line techniques introduced recently and analyzed them critically. At the end, on the basis of this survey, we have proposed the best suited environment for the use of each technique and a possible promising future direction.
Keywords :
diseases; medical image processing; pattern classification; tumours; Meningioma subtype classification; computer aided techniques; histopathology; multiclass problem; tumor samples texture; Accuracy; Feature extraction; Support vector machines; Training; Wavelet packets; Texture analysis; feature extraction; histopathological images; meningioma; multiresolution representation; wavelet packet;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Emerging Technologies (ICET), 2010 6th International Conference on
Conference_Location :
Islamabad
Print_ISBN :
978-1-4244-8057-9
Type :
conf
DOI :
10.1109/ICET.2010.5638382
Filename :
5638382
Link To Document :
بازگشت