SpectraClassifier 1.0: a user friendly, automated MRS-based classifier-development system

BMC Bioinformatics

Table 2 Comparing results for the validation of SC, using FS prior to LDA.

	Accuracy ± standard deviation in[44]	Accuracy ± standard deviation in SC
Short TE	88.82 ± 4.51	90.73 ± 1.97
Long TE	82.50 ± 5.31	85.12 ± 2.51
Long + Short TE	88.71 ± 4.54	90.31 ± 2.16

In this example, a classifier for low grade meningioma, aggressive (glioblastomas multiforme + metastasis) and low grade glioma was developed for short, long and the combination by concatenation [44] of long + short TE of SV MRS data from INTERPRET [5]. In [44], k-Random sampling train-test (kRSTT) with stratified test sets with 150 repetitions was the evaluation procedure used. In SC, a bootstrapping method with 1000 repetitions was the one used. Although both methods used to evaluate the classifiers are not exactly the same, both are equivalent sampling methods, therefore their results can be compared. As performance measure, the accuracy and the standard deviation were used.

ISSN: 1471-2105