Multiproject–multicenter evaluation of automatic brain tumor classification by magnetic resonance spectroscopy

Abstract

Justification Automatic brain tumor classification by MRS has been under development for more than a decade. Nonetheless, to our knowledge, there are no published evaluations of predictive models with unseen cases that are subsequently acquired in different centers. The multicenter eTUMOUR project (2004)(2005)(2006)(2007)(2008)(2009), which builds upon previous expertise from the INTERPRET project (2000INTERPRET project ( -2002 has allowed such an evaluation to take place. Materials and Methods A total of 253 pairwise classifiers for glioblastoma, meningioma, metastasis, and low-grade glial diagnosis were inferred based on 211 SV short TE INTERPRET MR spectra obtained at 1.5 T (PRESS or STEAM, 20-32 ms) and automatically pre-processed. Afterwards, the classifiers were tested with 97 spectra, which were subsequently compiled during eTUMOUR. ResultsIn our results based on subsequently acquired spectra, accuracies of around 90% were achieved for most of the pairwise discrimination problems. The exception was for the glioblastoma versus metastasis discrimination, which was below 78%. A more clear definition of metastases may be obtained by other approaches, such as MRSI + MRI. Conclusions The prediction of the tumor type of in-vivo MRS is possible using classifiers developed from previously acquired data, in different hospitals with different instrumentation under the same acquisition protocols. This methodology may find application for assisting in the diagnosis of new brain tumor cases and for the quality control of multicenter MRS databases.

Morphometric model for discrimination between glioblastoma multiforme and solitary metastasis using three‐dimensional shape analysis

“…Moreover, lower classification rates (60-F o r P e e r R e v i e w 6 80%) or <0.8 average AUC were reported to distinguish GBM and MET in other previous studies (13,14,16,17,(22)(23)(24)(25).…”

Pattern Recognition Analysis of MR Spectra

“…45,46 The explanation is that the superior limit of the voxel was close to the skull (see the brain circumvolutions), where contamination from subcutaneous fat was likely classification techniques. 43 Despite the fact that classes sometimes have overlapping spectra, linear and nonlinear classifiers usually perform similarly 44 for these types of discriminations. While this may seem disappointing for PR practitioners, it has an interesting corollary: spectra that are consistently misclassified by different PR techniques will be, most probably, outliers ( Figure 5).…”

Using magnetic resonance imaging and spectroscopy in cancer diagnostics and monitoring

“…94 1 H MRS has also been used to identify the most active regions of large complex tumors thereby aiding tumor biopsy 95,96 and for delineating tumor margins which can extend beyond the enhancing regions seen on conventional MRI. 95 The entire MRS profile has been shown to be a strong characteristic of tumor type and has been studied extensively as an aid to non-invasive diagnosis, the best results coming from the use of pattern recognition [97][98][99][100] Large multi-center studies have shown that this can be robust even when evaluated prospectively 101 and clinical decision support systems based on MRS have been developed. 102,103 High total choline and mobile lipids together with low myo-inositol (for an example brain tumor 1 H MRS see Fig.…”