Multicenter validation of [¹⁸F]-FDG PET and support-vector machine discriminant analysis in automatically classifying...

¹⁸F-Fluorodeoxyglucose (¹⁸F-FDG) positron emission tomography (PET) single-center studies using support vector machine (SVM) approach to differentiate amyotrophic lateral sclerosis (ALS) from controls have shown high overall accuracy on an individual patient basis using local a priori defined classifiers. The aim of the study was to validate the SVM accuracy on a multicentric level.

A previously defined Belgian (BE) group of 175 ALS patients (61.9 ± 12.2 years, 120M/55F) and 20 screened healthy controls (62.4 ± 6.4 years, 12M/8F) was used to classify another large dataset from Italy (IT), consisting of 195 patients (63.2 ± 11.6 years, 117M/78F) and 40 controls (62 ± 14.4 years; 29M/11F) free of any neurological and psychiatric disorder who underwent whole-body ¹⁸F-FDG PET-CT for lung cancer without any evidence of paraneoplastic symptoms. ¹⁸F-FDG within-center group comparisons based on statistical parametric mapping (SPM) were performed and SVM classifiers based on the local training sets were applied to differentiate ALS from controls from the other centers.

SPM group analysis showed only minor differences between both ALS groups, indicating pattern consistency. SVM using BE data set as training, classified 183/193 ALS-IT correctly (accuracy of 94.8%). However, 35/40 CON-IT were misclassified as ALS (accuracy 12.5%). Furthermore, using IT data as training, ALS-BE could not be distinguished from CON-BE. Within-center SPM group analysis confirmed prefrontal hypometabolism in CON-IT versus CON-BE, indicating subclinical brain changes in patients undergoing oncological scanning.

This multicenter study confirms that the ¹⁸F-FDG ALS pattern is stable across centers. Furthermore, it highlights the importance of carefully selected controls, as subclinical frontal changes might be present in patients in an oncological setting.