Predictive value of imaging markers at multiple sclerosis disease onset based on gadolinium- and USPIO-enhanced MRI and machine learning

Abstract

Objectives: A novel characterization of Clinically Isolated Syndrome (CIS) patients according to lesion patterns is proposed. More specifically, patients are classified according to the nature of inflammatory lesions patterns. It is expected that this characterization can infer new prospective figures from the earliest imaging signs of Multiple Sclerosis (MS), since it can provide a classification of different types of lesions across patients.

Methods: The method is based on a two-tiered classification. Initially, the spatio-temporal lesion patterns are classified. The discovered lesion patterns are then used to characterize groups of patients. The patient groups are validated using statistical measures and by correlations at 24-month follow-up with hypointense lesion loads.

Results: The methodology identified 3 statistically significantly different clusters of lesion patterns showing p-values smaller than 0.01. Moreover, these patterns defined at baseline correlated with chronic hypointense lesion volumes by follow-up with an R(2) score of 0.90.

Conclusions: The proposed methodology is capable of identifying three major different lesion patterns that are heterogeneously present in patients, allowing a patient classification using only two MRI scans. This finding may lead to more accurate prognosis and thus to more suitable treatments at early stage of MS.

Conflict of interest statement

Competing Interests: Gilles Edan received research support and compensation as a speaker from Biogen Idec, Serono and Sanofi-Aventis, Bayer Schering Pharma AG, and LFB; and has acted as a consultant for Teva Pharmaceuticals, Merck-Serono, Bayer-Schering, Biogenidec, and LFB. This does not alter the authors' adherence to all the PLOS ONE policies on sharing data and materials.

Figures

Figure 1. Classification work-flow showing all the steps of the proposed framework.

Figure 2. The feature extraction process for a single time point.

First all the lesions are delineated, then all the identified voxels are considered to compute the hollow index and to build the covariance matrix. Finally, the eigenvalues are obtained from this covariance matrix. The process is repeated for all time points and the lesions which match at the different time point are ordered in the same feature vector.

Figure 3. The same lesion at the same time point: (a) Gd-enhanced, (b) USPIO-enhanced and (c) pre-contrast.

It can be noticed that the USPIO enhancements are generally very mild compared to the Gd enhancements.

Figure 4. Illustration of a spatio-temporal evolution of the same lesion for both contrast agents and pre-contrast belonging to C 1.

In general, is the less specific which comprises lesions of different dimensions (small, medium, large) appearing at the first time point and then disappearing, and generally Gd-enhanced only.

Figure 5. Illustration of a spatio-temporal evolution of the same lesion for both contrast agents and pre-contrast belonging to C 2.

In general, includes relatively medium and large lesions present at both the first two time points, and with co-presence of ringing USPIO and focal Gd enhancement.

Figure 6. Illustration of a spatio-temporal evolution of the same lesion for both contrast agents and pre-contrast belonging to C 3.

In general, comprises relatively medium lesions present mainly at the first time point with non focal USPIO and Gd enhancement.

Figure 7. Patients according to their chronic hypointense lesions and TLLs by .

The red stars are patients of Group A reported in Table 2 which presented at least one lesion pattern or , the green diamonds are the patients of Group C with no active lesions at the two time points, and the black circles are the reminding patients of Group B.