Background

Compared to the conventional magnetization-prepared rapid gradient-echo imaging (MPRAGE) MRI, magnetization prepared 2 rapid acquisition gradient echoes (MP2RAGE) (Marques, J. P., 2010) shows a higher brain tissue and lesion contrast in multiple sclerosis (MS) patients. This specialized sequence is, however, mainly limited to research settings and not widely acquired in clinical routine.

Objectives

To synthesize realistic-looking MP2RAGE images from MPRAGE acquisitions via generative adversarial network (GAN) and verify if these improve the performance of automatic MS lesion and brain tissue segmentation tools.

Methods

We propose a GAN inspired by the pix2pix framework (Isola, P., 2018) which takes as input 2D slices of 3D MPRAGE images and generates a synthetic MP2RAGE (synMP2RAGE). Differently from pix2pix, the generator of the GAN combines three loss functions: a pixel-wise L1 loss, an adversarial loss, and a perceptual loss. Our framework is trained on 12 healthy controls and 8 MS patients and tested on 36 MS patients, for which an expert manually delineated cortical and white matter lesions. Imaging was performed with a 3T MRI scanner (Siemens Healthcare, Erlangen, Germany) with a 1x1x1.2 mm resolution. Evaluation is performed with reference-based metrics and through automatic segmentation of MS lesions (La Rosa, F., 2020) and brain tissue (Avants, B.B., 2011).

Results

Considering as reference the acquired MP2RAGE, synMP2RAGE achieves a peak signal-to-noise ratio of 31.39, normalized root mean square error of 0.13, and structural similarity index of 0.98, overperforming the MPRAGE (29.49, 0.17, 0.97, respectively) for all metrics. Compared to the initial MPRAGE it also significantly improves the lesion and tissue segmentation masks in terms of the Dice coefficient and volume difference (p-values < 0.001). On the contrary, no significant differences between the real and synMP2RAGE are found in the patient-wise comparison of the lesions’ segmentation (p-values > 0.05), whereas they are significant between MPRAGE and MP2RAGE (p-value < 0.001).

Conclusions

Our proposed framework successfully translates MPRAGE to MP2RAGE, synthesizing realistic-looking images which improve the performance of automatic segmentation tools tested on MS patients. In accordance with previous claims (Finck, T., 2020), these results confirm that GANs can be helpful in the automatic analysis of MRI images.

Background

Although conventional MRI acquisitions are of essence in the monitoring of MS, they show low specificity towards the microstructural nature of tissue alterations and exhibit rather low correlations with clinical metrics (“clinico-radiological paradox”). Conversely, recent advances in brain relaxometry allow characterizing microstructural alterations on a single-subject basis; the question yet remains whether such quantitative measurements can help bridging the gap between radiological and clinical findings.

Objectives

This study investigates whether automatically assessed alterations of T₁ relaxation times in brain lesions and normal-appearing white matter (NAWM) improve clinico–radiological correlations in early MS with respect to conventional measures.

Methods

102 healthy controls (65% female, [21-59] y/o) and 50 early-MS patients (76% female, [19-52] y/o) underwent MRI at 3T (MAGNETOM Skyra, Siemens Healthcare, Erlangen, Germany). The employed 3D protocol comprised MPRAGE, FLAIR (both used for lesion segmentation as in [Fartaria et al., 2017, MICCAI]), and MP2RAGE for T₁ mapping.

After the healthy controls’ data were spatially normalized into a study-specific template, reference T₁ values in healthy tissues were established by linear, voxel-wise modelling of the T₁ inter-subject variability [Piredda et al., MRM, 2020]. In the MS cohort, T₁ deviations from the established references were calculated as z-score maps.

Correlations between the EDSS and conventional measures, i.e. lesion volume and count, were compared against correlations with z-score-derived metrics in lesions and NAWM, namely the volume of voxels exceeding a given z-score threshold.

Results

Correlations between EDSS and lesion volume and count were found to be 0.23 and 0.18, respectively. Higher correlations were found between EDSS and the volume of voxels exceeding an absolute z-score threshold of 2, both in lesions and NAWM, with ρ=0.3 and ρ=0.33, respectively. Correlation further improved when considering only negative z-scores, ρ=0.36 for lesions and ρ=0.39 for NAWM. The highest correlation was found when considering absolute z-scores in the occipital lobe NAWM, ρ=0.47.

Conclusions

Microstructural alterations identified as T₁ z-scores were found to improve clinico–radiological correlation in comparison to conventional measures (lesion volume and count). Of notice, negative z-scores (i.e. abnormal T₁ shortening), which may be due to an increase in iron content, appear to be a potential predictor for the clinical state of an early MS patient.