Multiple sclerosis (MS) is the leading cause of non-traumatic disability in young adults, even if new immunomodulatory and immunosuppressive treatments can now slow down the disease activity and the progression of disability [1,2]. Diagnosis and follow-up of MS are based on magnetic resonance imaging (MRI), which can reveal demyelinating lesions in four typical locations (cortical/juxtacortical, juxtaventricular, infratentorial and medullary). This specific distribution, justifying the spatial dissemination criteria, is included in the 2017 McDonalds criteria for an early diagnosis of MS [3].

The capabilities of double-inversion recovery (DIR) sequence (a T2-weighted imaging with nulling of both white matter (WM) and cerebrospinal fluid signal [4]) for the detection of intracortical [5], juxtacortical [6], infratentorial [7], medullary and optic nerve [8] MS lesions is now well documented. These performance made DIR part of the optional sequences in the 2021 Magnetic Resonance Imaging in Multiple Sclerosis (MAGNIMS) European guidelines [9] and recommended by academic institutions such as OFSEP (Observatoire Français de la sclérose en plaques) [10]. Despite these technical improvements, the conventional three-dimensional (3D) DIR sequence remains relatively long (> 4 min), still limiting its routine use. More recently, the controlled aliasing in parallel imaging results in higher acceleration (CAPIRIHNA) technique has been proposed to further shorten the acquisition time of 3D MRI sequences, while attenuating the noise penalty compared to generalized autocalibrating partial parallel acquisition (GRAPPA) [11].

In this study, a highly-accelerated double inversion recovery (fast-DIR) sequence accelerated with CAIPIRINHA, was obtained using a very high acceleration factor (i.e., 3 × 3). Two features have been integrated in this accelerated sequence: (i), image iterative denoising [12] to compensate for the increasing noise inherent to parallel imaging acceleration, and (ii), magnetization T2-preparation [13] to further increase the contrast between MS lesions, WM and gray matter (GM) [14]. Improvement in 3D DIR imaging provided by T2-preparation was recently demonstrated in the detection of cortical, optic nerve, and cervical cord MS lesions [15], yet full potential of iterative denoising was expected to be realized only for highly accelerated acquisitions with otherwise low signal.

The purpose of this study was to compare image quality and diagnostic performance for the detection of juxtacortical MS lesions between fast-DIR and conventional DIR (conv-DIR) sequences. A secondary objective, was to compare the sensitivities the two DIR sequences for infratentorial MS lesion detection.