Optimization method of MRI scan parameters of a double inversion recovery sequence using a T1 map and a developed analysis algorithm

Abstract

BACKGROUND:

Optimizing scan parameters for double inversion recovery (DIR) sequences remains difficult.

OBJECTIVE:

To evaluate a new method for optimizing DIR sequence scan parameters using T1 mapping and a newly developed analysis algorithm.

METHODS:

Twelve healthy volunteers underwent T1 mapping and DIR magnetic resonance imaging. The following steps were used for image optimization including: 1) measurement of gray matter (GM), white matter (WM), and cerebrospinal fluid (CSF) T1 values to create a T1 map; 2) calculation of optimized scan parameters by using a developed analysis algorithm; 3) performance assessment of DIR magnetic resonance imaging by using the calculated optimized imaging parameters. Additionally, we used scan parameters from previous studies to obtain DIR images in order to evaluate our new method. The contrast between GM and suppressed tissues was compared between these images and those obtained using the optimized parameters.

RESULTS:

Using our optimization method, WM and CSF regions were suppressed uniformly for all scan conditions. The contrast was significantly higher in images obtained using this optimization method compared to those obtained using previously published parameters (p < 0.01).

CONCLUSIONS:

It is possible to obtain superior DIR images by using an optimization method that involves T1 mapping and a newly developed analysis algorithm.

Keywords

MRI brain DIR T1 map WAIR

1 Introduction

Gray matter (GM) imaging is important in the investigation of many neurological diseases, including schizophrenia, multiple sclerosis, stroke, Alzheimer’s disease, tuberous sclerosis, and epilepsy, all of which are associated with changes in cortical GM [1 –7]. Unfortunately, neocortical GM is difficult to capture because the neocortex is a thin, highly convoluted structure that folds back upon itself in an unpredictable manner [7]. High-resolution, thin-slice data are therefore necessary to investigate the neocortex. However, voxel size reduction during magnetic resonance (MR) acquisition reduces the signal-to-noise ratio; therefore, a compromise must be determined [7].

A further complication is the partial volume (PV) effect, whereby a single pixel may contain variable amounts of GM, cerebrospinal fluid (CSF), and white matter (WM) [7]. The PV effect can be reduced by reducing voxel size. Additionally, a double inversion recovery (DIR) sequence [8] has been designed to overcome the PV effect. Previous studies [9, 10] have successfully implemented DIR as a two-dimensional (2D) multislice sequence [7]. This DIR sequence extends the commonly used fluid-attenuated inversion recovery sequence for CSF suppression by adding a second 180° radiofrequency (RF) pulse, allowing the simultaneous suppression of signals from two different tissues with different T1 relaxation times [7, 8]. A WM-attenuated inversion recovery (WAIR) image with an enhanced GM region can be obtained using the DIR sequence while suppressing the WM and CSF regions. Similarly, a GM-attenuated inversion recovery (GAIR) image with an enhanced WM region can be obtained with the DIR sequence while suppressing the GM and CSF regions. To suppress two different tissues, the DIR sequence scanning parameters must be optimized. The typical parameters include two different inversion times (TI): a long inversion time (TI-1st), which is the duration between the two inversion pulses; and a short TI (TI-2nd), which is the duration between the second inversion pulse and the excitation pulse. Figure 1 shows the DIR sequence chart.

Fig.1

A sequence chart of the performed DIR MRI. TI-1st is the duration time between the two inversion pulses. TI-2nd is the duration time between the second inversion pulse and the excitation pulse.

It is difficult to optimize the typical parameters for individual DIR MR imaging (MRI). Therefore, nearly all published studies have used parameters from previous reports or parameters that are experimentally determined through phantom studies. In a report published by Redpath and Smith [8], they provide methods on how to determine typical parameters using repetition time (TR), echo time (TE), and the T1 value of each suppressed tissue. The purpose of this study was to evaluate a new method for optimizing DIR sequence scan parameters using a T1 map sequence and a newly developed analysis algorithm.

2 Methods and materials

2.1 Participants and materials

Twelve healthy volunteers (9 men, 3 women; age range, 22–39 years; average age, 25.1 years) were included in this study. All volunteers underwent T1 mapping and DIR MRI with a 1.5-T MRI scanner (Ingenia 1.5T; Philips Medical Systems, Best, The Netherlands). This study design was approved by our Institutional Review Board, and all volunteers provided written informed consent for participation.

2.2 Optimization method for DIR images

Our method for optimizing the DIR sequence scan parameters comprised of the following steps: 1) measurement of GM and WM T1 values on the T1 map obtained from a mixed sequence, 2) calculation of the optimized DIR sequence scan parameters using the developed analysis algorithm, and 3) performance of DIR MRI using the optimized parameters.

Measurement of T1 values

We obtained T1 and T2 maps of each volunteer using a mixed sequence. This mixed sequence utilizes a ratios and least squares algorithm [11] and simultaneously obtains T1 and T2 maps within only a few minutes. Scan parameters of the mixed sequence were as follows: spin echo relaxation time, 1000 ms; inversion time, 2260 ms; TE, 15 and 100 ms; TI, 500 ms; field of view (FOV), 256×256 mm; matrix size, 256×256; slice thickness, 3 mm. GM and WM T1 values were obtained from regions of interest that were drawn on the T1 map (Fig. 2). The T1 value of the CSF region is similar to water and therefore has a long T1 value that is difficult to measure. Therefore, the T1 value of the CSF region was manually defined as 4,250 ms [12].

Fig.2

Measurement of T1 values of WM and GM on the T1 map. The T1 map was automatically obtained using a mixed sequence. It was possible to optimize DIR scan parameters for each individual volunteer using measured T1 values.

Calculation of optimized scan parameters of the DIR sequence

Software was used to calculate optimum TI-1st and TI-2nd values from several initial parameters (TR, TE, and the T1 values of two different suppressed tissues). Figure 3 shows a flow chart of the steps used to calculate the optimum TI-1st and TI-2nd values. The software calculated the simulated signal intensity of WM, GM, and CSF from a range of TI-1st and TI-2nd values using a DIR theoretical equation derived from the Bloch equation. Among the usable range of TI-1st and TI-2nd values, the minimal values of the two suppressed tissues (the WM and the CSF) were determined. The optimal TI-1st and TI-2nd values were those that yielded the most minimal WM and CSF values.

Fig.3

A flow-chart of the steps performed to optimize the TI-1st and TI-2nd scan parameters.

Performance of DIR MRI using the optimized parameters

The TI-1st was the inversion time between the first and second 180° inversion RF pulses, while the TI-2nd was the inversion time between the second 180° inversion RF pulse and the 90° RF pulse. These optimized DIR sequence parameters were calculated from several initial parameters (TR, TE, and others) as well as from T1 values of the suppressed tissues, which were obtained from the mixed sequence. Optimized DIR images, such as WAIR, were obtained from the DIR sequence by using individually optimized TI-1st and TI-2nd parameters. The other scan parameters were as follows: TR, 6500, 9805, and 12400 ms; TE, 50 ms; FOV, 240×240 mm; matrix size, 240×240; slice thickness, 3 mm.

2.3 Physical evaluation

To evaluate our optimization method, we obtained DIR images using scan parameters reported in previous studies [10 , 13–15]. Both these scan parameters and the optimized scan parameters are shown in Table 1.

Table 1
Scan parameters for the DIR sequence. “Optimized” values were calculated using our optimization method. “Previous” values are those provided in previous reports

Setting TR [ms] Optimized Previous [10 , 14] Reference

Tl^lst [ms] Tl^2nd [ms] Tl^lst [ms] Tl^2nd [ms]

3D, 1.5T 6500 2358 370 2350 350 [13] Radiology,

2006;241 : 873–879

2D, 1.5T 9805 2870 374 2600 100 [10] Magn Reson

Imaging, 1998;16 : 127–135

2D, 1.5T 12400 3086 375 3100 330 [14] J Magn Reson

Imaging, 1998;8 : 544–547

Setting	TR [ms]	Optimized	Previous [10 , 14]	Reference
3D, 1.5T	6500	2358	370	2350	350	[13] Radiology,
						2006;241 : 873–879
2D, 1.5T	9805	2870	374	2600	100	[10] Magn Reson
						Imaging, 1998;16 : 127–135
2D, 1.5T	12400	3086	375	3100	330	[14] J Magn Reson
						Imaging, 1998;8 : 544–547

WAIR images were obtained using both optimized scan parameters and previously reported scan parameters, and contrasts between GM and suppressed signals (C-GM) were measured. A high C-GM signifies that GM can be clearly observed. C-GM was calculated using the following formula: $C - GM = {SI (GM) - (SI (WM) + SI (CSF))} / {SI (WM) + SI (CSF)}$ (1) where SI(GM), SI(WM), and SI(CSF) are the respective signal intensities of the GM, WM, and CSF regions.

2.4 Observer evaluation

The WAIR images obtained using the optimized parameters and those obtained using previously published methods were independently evaluated by three reviewers, each with over 8 years of clinical experience, to assess overall image quality. The coordinator of the study showed the images in a random order on a medical liquid crystal display (LCD) for individual evaluation of 5 parameters of image quality by the reviewers, each of whom were blinded to the parameters of the WAIR images. The reviewers graded images for degree of tissue signal suppression (WM and CSF), contrast of GM, image noise, and sharpness (blurring of WM) using a 5-point scale (1, poor; 2, fair; 3, good; 4, very good, 5, excellent).

2.5 Statistical analysis

Statistical analysis was performed using commercial software (Prism 5; GraphPad Software, Inc., San Diego, CA. MATLAB R2016b; Mathworks, Inc., Natick, MA). The contrasts (C-GM and C-WM) between the current method and previous methods were compared using a Wilcoxon rank-sum test. In addition, the five image quality parameters that were scored by observer evaluation were compared via parameter-method using a Wilcoxon rank-sum test.

3 Results

3.1 Results of T1 values of the GM and WM and optimized scanning parameters

Table 2 shows the GM and WM T1 values obtained by mixed sequence. The mean GM value and standard deviation was 545.2±15.8 ms, while the mean WM value and standard deviation was 1199.7±75.0 ms. Table 2 also shows the optimized TI-1st and TI-2nd for each TR condition. These optimized parameter values were similar among all volunteers and were closely related to TR. As a result, personalized scanning parameters were calculated using individual T1 values and our optimization method for each TR condition.

Table 2
T1 values of GM and WM and optimized TI-1st and TI-2nd scanning parameters obtained using the measured T1 values and our optimization method

Case No T1 value [ms] Optimized scanning parameters

TR 6500 [ms] TR 9805 [ms] TR 12400 [ms]

WM GM TI^1st TI^2st TI^1st TI^2st TI^1st TI^2st

l 550 1211 2360 373 2874 378 3091 379

2 567 1089 2369 384 2885 389 3103 391

3 524 1186 2349 357 2857 360 3072 362

4 547 1320 2360 372 2S72 376 3089 377

5 547 1229 2360 372 2872 376 3089 377

6 556 1184 2364 377 2878 382 3095 383

7 558 1190 2365 379 2879 383 3097 385

8 524 1116 2349 357 2857 360 3072 362

9 530 1202 2352 361 2861 365 3076 366

10 551 1163 2362 374 2875 379 3092 380

11 571 1206 2370 386 2890 393 3110 395

12 533 1024 2350 360 2860 365 3080 368

13 530 1103 2352 361 2861 365 3076 366

Case No	T1 value [ms]	Optimized scanning parameters
l	550	1211	2360	373	2874	378	3091	379
2	567	1089	2369	384	2885	389	3103	391
3	524	1186	2349	357	2857	360	3072	362
4	547	1320	2360	372	2S72	376	3089	377
5	547	1229	2360	372	2872	376	3089	377
6	556	1184	2364	377	2878	382	3095	383
7	558	1190	2365	379	2879	383	3097	385
8	524	1116	2349	357	2857	360	3072	362
9	530	1202	2352	361	2861	365	3076	366
10	551	1163	2362	374	2875	379	3092	380
11	571	1206	2370	386	2890	393	3110	395
12	533	1024	2350	360	2860	365	3080	368
13	530	1103	2352	361	2861	365	3076	366

3.2 DIR images obtained with the optimized method

Figure 4 shows representative WAIR images obtained using optimized parameters and those obtained using previously published parameters. Using the optimized DIR sequence scan parameter method, WM and CSF were uniformly suppressed and GM was readily apparent in all scan conditions. In contrast, WM and CSF were not correctly suppressed when the scan parameters from previous studies were used. Figure 5 shows examples of WAIR images with incorrectly suppressed WM and CSF that were obtained using previously reported parameters. When the TR was 6500 ms, the CSF was not correctly suppressed in some cases. Furthermore, when the TR was 9805 or 12400 ms, the WM was not correctly suppressed in some cases. In particular, when the TR was 9805 ms, almost all cases showed WAIR images with poor contrast using the previously published imaging methods. In contrast, it was possible to acquire WAIR images in which WM and CSF were properly suppressed when using optimized scan parameters.

Fig.4

WAIR images obtained using optimized parameters and those obtained using previously reported parameters in one volunteer. The window width was the length between the 5th and 95th percentile values for each image. The window level was the median value of each image. Using optimized parameters, the WM and CSF regions were uniformly suppressed and GM was clearly apparent in all scan conditions. In contrast, GM and CSF were not correctly suppressed when the scan parameters from previous reports were used.

Fig.5

WAIR images in which WM and CSF were not correctly suppressed when using previously reported parameters. The upper row shows the WAIR images obtained using optimized parameters in the same volunteers. By using optimized scan parameters, it was possible to acquire WAIR images in which WM and CSF signals were correctly suppressed.

3.3 Results of the physical evaluation

Figure 6 shows the C-GM results from images obtained under different TR conditions using optimized parameters and those obtained using previously reported parameters. At all TR conditions, the C-GM values were significantly higher when using the optimized parameters compared with the previously published parameters (p < 0.05). This result suggests that our method for optimizing parameters allows for the correct suppression the MRI signals of two different tissues.

Fig.6

GM contrast (C-GM) values for all TRs. For all scan parameters, C-GM was significantly higher in images obtained using optimized parameters compared with those obtained using previously reported parameters (p < 0.05).

3.4 Qualitative evaluation results

Five image quality parameters, including overall image quality, were qualitatively evaluated by three reviewers for both the images obtained using individually optimized parameters as well as those obtained using previously published imaging parameters. The averages of these scores are displayed in Table 3. The scores for WM signal suppression and those for GM contrast were significantly higher for the images obtained using optimized parameters compared to those obtained using previously reported parameters (p < 0.01). When TR was set to 6500 ms, the scores of CSF suppression were higher for the images obtained using optimized parameters compared with those obtained using previously reported parameters (p < 0.05). In other TR conditions, the rating scores of CSF suppression were not significantly different; however, rating scores of CSF suppression were higher than 4.7 in all conditions for images obtained using optimized parameters. There were no significant differences between parameter methods for either noise or sharpness.

Table 3
Average rating scores obtained by observer evaluation. Asterisks indicate significant differences between the images obtained using optimized parameters and those obtained using previously reported parameters (p < 0.05)

TR = 6500 [ms] TR = 9805 [ms] TR = 12400 [ms]

Optimization Previous p Optimization Previous p Optimization Previous p

WM suppression 4.67±0.47 3.56±1.04 <0.01^* 4 69±0.62 1.69±1.24 <0.01^* 4.77±0.42 3.02±1.03 <0.01^*

CSF suppression 4.77±0.42 3.67±1.80 <0.05^* 4 73±0.44 4.58±0.84 0.94 4.71±0.43 4.90±0.31 0.63

GM contrast 4.73±0.44 3.81±0.86 <0.01^* 4 75±0.48 1.85±1.15 <0.01^* 4.75±0.43 3.46±0.89 <0.01^*

Noise 2.83±0.51 2.96±0.20 0.23 2.98±0.14 2.94±0.24 0.31 2.88±0.44 2.96±0.20 0.39

Sharpness 2.98±0.14 2.96±0.20 0.57 2.98±0.14 2.96±0.20 0.57 2.98±0.14 2.96±0.20 0.57

	TR = 6500 [ms]	TR = 9805 [ms]	TR = 12400 [ms]
WM suppression	4.67±0.47	3.56±1.04	<0.01^*	4 69±0.62	1.69±1.24	<0.01^*	4.77±0.42	3.02±1.03	<0.01^*
CSF suppression	4.77±0.42	3.67±1.80	<0.05^*	4 73±0.44	4.58±0.84	0.94	4.71±0.43	4.90±0.31	0.63
GM contrast	4.73±0.44	3.81±0.86	<0.01^*	4 75±0.48	1.85±1.15	<0.01^*	4.75±0.43	3.46±0.89	<0.01^*
Noise	2.83±0.51	2.96±0.20	0.23	2.98±0.14	2.94±0.24	0.31	2.88±0.44	2.96±0.20	0.39
Sharpness	2.98±0.14	2.96±0.20	0.57	2.98±0.14	2.96±0.20	0.57	2.98±0.14	2.96±0.20	0.57

4 Discussion

In this study, we describe a method for optimizing DIR MRI sequence scan parameters that incorporates T1 mapping and a newly developed analysis algorithm. The WAIR images obtained using the resulting optimized parameters in all different scan conditions were similar to each other (Fig. 4), and these images showed significantly higher GM contrast compared to the images obtained using previously reported parameters. In addition, the rating scores of WM and CSF suppression were remarkably high for the images obtained using optimized parameters, and GM contrast rating scores were significantly higher in these images compared with those obtained using previously reported parameters (Table 3). Figure 5 shows WAIR images obtained using previously reported parameters that exhibit incorrectly suppressed CSF and WM. These cases required individually optimized scan parameters in order to obtain WAIR images in which WM and CSF signals were appropriately suppressed. Using our method, it was possible to optimize the scanning parameters for individual participants during MRI examinations. Although it is normally difficult to briefly measure the T1 values of each region, it is possible to obtain T1 values of these regions using a mixed sequence that provides T1 and T2 maps within only a few minutes. Many previous studies have described T1 values [12 , 17]; however, there is a large range for these reported values. Further, T1 values are markedly different in female and male subjects [16]. Therefore, individual optimization of scanning parameters is necessary in order to correctly suppress two different tissues in DIR MRI. We were able to calculate optimized TI-1st and TI-2nd values from the known T1 values of the different suppressed tissues using a newly developed analysis algorithm.

DIR has been most often used to study cortical GM lesions, and the accrual of GM lesions has been shown to be correlated with disability in patients with established multiple sclerosis (MS) [18 –21]. The presence of these lesions may also improve the specificity of current MS MRI diagnostic criteria [18 , 21–23]. A recent combined histopathological and MRI study reported that 90% of cortical MS lesions observed via high-resolution three-dimensional (3D) DIR were confirmed histopathologically [24]. Using WAIR images, it is possible to evaluate and diagnose cortical GM lesions without the complication of the PV effect [7]. Therefore, to correctly diagnose GM lesions, it is desirable to obtain high-resolution WAIR images with correct suppression of GM and CSF signals. Our optimization method provided correctly suppressed DIR images in both standard 2D DIR as well as in high-resolution 3D DIR MRI.

This study has several limitations. First, this was a volunteer study due to the difficulty of obtaining clinical MR images using non-optimized scan parameters. A second limitation is that all MRI vendors lack a convenient sequence, such as the mixed-sequence, which measures T1 values of WM and other tissues in a short period of time, but our optimization method requires measurement of T1 values of the suppressed tissues without the CSF. Fortunately, it has recently been made possible to measure T1 values using the Look-Locker sequence, which can measure T1 values of each tissue in a short time [25]. In the future, easy-to-use measurement sequences, such as the mixed sequence and the Look-Locker sequence, may grow in popularity.

5 Conclusion

Based on our study findings, we conclude that it is possible to obtain correctly suppressed DIR images by using an optimization method for MRI scan parameters that is based on a mixed sequence and a newly developed analysis algorithm. This finding may be valuable for future imaging studies involving degenerative diseases.

Footnotes

Acknowledgments

This work was supported by JSPS KAKENHI Grant Number JP15K19206.

References

Mukherjee

, Bahn

M.M.

, McKinstry

R.C.

, Shimony

J.S.

, Cull

T.S.

, Akbudak

, et al., Differences between gray matter and white matter water diffusion in stroke: Diffusion-tensor MR imaging in 12 patients, Radiology 215(1) (2000), 211–220.

Eriksson

S.H.

, Rugg-Gunn

F.J.

, Symms

M.R.

, Barker

G.J.

and Duncan

J.S.

, Diffusion tensor imaging in patients with epilepsy and malformations of cortical development, Brain 124(Pt 3) (2001), 617–626.

Ridler

, Bullmore

E.T.

, De Vries

P.J.

, Suckling

, Barker

G.J.

, Meara

S.J.

, et al., Widespread anatomical abnormalities of grey and white matter structure in tuberous sclerosis, Psychol Med 31(8) (2001), 1437–1446.

Wilke

, Kaufmann

, Grabner

, Putz

, Wetter

T.C.

and Auer

D.P.

, Gray matter-changes and correlates of disease severity in schizophrenia: A statistical parametric mapping study, Neuroimage 13(5) (2001), 814–824.

Frisoni

G.B.

, Testa

, Zorzan

, Sabattoli

, Beltramello

, Soininen

, et al., Detection of grey matter loss in mild Alzheimer’s disease with voxel based morphometry, J Neurol Neurosurg Psychiatry 73(6) (2002), 657–664.

Miller

D.H.

, Barkhof

, Frank

J.A.

, Parker

G.J.

and Thompson

A.J.

, Measurement of atrophy in multiple sclerosis: Pathological basis, methodological aspects and clinical relevance, Brain 125(Pt 8) (2002), 1676–1695.

Boulby

P.A.

, Symms

M.R.

and Barker

G.J.

, Optimized interleaved whole-brain 3D double inversion recovery (DIR) sequence for imaging the neocortex, Magn Reson Med 51(6) (2004), 1181–1186.

Redpath

T.W.

and Smith

F.W.

, Technical note: Use of a double inversion recovery pulse sequence to image selectively grey or white brain matter, Br J Radiol 67(804) (1994), 1258–1263.

Bedell

B.J.

and Narayana

P.A.

, Volumetric analysis of white matter, gray matter, and CSF using fractional volume analysis, Magn Reson Med 39(6) (1998), 961–969.

10.

Turetschek

, Wunderbaldinger

, Bankier

A.A.

, Zontsich

, Graf

, Mallek

, et al., Double inversion recovery imaging of the brain: Initial experience and comparison with fluid attenuated inversion recovery imaging, Magn Reson Imaging 16(2) (1998), 127–135.

11.

In den Kleef

J.J.

and Cuppen

J.J.

, RLSQ: T1, T2, and rho calculations, combining ratios and least squares, Magn Reson Med 5(6) (1987), 513–524.

12.

Rooney

W.D.

, Johnson

, Li

, Cohen

E.R.

, Kim

S.G.

, Ugurbil

, et al., Magnetic field and tissue dependencies of human brain longitudinal 1H2O relaxation in vivo, Magn Reson Med 57(2) (2007), 308–318.

13.

Pouwels

P.J.

, Kuijer

J.P.

, MuglerJ.P., 3rd, Guttmann

C.R.

and Barkhof

, Human gray matter: Feasibility of single-slab 3D double inversion-recovery high-spatial-resolution MR imaging, Radiology 241(3) (2006), 873–879.

14.

Bedell

B.J.

and Narayana

P.A.

, Implementation and evaluation of a new pulse sequence for rapid acquisition of double inversion recovery images for simultaneous suppression of white matter and CSF, J Magn Reson Imaging 8(3) (1998), 544–547.

15.

Bender

and Klose

, Double inversion recovery: Impact of incidental magnetic transfer effects on optimal inversion times, Invest Radiol 45(4) (2010), 196–201.

16.

Wansapura

J.P.

, Holland

S.K.

, Dunn

R.S.

and Ball

W.S.

Jr , NMR relaxation times in the human brain at 3.0 tesla, J Magn Reson Imaging 9(4) (1999), 531–538.

17.

Shin

, Gu

and Yang

, Fast high-resolution T1 mapping using inversion-recovery Look-Locker echo-planar imaging at steady state: Optimization for accuracy and reliability, Magn Reson Med 61(4) (2009), 899–906.

18.

Filippi

, Rocca

M.A.

, Calabrese

, Sormani

M.P.

, Rinaldi

, Perini

, et al., Intracortical lesions: Relevance for new MRI diagnostic criteria for multiple sclerosis, Neurology 75(22) (2010), 1988–1994.

19.

Geurts

J.J.

, Pouwels

P.J.

, Uitdehaag

B.M.

, Polman

C.H.

, Barkhof

and Castelijns

J.A.

, Intracortical lesions in multiple sclerosis: Improved detection with 3D double inversion-recovery MR imaging, Radiology 236(1) (2005), 254–260.

20.

Polman

C.H.

, Reingold

S.C.

, Banwell

, Clanet

, Cohen

J.A.

, Filippi

, et al., Diagnostic criteria for multiple sclerosis: 2010 revisions to the McDonald criteria, Ann Neurol 69(2) (2011), 292–302.

21.

Sethi

, Muhlert

, Ron

, Golay

, Wheeler-Kingshott

C.A.

, Miller

D.H.

, et al., MS cortical lesions on DIR: Not quite what they seem? PLoS One 8(11) (2013), e78879.

22.

Sethi

, Yousry

T.A.

, Muhlert

, Ron

, Golay

, Wheeler-Kingshott

, et al., Improved detection of cortical MS lesions with phase-sensitive inversion recovery MRI, J Neurol Neurosurg Psychiatry 83(9) (2012), 877–882.

23.

Calabrese

, Battaglini

, Giorgio

, Atzori

, Bernardi

, Mattisi

, et al., Imaging distribution and frequency of cortical lesions in patients with multiple sclerosis, Neurology 75(14) (2010), 1234–1240.

24.

Seewann

, Kooi

E.J.

, Roosendaal

S.D.

, Pouwels

P.J.

, Wattjes

M.P.

and van der Valk

, et al., Postmortem verification of MS cortical lesion detection with 3D DIR, Neurology 78(5) (2012), 302–308.

25.

Freeman

, Gowland

and Mansfield

, Optimization of the ultrafast Look-Locker echo-planar imaging T1 mapping sequence, Magn Reson Imaging 16(7) (1998), 765–772.