0:02

Hello, and welcome to Noon Conference hosted by Medality.

0:05

Noon Conference connects the global radiology community through free live

0:09

educational webinars that are accessible for all and is an opportunity to learn

0:13

alongside top radiologists from around the world.

0:16

Today, we're honored to welcome Dr.

0:17

Khalid Gad for a lecture entitled "MRS in Practice: Do It

0:21

Right, Read It Right, Use It Right."

0:24

Dr. Gad serves as the clinical lead of functional and advanced neuroimaging at Ibn

0:28

Sina Neuro Specialty Hospital in Kuwait.

0:31

He has gained extensive international exposure in neuroimaging, including a

0:34

neuroradiology fellowship at Johns Hopkins and a master's degree in advanced

0:38

neuroimaging from University College London.

0:41

He remains academically active and engaged in teaching, supported by qualifications

0:45

that include the FRCR and a master's in health professions education from

0:49

Maastricht University. At the end of the lecture, please join Dr.

0:53

Gad in a Q&A session where he will address questions you may have on today's topic.

0:57

Please remember to use the Q&A feature to submit your questions so we can get to as

1:00

many as we can before our time is up.

1:02

With that, we are ready to begin today's lecture. Dr.

1:04

Gad, please take it from here.

1:07

Well, thank you very much. Hello, everyone, and thanks Medality for

1:11

having me today, delivering a talk on MR spectroscopy,

1:15

which is a very valuable technique in neuroimaging.

1:19

However, in my daily practice, I often

1:24

hear some sort of skepticism regarding

1:27

the clinical utility of the technique.

1:30

Maybe due to unfamiliarity sometimes or maybe due to concerns

1:33

about the quality of the technique, which has some

1:37

limitations or challenges, and also variability among different sites or

1:41

even among different patients or cases.

1:45

So yes. So I totally understand these limitations or challenges, but

1:49

I very much understand that there are three main key factors

1:53

that can make MR spectroscopy easy to interpret.

1:57

Ensuring technical quality and to learn how to read the

2:01

peaks very well,

2:03

and also to know when to use MRS under appropriate

2:07

clinical context and conditions.

2:10

And this is going to be the framework of my talk today.

2:12

Let's start with ensuring technical quality.

2:15

Very fresh,

2:17

very basic background here.

2:20

MRS is basically two types, single-voxel versus

2:23

multi-voxel spectroscopy, where single voxel uses

2:27

one

2:29

voxel

2:31

that is placed over the lesion or the tissue and gives you a good

2:34

SNR,

2:36

but with limited spatial information, of course.

2:39

Compared to multi-voxel spectroscopy, where you have a grid

2:43

of voxels placed over the tissue or over the lesion, and of course, you

2:47

get much better spatial information by these spatial

2:51

encoding steps, either in 2D or 3D, but at the expense of the SNR,

2:55

which is relatively limited with multi-voxel spectroscopy.

2:59

And this is what we call chemical shift imaging

3:03

or MRS imaging, where results can be represented by

3:07

either single spectra related to every single voxel

3:10

or a spectral map, with all voxels can be seen at the

3:14

same time, or metabolite images, where you can color code

3:18

those different metabolites or sometimes the ratios between metabolites.

3:23

Well, so how do we acquire

3:26

the MRS scan? So basically, and everybody knows that there are

3:30

two basic MRS sequence designs. One of them is the

3:34

PRESS, and the other one is called STEEM.

3:36

And it's basically the way the order of sequences and the

3:42

pulse design here. The bottom line is that you have a

3:45

90-degree excitation pulse, and you have either three

3:49

90 degrees pulses, radio frequency pulses, or you have one

3:53

excitation pulses, then followed by two 180 refocusing

3:57

pulses. So the timing intervals between those pulses

4:01

and the

4:03

design of these radio frequency pulses is the core

4:08

difference between both the PRESS and the STEEM. So what is PRESS?

4:11

So PRESS or point resolve spectroscopy

4:14

is the most popular in clinical work.

4:18

So it's the popular sequence design in our daily practice, and it gives

4:22

you powerful SNR. SNR is usually double for

4:26

the SNR as compared to the other sequence design, which

4:29

is the STEEM or stimulated echo acquisition mode.

4:33

That is not frequently used in our clinical work.

4:35

So it's basically used in research studies.

4:39

And it gives you a smaller SNR, almost

4:42

50% less than PRESS, but at the same time, it can enable you to

4:46

do very short TEs under 20

4:49

milliseconds. The third type, and it's most widely adopted in

4:54

3T systems nowadays, is the SEMI laser

4:57

spectroscopy. And the SEMI laser

5:01

spectroscopy is the best

5:05

choice for

5:07

the best spatial localization. So it gives you much better spatial

5:11

localization in less scan time that ends up having

5:15

lower SAR, of course.

5:18

And then we'll talk today about the spectral editing techniques very briefly.

5:21

And most popular technique used is called MEGA PRESS,

5:25

where it uses selective spectral

5:30

pulses to disrupt the natural evolution of unique

5:33

metabolites. So it can give you opportunity to see metabolites that we

5:37

haven't been able to do in the past, and we'll talk about this

5:41

shortly or briefly todaySo the

5:45

question that always comes from my techs, and I always encourage my residents and

5:49

fellows to do that, to be reachable during the MRI scan, to be

5:52

approachable to the techs. And the reason is that there are some

5:56

MRI parameters that require the neuroradiologist to be

6:00

available and give your input. And these

6:04

parameters, the three main parameters are either the voxel, the

6:07

voxel itself, the placement of the voxel, and of course the voxel size, or

6:12

the number of excitations, and also the TE.

6:15

So remember these three factors very well.

6:17

So why voxel size is important? Because it is

6:21

proportional to the SNR. And it's

6:25

straightforward. So whenever you want to increase your SNR by double-fold,

6:30

you simply increase your voxel size by a factor of two.

6:34

And this is the best strategy and most effective strategy to get a good

6:38

SNR. But the challenges are always that you don't

6:41

want your voxel to be contaminated by unnecessary

6:45

voxels or tissues around, like normal brain tissue, for example.

6:48

You need to focus on the lesion. And sometimes you need to avoid other adjacent

6:52

tissues, such as the skull, for example, to avoid fat and to avoid

6:56

susceptibility. Maybe if the lesion is close to the CSF, you need also to avoid

7:00

that because you'll get much more water into your

7:04

voxel, which you need to avoid, and even CSF has also some

7:08

higher concentrations of lactate, which is, again...

7:11

So the bottom line is that make your voxel size the maximum,

7:15

but without being contaminated from unnecessary tissues.

7:19

So if you can't go by large size, then the

7:23

other strategy you can use also is to increase your

7:27

number of excitations, which will also increase your SNR here.

7:31

But the thing is, the issue is the SNR is

7:35

proportional not to the number of excitations, but to the square root of the number

7:39

of excitations. Which means that

7:42

if you want to increase your SNR

7:45

two times, you need to increase the number of excitations by a factor of

7:49

four.

7:50

And of course, the price will be the scan time.

7:53

You get less tolerance from the patients, more motion artifacts.

7:57

You will affect the workflow of your cases, and so on.

8:01

So this is not the best strategy. The best strategy is always to increase your

8:04

voxel size, as long as you keep it away from

8:08

other unnecessary tissues around. So what about the TE?

8:13

So we basically use three TEs in clinical MRIs, either short

8:17

TE, and

8:20

short reminds me with so many peaks.

8:22

So you get so many peaks when you do short TE.

8:25

But when you do long TE, you get less peaks.

8:28

And if you use intermediate TE, you basically

8:32

invert your peaks. So some peaks can get inverted with

8:36

the use of intermediate TE. And you can see here how

8:40

short TE can increase the number of peaks, but also

8:44

can get some sort of distortion at the baseline.

8:46

So you will end up having much more peaks.

8:50

Some of them are true and others also just represent noise.

8:54

While using intermediate or long TE, you will remove the

8:58

noise, you will get better SNR, especially with intermediate TE,

9:02

you will get better SNR compared to long TE.

9:05

But intermediate TE, the best advantage or the most

9:09

popular use of intermediate TE is to invert some peaks, as in this case,

9:14

where you can see the lactate here resonates at 1.3 ppm in long

9:18

TE, but intermediate TE, it will invert,

9:22

giving you more confidence saying that this is lactate.

9:26

So what else do you need to consider when it comes to the quality of the

9:30

spectrum? You need to consider whether you use 3 Tesla or

9:34

1.5. Of course, 3 Tesla is better than 1.5 Tesla.

9:37

And the reason is not just simply because of the SNR, but

9:41

imagine that when you do a 3 Tesla, you stretch your

9:45

baseline by a factor of two. So you double your

9:48

baseline

9:50

spectral quality. So any

9:53

overlapping peaks or those tiny peaks here that you

9:57

can't visibly recognize

10:01

using 1.5 Tesla will be easy to recognize and will be stretched

10:04

apart from each other using 3 Tesla, and this is the most

10:08

important use of 3 Tesla.

10:12

The other thing, of course, is magnetic inhomogeneity, which is controlled by

10:15

shimming, and most of our machines today do automatic shimming

10:20

to control the quality and the bandwidth of your peaks,

10:24

and then doing water and sometimes fat suppression is

10:27

critical before doing spectroscopy because water is abundant in the

10:31

brain tissue. So once you suppress water by using CHESS or

10:35

some sort of a frequency selective pulse, you make

10:39

everything else visible by 100 times.

10:43

So it's very important to suppress water, but it's also very important to

10:46

avoid water. So imagine that this is the message I

10:50

want to deliver here. It's important to suppress water, but it's more important

10:54

to avoid water in your voxel. So what we do is

10:58

basically, we suppress outer volume by applying our

11:02

sequences. So once we suppress the outer volumes

11:06

by using these sequence designs and different

11:10

gradients, we end up having the voxel

11:13

clean and not contaminated by other tissues.

11:17

And when we do, like in this example, if you do multi-voxel

11:20

spectroscopy, we can use what we call saturation bands at different

11:24

planes to suppress all the voxels around

11:28

the lesion, not in the brain, but basically, we need to suppress

11:32

the skull here and also the air outside, so you end up having

11:36

a clean spectrum here with outer volume suppression compared to this

11:40

part of the image where no outer volume suppression is

11:43

used.Well, so there are some metabolites that I call

11:46

troublemakers or unwanted guests that you need to avoid in your voxel.

11:51

So maybe air, if your voxel is close to the maxillary

11:55

sinus, for example, or the skull base or the mastoid air cells, this is a

11:58

source of susceptibility, and you need to always to avoid that.

12:02

CSF, as I mentioned before, because of the water and because of also the higher

12:06

concentration of lactate. Bone, of course, because of susceptibility and lipid.

12:10

Blood, calcium. So all these guests are unwanted in your voxels

12:14

and this

12:16

is why your voxel size is always challenged by all

12:19

these tissues. What about gadolinium?

12:23

There

12:24

used to be some sort of debate because basically and theoretically,

12:28

gadolinium may affect the spectrum by causing some

12:32

broadening of the peaks and reducing the SNR relatively.

12:35

But in practice, what we see and what many experts

12:39

also see is that the effect is not major.

12:43

So the difference is very tiny. Which means that yes,

12:47

it's better to do your MRS without giving contrast or pre-contrast, but

12:51

sometimes you cannot do that because you can't see the lesion.

12:54

You need to see the lesion first. You need to differentiate the lesion from the

12:57

cystic part or from the necrotic areas.

12:59

You need to differentiate the lesion from the edema around it.

13:02

So sometimes it's impossible, so you have to give contrast first.

13:05

And in these cases, I would say that if and only if you

13:09

can postpone the MRS to make a gap

13:13

or a lag for 10 to 15 minutes after giving gadolinium, this

13:17

would give you the best results. But again,

13:20

that doesn't need to overwhelm you.

13:23

So if you have nothing to do other

13:27

than doing spectroscopy just after giving contrast,

13:31

you would expect a good result as well, and the difference is not

13:35

major. Well, let's move to the next key

13:38

factor. The next key factor is to learn how to read the peaks.

13:43

Reading MRS, for me, it looks like

13:47

looking at a silhouette of a city, looking at a city silhouette

13:51

of one of the cities you've visited before.

13:52

So if you've visited one of these cities before, it will be very easy for

13:56

you to recognize what is that city just by

14:00

looking at the silhouette of the city.

14:04

Of course, the one here in the top right corner of the image

14:08

is New York, just by looking at the silhouette.

14:11

And experts do the same thing. They look at spectroscopy, and they can

14:15

just figure out what's happening here. Is this a benign lesion?

14:19

Is this an neoplastic process? This is what they do.

14:22

But today, we need to learn as beginners, because I assume that

14:25

many of the audience today

14:29

are beginners, so we need to learn as beginners how to read

14:32

spectroscopy. Let's start by this.

14:35

Is this a short or long TE? You can see less peaks,

14:39

less numbers of peaks. You have only three peaks visible, so it's most

14:42

likely a long TE,

14:45

not short TE. So my way, my strategy reading

14:48

spectroscopy, I always use the number two because this is my

14:52

magic number. How is that possible?

14:54

Let's use two and start at the scale of the ppm here

14:59

at number two, and this is where you expect to see

15:03

the highest peak. This is NAA, the peak that

15:07

has two As, which is NAA. Then

15:10

move to the left side of this peak.

15:14

You will find two other peaks, and they both have

15:18

the letter C, so you'll have two Cs because they represent two other

15:22

peaks. One is choline, and the other one is creatine.

15:26

And then move to the other side of this large peak and make sure

15:30

that you don't have this two top peak, which is

15:34

called lactate on the right side of the image here.

15:38

So this is my magic two number, and this is the way I

15:42

start reading spectroscopy. Start at two, you'll find

15:46

this NAA that has two As, and then go to the other

15:50

side here, you'll see two peaks. One is choline, and the other one is

15:53

creatine. And then move to the left side or the right side and make sure that you

15:57

don't have this two top peak that

16:00

represents lactate. So this was long

16:04

TE, the clean one, the simple one. What about

16:08

short TE that has many peaks? I will always

16:11

begin with the same strategy. I will look at the three major

16:15

peaks here. I'll start at two, finding the NAA, then

16:19

go to this side, looking at choline and creatine, and then

16:23

move to the far left side of the image because this peak

16:27

here usually represents another creatine peak other than this

16:31

one. This is the main creatine peak, and this is the other creatine peak.

16:35

So I will always start here because I will need this

16:38

creatine to complete the rest of my exercise.

16:43

Well, then I will split the spectrum into three zones.

16:47

Zone number one, where I should look for my inositol and

16:51

other peaks, such as glycine and taurine.

16:54

And then zone number two, where

16:57

GLX, which is glutamine and glutamate, and other amino

17:01

acids are usually found. And then zone number three,

17:05

remember, lactate lives here, but lipid also is here,

17:09

and alanine and some amino acids as well.

17:12

So this is the three-zone approach that I used following

17:16

the three-peak approach using the number T or

17:19

the magic number T. And this is the way I read spectroscopy

17:23

as a beginner.

17:26

There's one piece of information that is critically important.

17:30

You remember that we mentioned that lactate inverts on TE, but it's not

17:33

only lactate. There are many metabolites that can invert

17:37

on intermediate TE or depending on

17:41

the TE value. And for example, my

17:45

inositol can invert in zone one using a shorter TE

17:49

even than the one that we use for intermediate TE in clinical practice.

17:53

But most importantly, and remember this very wellIt is

17:56

smaller. So the peak gets smaller when you use intermediate

18:01

TE, the clinical intermediate TE that we use every day.

18:05

In zone number two, both glutamate and glutamine can invert, and in

18:09

zone number three, as I mentioned, lactate can invert,

18:12

but also alanine can do the same thing.

18:17

Well, so what about ratios, integrals, amplitudes,

18:21

and what we call Hunter's angle? Everyone knows what's Hunter angle.

18:24

Hunter angles is this angle between the peaks of choline and

18:28

NAA, and normally you should have it

18:32

like that. So you should have the choline much shorter than

18:36

NAA, because NAA is a neuronal marker, but choline is a cellular

18:41

proliferation marker. So in normal conditions in the brain,

18:45

NAA should be higher than choline.

18:47

But the challenge is here, are you using amplitude or

18:51

integrals? What is the difference?

18:54

When you do shimming, if you have good shimming, you will end up having

18:58

this peak, very narrow bandwidth.

19:01

And in these cases, the amplitude equals the integral.

19:05

The integral is actually the area under the curve.

19:08

But if you have poor shimming as this one coded by yellow here,

19:12

you will end up having a much shorter peak here because of the

19:16

reduced amplitude. However, the area under the curve is the same,

19:20

so the integral is the same, but the amplitude is much lower.

19:25

And this is very confusing. So I would

19:28

suggest that you always use integrals, not

19:31

amplitudes. And if you use amplitude in your clinical practice, you

19:35

should be very careful, and you should

19:38

have some sort of experience reading that.

19:41

And the way you differentiate is by looking at this number.

19:44

Look carefully at this number. This is the peak value,

19:48

quantitative value, but it's actually preceded by the letter I.

19:52

Whenever you see this I, this means that this number represents

19:56

here the integrals. But if it is not present, if you don't see this

20:00

I, that means that this number represents the

20:04

amplitudes, and of course, this is

20:07

actually, you can change this in the settings of the software on your

20:11

scanner. So my piece of advice here

20:15

is, it's better to use integrals than amplitudes while

20:19

reading the spectrum.

20:21

One more thing to consider, and very importantly, there are some

20:25

variations related to age, the region, the

20:29

side of the brain, and even the gender.

20:32

So when you look, for example, at the

20:36

NAA here, NAA is lowest

20:40

in the first year of life, and it follows the myelination pattern,

20:44

of course, central to peripheral, inferior to superior, and posterior

20:48

to anterior. However, myo-inositol is

20:51

highest in the first year of life because it

20:55

represents the growth of the brain, the glial proliferation needed by the

20:59

brain at this time, and then it reduces thereafter.

21:02

So there are some variations. Creatine, for example, is higher in the gray

21:06

matter than in the white matter, and it's highest in the

21:09

cerebellum and lowest in the pons.

21:12

So you need to consider these variations before reading or interpreting

21:16

your MRS scale.

21:18

Well, let's move to the third part of my talk today.

21:22

Do you know when to use it under appropriate clinical context

21:26

or clinical scenarios? And let me share with you a couple of clinical scenarios

21:30

in the following slides. The most basic question while

21:34

using MRS: Is this a tumor or not?

21:38

Is this a glioma or not?

21:40

So basically, when we use MRS to identify a tumor,

21:45

the most important metabolite we are looking at is

21:48

choline, and I think everyone knows that.

21:51

So choline is your most important peak here to

21:55

identify a tumor, and you can use ratios between choline and

21:59

NAA or choline and creatine. Both are useful

22:03

to make confidence regarding this diagnosis.

22:07

And there are many ratios, as you can see here, suggested ratios.

22:11

And you know that using ratios is always, you should use it

22:15

with extra care because it's always a combination of sensitivity

22:19

and specificity. So whenever you have a higher ratio

22:23

here, you are more specific, but you are less

22:26

sensitive and

22:28

vice versa.

22:30

NAA also gets reduced because it represents

22:34

neuronal destruction or neuronal loss, which happens with tumor.

22:38

Myo-inositol is one of the metabolites or the peaks that is very

22:42

important to help you make a diagnosis of

22:46

a low-grade glioma, and usually in high-grade tumors, it

22:50

gets reduced. And also there are useful ratios.

22:54

One of the suggested ratio is using a myo-inositol over NAA.

22:58

If it is higher than 0.6 or 0.7, you are more

23:02

confident saying that this is a low-grade glioma, not a

23:06

high-grade one. And finally, lactate is very

23:10

important because it represents necrosis or

23:13

hypoxia, which is one of the hallmarks of

23:16

high-grade tumors, high-grade gliomas, as you can see here.

23:21

Well, this is an example of a pediatric low-grade glioma in a

23:25

patient with epilepsy. And you can see here how the spectrum

23:29

tells you first that choline is high

23:33

here, and NAA is low, so you are probably dealing with a

23:37

neoplastic tumor. But when you look at the short TE

23:41

here, this is a short TE and this one is intermediate TE.

23:44

When you look at short TE, you will look at the myo-inositol and

23:48

find it very high here. So you have now high choline, low

23:52

NAA, and no lactate because using intermediate TE will

23:56

help you identify lactate that should be inverted here.

23:59

So you don't have signs of hypoxia ischemia.

24:01

You have high choline, so it's a sign of cellular proliferation, and

24:05

you have myo-inositol, which is a sign of gliosis or glial

24:09

proliferation, characteristic of low-grade glioma, supported

24:13

by the perfusion map here that shows reduced

24:17

rCBV.What about grade three? It's not easy to

24:20

say grade three just relying on spectroscopy.

24:24

But

24:25

from the pathology perspective,

24:27

you also need to understand that whenever you get

24:31

higher grade

24:33

of the glioma, you will end up having more lactate here because there is

24:37

now evidence, there's an element of ischemia or hypoxia in the tumor.

24:41

The tumor keeps growing, and there's not enough blood supply

24:45

to support this growth. So there should be some sort of an anaerobic metabolism,

24:49

and there should be

24:51

some evidence of ischemia here or tissue hypoxia.

24:54

So we'll see this lactate here at 1.3 PPM.

24:58

Again, you have high choline, you have even lower NAA than the previous

25:01

case. So this is a tumor, and probably the ratio is high,

25:05

so you'll have some clue that this tumor could be a

25:09

high-grade tumor. And you still have myo-inositol.

25:13

But myo-inositol is not very high here, so this is probably

25:17

not a low-grade tumor. So you will use this spectrum to

25:21

support other imaging findings. Remember this very well.

25:24

We do MRS to support other techniques, not to replace other

25:28

techniques. And this is a case of diffuse astrocytoma,

25:32

anaplastic or WHO grade three.

25:35

The grade four astrocytomas or GBM in a case of

25:39

IDH wild type glioma here will have the

25:42

characteristic features of a high-grade aggressive tumor.

25:46

Much elevation of choline, very low NAA or

25:51

destroyed because of the destruction of neurons.

25:53

Look at this voxel here, just in the core of the lesion.

25:56

And you have lactate here that is inverted and has doubled, very

26:00

characteristic at 1.3. And then if you move your

26:04

voxel in this multi-voxel spectroscopy technique, if you move it

26:08

around the lesion or adjacent to the lesion in this area of abnormal signal,

26:12

but not the tumor itself, you will end up having also a similar

26:16

pattern, although less aggressive than this one.

26:20

So this voxel still shows high choline and low NAA, which mean

26:24

that the peri-tumoral zone also is

26:27

pathologic. So you have the positive peri-tumoral zone, which

26:31

differentiates primary high-grade gliomas

26:35

from metastatic lesions because the peri-tumoral zones in

26:39

metastatic lesions

26:41

are always negative.

26:43

So the characteristic signature of GBM or high-grade gliomas

26:47

is the following: elevated choline NAA, elevated

26:51

choline creatine, non-elevation of myo-inositol, but be careful,

26:55

this must be read in short TE. So these

26:59

spectra are actually intermediate.

27:01

Look at the TE, it is 135.

27:04

And here also is 135. So to look for myo-inositol, you should

27:08

look at short TE. And then you have inverted lactate, as you can see

27:13

here, and then you have positive peri-tumoral zone.

27:16

These are characteristic of a GBM or a high-grade

27:20

glioma. And the reason behind the positivity of the

27:23

peri-tumoral zone in these tumors or lesions is the

27:27

fact that in a case of high-grade tumor,

27:30

usually cells around the lesion, they migrate

27:34

from the necrotic core beyond the vascular plate, trying to

27:38

search for blood supply. So these cells

27:42

want more food, want more nutrients.

27:45

So they get out of this border of this zone, and they fill

27:49

up the space here. So although you sometimes see a negative

27:52

imaging appearance of the area around the lesion, it

27:56

is positive histologically, which means that you

28:00

still have malignant cells here, and this is why the peri-tumoral zone is positive

28:04

in high-grade gliomas. Back to the ratios.

28:07

So there are suggested useful ratios in the literature, and you'll find

28:11

many of that. But again, use it at your own risk,

28:16

because

28:17

you will end up having a combination of sensitivity and specificity.

28:21

And I would say that with practice, you will build up your

28:24

own ratios. There is no consensus, full

28:27

consensus on using a single value or a single

28:31

quantitative measure of

28:35

spectroscopy in grading tumors.

28:38

But I think the concept is still clear.

28:41

Where else can we use spectroscopy in the context of a brain tumor?

28:45

To guide biopsy. This is extremely important.

28:48

We usually follow the highest choline voxel.

28:51

We use multi-voxel. We ask our neurosurgeons

28:55

to be with us, and we actually

28:58

do some sort of

29:01

tracing these voxels together, and we guide them, such as in

29:04

this case. They can see the

29:07

most aggressive part or the worst part in terms of elevation of

29:10

choline, the highest choline here, and sometimes the

29:14

presence of lactate as well because it represents necrosis.

29:18

In this case, it's very straightforward because this part of the lesion is

29:22

strongly enhancing. But when you compare it to this case, in

29:26

this case,

29:27

there's a non-enhancing high-grade lesion here.

29:30

So this is a WHO IV IDH mutant astrocytoma, which as you can

29:34

see here is--

29:37

You can barely see some enhancement, minimal, if any,

29:40

enhancement. But when you use spectroscopy, you can identify

29:44

the voxel with the highest grade, because this is important.

29:48

These tumors are

29:50

frequently not simply grade four.

29:53

You'll have a mixture of tissue inside.

29:56

So you'll have grade three, grade four, and maybe grade two as well.

29:59

So you don't want to end up having under-representation of the

30:03

tumor or

30:05

error in the sampling of the tumor giving it a lower

30:09

grade than it truly

30:12

is.So when you look at this voxel, you will see the high choline here start

30:16

at two, and NAA is low here, move to the side.

30:20

There are two peaks here. One of them is choline, is very high, and move to the

30:23

other side, try to find the other peak that has two tops,

30:27

lactate, but it is inverted because this is intermediate

30:31

TE 144. And this is the worst or the most

30:34

aggressive voxel here. And you can correlate it with the ADC that

30:38

has restricted diffusion at this location, and you can even correlate it with

30:42

the rCBV map on this perfusion study that shows you elevation of rCBV

30:46

at the same corresponding location.

30:50

Well, let's look at this case together.

30:52

We have here, let's start at two, very low NAA, move to the

30:56

side, we have two other peaks. One of them is choline, which is pretty

31:00

high. And then move to the other side. Yes, I found it.

31:04

The two doublet here of this peak that is inverted.

31:08

So it should be lactate. High choline, something in the region of

31:11

lactate and very low NAA. It should be a high aggressive, high-grade

31:15

tumor. But actually, it is not,

31:19

because this is actually a meningioma.

31:22

And the reason we see this here, because this is not lactate, this

31:26

is alanine. And alanine,

31:29

it always inverts with meningiomas.

31:32

It always inverts at

31:33

1.448, near 1.5

31:37

ppm,

31:38

with intermediate TE.

31:41

So it is not

31:43

lactate. It's a characteristic feature of

31:47

meningioma. And the reason why you don't see any NAA, because

31:51

the meningioma is big, is huge. So your voxel is not contaminated by any

31:55

brain tissue, and you don't expect to see any neuronal tissue

31:59

in a meningioma, so NAA is not found here.

32:02

But of course, you have choline because in meningioma, you have cellular

32:05

proliferation. But you see this tumor is grade one tumor.

32:09

It's a benign tumor compared to the thought that we had

32:12

before. So the message here or the lesson here is that

32:15

always look carefully at the metabolite peak and make sure

32:19

that it resonates at the expected ppm value.

32:23

If this is lactate, you should find it here at

32:26

1.3, but it's not. It's not lactate because it

32:30

doesn't resonate exactly at the location you expect.

32:34

And if you don't know what is that, just Google it.

32:36

When you Google it, you will end up having a good idea about

32:40

the location of every peak, especially new peaks that you see in your

32:45

spectrum. Another example of this spectra that you have

32:49

a very high choline here. Again, low NAA, you have

32:52

lipid lactate. This is not intermediate TE, so probably this is a

32:56

short TE. And then you have something here.

33:00

This is a new peak. I would expect some myo-inositol here, which is

33:04

remarkably reduced. That's fine. But what is this?

33:07

I don't expect to see this here. And in the setting, of course, you will

33:11

not see this spectra. Nobody will give you the spectra,

33:15

especially in exams, and ask you what is this or what is that lesion.

33:19

You should also be able to see the other imaging techniques.

33:22

And this is what I said, MRS will support

33:25

the diagnosis or will support your thoughts in making

33:29

your differential diagnosis. But when you look at the lesion itself,

33:33

now you know that this is a medulloblastoma. So what is that peak?

33:37

If you look it up, you will know that

33:40

taurine peaks here. Taurine resonates here, and

33:43

taurine is a characteristic marker of medulloblastoma,

33:48

not all molecular subtypes, but actually

33:52

three and four. So it resonates at 3.4,

33:56

and it's found also in molecular subgroups three and

34:00

four, interestingly.

34:02

Well, so what about radiation necrosis?

34:05

So this is actually something that we see every day.

34:08

So the bottom line is lactate will dominate, but with

34:12

also lipids. So lipid lactate will dominate in cases of

34:16

radiation necrosis. You see this patient, this is before treatment,

34:19

and these three images are after treatment.

34:22

And you have the perfusion study here that tells you that

34:25

possibly there's no recurrence, but you have

34:29

ugly-looking enhancement here, suspicious.

34:31

Also in the FLAIR, you have a normal FLAIR signal. But look at the spectroscopy.

34:35

You see here the very high

34:39

lipid lactate elevation here, while all other

34:43

peaks are reduced, are damped down.

34:46

While in the lesion margin, not in the core of the lesion, but even in the

34:50

lesion margin, you still see this peak of lipid lactate,

34:54

and you see some elevation of choline as well, but not

34:57

as

34:59

much as expected in a true recurrence.

35:02

In a true recurrence, the choline-- because this is a

35:06

high-grade lesion, this is an aggressive tumor.

35:08

If you have a true recurrence, you would never end up having this

35:12

slight elevation of choline here.

35:14

So you would expect to see some elevation in choline because of the

35:17

inflammatory response due to radiation-induced or

35:21

treatment-induced changes. So we'd expect to see this, but

35:25

again, at the lesion margin, in the core of the lesion, everything will

35:29

be down except lipid and lactate, which will

35:32

dominate. What about true tumor recurrence?

35:37

High choline must be seen. This is straightforward.

35:41

A case of radiation, radiosurgery after gamma knife treatment,

35:45

and the patient came back with this enhancing lesion surrounded by edema,

35:49

and you see this very high elevation of choline, remarkably

35:53

reduced NAA, and you have a large peak of

35:57

lactate here. Why it's not inverted?

36:00

This is not short TE. Look at the TE here. It's 270.

36:04

So this one is long TE. So you'd never expect lactate to invert

36:08

here.

36:09

Well,

36:11

now I'd like to show you the next slide here,

36:16

and in this slideThis case is,

36:20

I would say that you frequently see this scenario.

36:24

I think so. So a patient who comes with this enhancing lesion,

36:28

suspicious.

36:30

The lesion is large, however, there's no mass effect

36:34

that matches the size of the lesion.

36:37

And the lesion actually follows or the abnormal enhancement follows a vascular

36:41

territory, where you have also diffusion restriction along the

36:44

vascular territory of the right PCA.

36:48

And then you have very high lactate.

36:50

So when you did spectroscopy in the beginning, they thought that there was some

36:53

technical problem. The spectrum is

36:56

unreadable, uninterpretable, and you can't

37:00

get

37:01

any interpretation or any conclusion.

37:04

But in reality, look carefully at this lesion.

37:08

You have some sort of lactate here.

37:11

And there's one piece of information that is very critical in this case, I'd like

37:14

to add.

37:15

On 3 Tesla,

37:18

lactate should invert on intermediate E, but it's not

37:22

consistent as in 1.5 Tesla. In 1.5 Tesla,

37:26

lactate will always invert on intermediate E, but in 3 Tesla,

37:30

it's not always consistent. Sometimes it doesn't, sometimes it

37:34

does, and sometimes one doublet, one peak inverts,

37:38

but while the other one does not. So this is a case of infarction.

37:42

Yes, true, you have high lactate. Why?

37:45

Because you have anaerobic metabolism.

37:46

You have severe tissue ischemia and hypoxia, and you end up having

37:51

marked elevation of lactate while the other metabolites or the other peaks

37:55

are reduced, at least because of the tissue

37:58

destruction, but also because they are relatively

38:02

visible compared to the high peak of the lactate here.

38:06

So this is a case of infarction. So do we

38:10

frequently or do we usually do MRS in cases of infarction? The answer is no.

38:14

But sometimes it happens, especially when in subacute infarction, when you get a

38:17

case like this one with strong enhancement, like everyone is suspicious, and

38:21

you need to avoid biopsy at least of these lesions, because biopsy is of

38:25

course unnecessary, and it can

38:28

carry diverse adverse

38:31

effects.

38:33

Well, in the case of brain abscess spectroscopy is usually useful

38:37

together with perfusion because in perfusion you will have reduced

38:41

rCBV, and this will help you differentiate these ring-enhancing lesions

38:45

usually compared to tumors, but also in spectroscopy,

38:49

spectra will make you confident, will add

38:52

a lot to the value of the study

38:56

or to the confidence of the diagnosis. What do you usually see?

39:01

Lactate in these cases, again, because of the necrosis.

39:04

But we have here suppurative necrosis. It's not liquefactive.

39:08

It's not as the type of necrosis that we usually see in tumors,

39:12

but in abscesses, we see liquefactive necrosis or

39:16

suppurative necrosis. So you will have lactate at the expected location,

39:20

1.3 PPM, and you will have many amino acids.

39:23

And always remember this, A abscesses

39:26

is A amino acids. You will have alanine, you have acetate, you have

39:30

aspartate, and you don't need to remember all these resonance

39:33

frequencies. You can just look them up at Google when you see a suspicious

39:37

case of abscess. You will have succinate, which is very useful

39:41

at 2.4 PPM in cases of suppurative infections and sometimes

39:46

you will have other metabolites or other peaks as well.

39:50

What about tumefactive them? I work in a place-- We have

39:54

a national referral center of MS cases

39:58

in Kuwait, and we don't do MRS. I can't say that we

40:02

should do MRS in every single case of MS, but what happens is

40:06

sometimes you receive a patient like this with a single lesion,

40:10

single ring-enhancing lesion. I understand that this is an open ring of

40:13

enhancement, and this should be a clue, but again, the patient doesn't have

40:17

any history of demyelination, and this is the first presentation, and the

40:21

lesion is surrounded by abnormal signal on FLAIR.

40:24

So this might add some value to the diagnosis.

40:28

Itself alone, it cannot make the diagnosis,

40:32

but in conjunction with other modalities, of course, it's going to be very

40:36

useful. What do you expect to see in cases of MS?

40:39

Usually nonspecific features,

40:43

like macromolecules, elevation of macromolecules here in zone three,

40:47

you have

40:48

elevation of glutamate here in zone two, and you

40:52

can see this peak here, and you can also have some lactate which inverts here

40:56

on intermediate TE because of this tumefactive

41:00

lesion. And of course, there will be some sort of elevation of choline.

41:03

Again, these peaks are

41:07

nonspecific, but in conjunction with other techniques should be useful.

41:12

Metabolic CNS disorder. This is very useful and very

41:15

relevant application of MRS.

41:19

So in neurodegenerative diseases, the rule is that you have

41:23

lower NAA.

41:25

In having Canavan disease, of course, this is the

41:28

only CNS disorder that will show you

41:32

remarkable elevation of NAA.In mitochondrial disorders,

41:36

always look at lactate because it's going to be elevated because of the

41:39

anaerobic metabolism, and hepatic encephalopathy, and so on.

41:43

And always try to find the new peak.

41:46

And

41:47

to be honest, these cases are usually referred with a provisional diagnosis.

41:50

So you have some sort of a thought in your mind

41:54

before performing MRS. And this is very important because sometimes you

41:58

need to manipulate the technique in order to be able

42:02

to pick the new peak or to detect the new peak

42:06

that is characteristic of the lesions. One example is phenylketonuria.

42:10

And when you look carefully at phenylalanine, it resonates at seven point three six

42:14

ppm. And this is extremely important because in our clinical

42:18

practice, we usually scale or we usually look at ppm

42:22

starting from four or slightly higher than four, but we never

42:26

go to seven point three six. And this is not

42:30

something that you need to do before the scan.

42:32

You can change the settings. You can ask your tech to

42:36

expand the bandwidth to cover

42:39

this part of the scale so you can pick the

42:43

phenylalanine, which resonates at seven point three six ppm, and it's going to be

42:47

very characteristic, and it will help you diagnose

42:51

these cases. So in phenylketonuria, remember to expand the display.

42:55

The other piece of advice that I'd like to

42:59

give here is the use of intermediate TE, because if you use short

43:02

TE, you'll have much disruption of the baseline.

43:06

And sometimes phenylalanine is very short, not like this case.

43:09

In mild cases of phenylketonuria, you can have that

43:13

here with a lower amplitude, so it's going to be mixed with

43:17

the noise or with the disruption of the baseline.

43:19

So always try to use intermediate TE because this is the

43:23

best TE that will give you the highest signal

43:27

compared to the highest SNR ratio and will mask

43:31

the noise here in the baseline. Well, so in succinate, in case of succinate

43:35

dehydrogenase deficiency or complex

43:39

two deficiency,

43:40

other than finding elevation of lactate, succinate also will

43:44

be a useful marker here. So what about glycine in

43:48

cases of nonketotic hyperglycinemia?

43:51

This is another piece of advice and another technical manipulation that you

43:55

need to do because in cases of hyperglycinemia,

43:58

it's

44:00

very important and critical to use intermediate TE.

44:03

And the reason is, look carefully at the ppm value here,

44:07

where glycine resonates. It resonates at three point five five.

44:10

For those who do MRS, what else resonates at this

44:14

location? Very close to this location. It's myo-inositol.

44:18

Myo-inositol resonates as a multiplet from three point five to three

44:22

point six. Let me say three point five four

44:26

or three point five six. So it's very close to

44:30

glycine. So using intermediate TE, and I told you in the beginning that

44:33

myo-inositol can invert on intermediate TE, but the

44:37

clinical intermediate TE that we do at one hundred one three five

44:41

or one hundred

44:43

forty-four milliseconds, usually myo-inositol gets

44:47

reduced, remarkably reduced. So it will make your glycine

44:51

more visible because glycine will not be affected, but myo-inositol will be

44:55

reduced. So it's a strategy to

44:58

make some sort of nulling of the myo-inositol in

45:02

favor of the visibility of the glycine in these cases.

45:04

So imagine when you have these cases, your tech must talk to you, and

45:08

you can have very important inputs to

45:12

the technical optimization of the study.

45:16

In cases of maple syrup urine disease also, it's very useful because you

45:19

basically look at new peaks, beta-chain amino acids, beta-chain ketoacids.

45:24

Again, this is very helpful to make the diagnosis.

45:27

So the role of MRS in metabolic and MS disorders can be

45:30

supporting diagnosis, of course, with other imaging findings,

45:35

to do the serial follow-up in these patients, especially those who are under

45:38

dietary restrictive treatment. So you need to make sure that things are getting

45:41

better on the follow-up, and you can also screen siblings without doing

45:45

any bad stuff or any extensive

45:49

investigational workup, such as in patients with

45:53

X-linked adrenoleukodystrophy.

45:56

And again, remember that intermediate TE is your real friend,

46:00

not only because of glycine that can be more visible, but also look

46:04

at this. In cases of phenylketonuria, intermediate TE will be very

46:08

useful because it will remove the baseline distortions.

46:11

And look at the neurodegenerative disorders.

46:14

The hallmark here is the loss of the

46:17

NAA. So loss of NAA in neurodegenerative.

46:20

So again, intermediate TE gives you the best SNR and gives you the

46:24

best visibility of choline and NAA.

46:26

So the bottom line is in any case of metabolic and MS disorder,

46:30

make sure that you have intermediate TE in your

46:35

wallet.

46:37

These are two examples from my colleagues in Kuwait.

46:40

The one on the right side here is a complex one deficiency, one type

46:44

of mitochondrial respiratory chain disorder.

46:46

And again, look at the inverted lactate at

46:50

intermediate TE. While the case on the right side here is a case of maple

46:54

syrup urine disease. And once you see new peaks at

46:57

unexpected location here, look them up.

47:01

Look it up in Google, and it will

47:05

support your thought. So you have a beta-chain amino acid, a beta-chain keto acid,

47:09

and this is one of the

47:10

very useful applications of MRS in metabolic and MS

47:14

disorders. What about future directions?

47:17

Very briefly, we've started in many clinical sites

47:21

nowimplementing the edited MRS or the

47:25

spectral editing technique of MRS.

47:27

And the most popular one is the mega press spectroscopy, where we

47:31

use spectral selective pulses, and it

47:35

requires some sort of higher gradient performance.

47:39

So it cannot be applied on any MR system.

47:41

Your system must be a 3 Tesla, and there should be some sort of optimization

47:45

in terms of hardware and also software by which you can

47:50

edit the pulse based on the J-coupling differences between metabolites.

47:54

And you will end up having new metabolites, as in this example of

47:57

2-hydroxyglutarate, which is very important because it's a marker of

48:01

IDH mutation in astrocytomas. So once you

48:05

see it here, at the expected location,

48:09

you will get more confidence and the accuracy is actually

48:13

very high. It reaches up to 90% or even more.

48:17

And there are other examples here, such as glycine, that can give you

48:21

an idea about the

48:24

outcome in cases of gliomas, especially the survival, which

48:28

is lower in cases where glycine is elevated regardless of the

48:32

IDH mutational status. There are many reports

48:36

about cystathionine, which is a marker that

48:40

can predict 1p/19q code deletion.

48:43

And here, of course, I'm talking about oligodendroglioma.

48:46

Again, this is critically important to differentiate oligo

48:50

from astro even before biopsy. And then GABA

48:54

in psychiatry, glutathione in psychiatry, and also Alzheimer's disease and so

48:58

on. So it's promising, it's evolving, and we'd expect to

49:02

see more applications of edited MRS in the very

49:05

near future. What about multinuclear MRS?

49:10

Yes, it's the hydrogen or the 1H MRS that

49:14

we basically use in our

49:16

practice and in

49:18

many of the clinical sites and even researchers.

49:20

But you can imagine that there are many other

49:24

types or nuclei that we can use doing

49:28

MRS depending on

49:32

many factors. One popular one is actually the phosphorus

49:36

spectroscopy. And here's an example of phosphorus spectroscopy that

49:40

tells you new information about the phosphate in the brain

49:44

here. So this is the ATP, adenosine triphosphate, with the three

49:47

components of it. This is the phosphocreatine, which is the

49:52

battery of your energy in the brain, and this is the

49:56

inorganic phosphate, which we call the leftover

49:59

energy after consuming the energy of the phosphocreatine and the ATP.

50:03

So these are very important, and there are

50:07

variations in these values in health and disease.

50:10

And you can, for example, estimate the pH of the brain, which

50:14

changes with brain tumors, because with more aggressive tumors, the

50:18

brain usually gets more alkaline, and the pH will be

50:22

higher by just measuring the

50:26

frequency shift difference between the inorganic phosphate and phosphocreatine and

50:30

so on. You can have a magnesium

50:33

map by measuring the difference between the alpha and the beta

50:36

components of ATP and so on. So these are very useful and again,

50:40

promising techniques in the future.

50:42

Artificial intelligence is being also

50:46

implemented in MRS, such as like many other

50:50

modalities and like many other applications.

50:52

And it's useful not only in the

50:56

interpretation of the results, but starting from the data

50:59

acquisition, the processing and the analysis and the denoising

51:03

of the spectra and the reconstruction, and

51:07

then the peak picking and interpretation of results.

51:10

So many domains where AI can be used

51:14

in the field of MRS. This is an example of a classifier

51:18

system based on machine learning that we used in the past

51:23

to help us differentiate high-grade from

51:26

low-grade from non-neoplastic lesions.

51:30

And you can see here the ROC curve analysis and see the area under the curve with

51:33

the orange curve here that represents

51:39

reading these images by the radiologist assisted by the tool or the

51:43

software, compared to reading it

51:46

without using the tool, only by the radiologist.

51:49

And you can see here again the ability to diagnose low-grade tumors, for

51:53

example, and differentiate them from other pathology.

51:57

And this is the performance here. You can see the area under the curve is

52:01

higher here when we use the software or when we

52:04

get assisted by the software compared to the performance of radiologists

52:08

without using this tool. Well, so this brings

52:12

us to the

52:14

last slide here. This is my takeaway from this talk.

52:19

That MRS should remain a valuable and even

52:23

likable tool. And as I said, not to replace other imaging

52:27

modalities or techniques. You should use it to support your thoughts and your

52:31

differential diagnosis in conjunction with all other imaging

52:34

techniques. And this is possible only if three key factors are met:

52:38

ensuring technical quality, mastering the interpretation of the

52:42

peaks, reading the metabolic peaks very well, and then to apply

52:46

it within the right clinical indication and

52:49

under the appropriate clinical context.

52:52

So

52:54

MRS can work well with you when you get it right, when you

52:58

read it right, and when you use it right.

53:01

This is my wonderful new radiology team in Adessina

53:05

Hospital in Kuwait, and thank you very much.

53:10

All right.

53:12

Thank you for sharing all that with us, Dr. Gad.

53:14

At this time, we'll be opening the floor for any questions from our audience.

53:18

You can submit your questions to the Q&A

53:19

feature.They

53:25

did hide that recently, Dr. Gad. You're going to have to go to your

53:29

more menu.

53:31

Yeah. I can see the Q&A section, but I can also see in the

53:35

chat box other comments as well. I'm not sure which.

53:38

I think I should start with the Q&A section, right?

53:42

Go for it.

53:44

Yeah. Well, so any experience of MRS in neurodegenerative

53:47

conditions?

53:50

Yes. Again, it can be used in

53:54

conjunction with-- So if you are talking about children in

53:57

pediatric neurodegenerative disease, yes.

54:00

So yeah, we use it, and the hallmark is always

54:03

the reduction of NAA or

54:07

NA ratios,

54:10

or choline, for example.

54:12

And also

54:14

glutathione, one of the emerging

54:18

metabolites that can be used, especially with edited MRS, in cases of

54:22

Alzheimer's disease, for example.

54:24

But in my experience, it doesn't have

54:28

a major role in making the diagnosis.

54:32

Again, it

54:34

adds to

54:36

your confidence, or it helps you support

54:40

the diagnosis in some conditions.

54:44

What conditions can give both low choline and myo-inositol

54:48

peaks, and is vascular disease one of them?

54:52

Yeah, this is a good question. So any case where you would

54:56

expect some

54:58

gliosis, reactive glial tissue.

55:02

Because glial tissue is the fastest tissue to react.

55:06

Of course, neuronal tissue does not change

55:09

significantly, but glial tissue does.

55:12

So you would expect this

55:17

to happen

55:20

maybe in cases of encephalitis,

55:23

especially if it is not acute encephalitis,

55:27

if it is chronic encephalitis. Because in acute encephalitis, I would expect also

55:30

some sort of elevation of choline because of cellular proliferation.

55:34

And again, in cases of

55:37

MS, as the example I presented,

55:40

this one with tumefactive MS,

55:43

you should see some myo-inositol elevation, while the choline doesn't have to

55:47

be elevated that much. But let me be clear with you about

55:51

that. This is going to be a trap, when you have a low

55:55

choline and

55:57

elevated myo-inositol peak. So I think this is what you mean.

56:01

So low choline and elevation of myo-inositol.

56:05

But the value of looking at myo-inositol is actually when you

56:09

have a high choline. So let me make it clear.

56:12

So look at myo-inositol when you see a high choline, because it will

56:16

help you differentiate high-grade lesion from a low-grade lesion, and this is the

56:19

most relevant

56:21

approach or application of

56:25

myo-inositol MRS when it comes to myo-inositol.

56:29

So could you kindly repeat the-- Yes.

56:32

So integral amplitude. So the basic thing is that some

56:35

systems, because we always look at the spectra, but the

56:39

number that is written next to each peak of the

56:43

spectra represents the

56:47

amplitude or the integral. If it has an i, if it has the letter i

56:51

next to it, it means that this number represents the integral, which

56:55

is the area under the curve, and in my opinion, and based on my experience,

56:59

this is much better, especially if your bandwidth is not

57:04

very sharp. I mean, the peaks are not very sharp because of

57:07

shimming problems, because of anything.

57:09

Once you have broadening of the spectra, never rely on the

57:13

amplitude. Believe me, always rely on the integral,

57:17

and this is something that you can control in the setting.

57:20

But if you have a narrow bandwidth and you have a sharp peak, then

57:24

I assume that the amplitude should equal the integral.

57:27

I mean, the amplitude should reflect

57:31

the integral. If this one is high, the other one should be high as

57:35

well. How can we explain spectral degradation in the peripheral

57:39

voxels of CSI?

57:44

Yes, I would say that this might happen because of the

57:49

outer volume suppression, because whenever you have outer volume suppression, you

57:53

saturate this band and you saturate also the voxel

57:57

just adjacent to this band, and this is usually what

58:00

happens.

58:02

So

58:04

this is why you get some sort of degradation in the periphery of the voxels of the

58:07

CSI, and this is why your grid should cover the lesion and should

58:11

cover the area around the lesion, the peri-lesional area, and should

58:15

cover some backup voxels. Let me

58:19

call them backup voxels or spare voxels.

58:22

You can use them,

58:25

or you can just ignore them, because you should expect some sort of

58:29

degradation at this part. The other thing that in practice,

58:33

most of the time, the lesion is when you put a grid on the lesion and the lesion

58:37

is quite sizable, most likely you will end up

58:40

abutting the surface of the brain or an

58:44

area that you want to avoid naturally,

58:48

either the skull,

58:50

either the maxillary sinus, the skull base, or even

58:54

outside the skull itself. So you'll end up having some air.

58:57

So it's either air, bone, or maybe CSF, nobody

59:00

knows, and this might cause some distortion and degradation of

59:04

the spectrum. What else do we have here?

59:08

Oh, thank you.

59:10

Where are we with AI reading spectra?

59:12

So to be honest, so I have some experience with AI, but not

59:15

exactly in MRS. But what I can say, yeah, I'm not sure when did

59:19

you exactly post this question, but I think my slide on AI

59:24

should have answered this question

59:27

briefly.

59:29

Thank you. Due to time limitations, one is only able to use one or two

59:33

single voxelThe MRI sequence, which would you pick?

59:36

Short, intermediate? Yeah, this is a very good question.

59:39

Well, so

59:42

the best approach is to do short and intermediate.

59:46

So I always say that if you don't know the lesion, if this is your

59:49

first encounter with the patient, you don't know anything about the lesion, you

59:53

have to do both.

59:55

Okay?

59:57

But if you know that this is a tumor case and you follow up this

60:00

tumor, either for possibility of recurrence, possibility of

60:04

radiation necrosis, treatment-induced change, anything like that,

60:07

then let's focus on intermediate. Why to do short TE?

60:11

You're not getting any new information. Just do intermediate.

60:14

Every time the patient comes for follow-up, do intermediate.

60:17

But if this is the first time... So

60:20

the other answer to your question, which is very practical, if you're really short

60:24

of time, if you don't have

60:27

time for it, to do another six-minute or five-minute

60:31

MRI scan, then it's fine. Let's do short, and if you don't

60:35

see anything near the lactate region, near zone three,

60:40

then you can just rely on the other metabolites on short TE.

60:44

Or if you get a clean spectrum as well, because many times

60:49

under many situations, we do the short, and we get some sort of

60:52

distortion of the baseline. So well, so from your experience,

60:56

are there suggested anatomical locations in the evaluation of

61:00

inborn errors of metabolism? It depends.

61:03

So for example, you would, in a case of leukoencephalopathy, just

61:07

go for the abnormal signal.

61:09

In a complex

61:11

one deficiency,

61:16

you should see the lesions involving the basal ganglia and the brain stem.

61:21

Same thing, in a case of Ley disease.

61:23

So the answer, it

61:27

depends on the expected pathology or

61:31

if you have an abnormal signal, then it's going to be very useful to just

61:35

follow the abnormal signal. Ways to show

61:39

2HG,

61:41

yes, doing mega press spectroscopy,

61:45

but you have to

61:47

have a software, a toolbox to

61:51

do the off resonance and on resonance, and then to do the

61:54

difference spectrum, because there is a way to read that, and this is

61:58

not available on the vendors and the vendors'

62:02

workstations. So if you have a vendor workstation,

62:06

either the Single via by Siemens or the Spectra

62:10

View by Philips or the

62:12

one by GE, I can't remember that exactly the name.

62:15

So you would not have this

62:18

tool or this option. So the only way to do it is to use a

62:22

third-party software. There are many free third-party software,

62:26

and there are also paid or subscription

62:30

ones as well to document. But again, you cannot do it

62:35

just on your standard scanner, standard three tesla scanner.

62:39

You have to optimize it. You have to, I think, to

62:42

purchase the

62:44

technique itself or the option, because

62:48

there has to be some sort of hardware configuration and optimization

62:52

before scanning mega press. To document MRS ratios, you

62:56

suggest using short or intermediate

63:00

voxel. I think you mean TE.

63:03

Yes, this is a good question again, but, well, so

63:07

if you are talking about the big ones, the major ones,

63:11

choline, NAA, creatine, I would suggest

63:15

doing intermediate TE, because in short TE, you should

63:18

expect some distortion of the baseline, which might affect

63:22

your metabolites, because when you have distortion, especially in three tesla,

63:26

if you get distortion in the baseline,

63:29

you will end up having spectral broadening, and

63:33

may overlap with the other peaks as well.

63:35

So even regardless of the field strength, to make this clear again, yes,

63:40

intermediate TE is the one that gives you the highest

63:44

SNR,

63:45

and the cleanest

63:48

peaks or amplitudes,

63:50

especially for the major ones, choline,

63:54

creatine, and NAA.

63:57

At Philips MRI, many ratios are present for choline NA which--

64:01

Yes, this is true, and this is what I mentioned.

64:04

It's always a combination of sensitivity, specificity, okay?

64:07

So what I would suggest, even when you look at the papers, I don't

64:11

find any consensus or agreement on using a specific ratio or a

64:15

cutoff point, if you mean that. So it

64:18

requires practice, and it's very useful when you do comparison

64:22

with the previous

64:24

studies. And again, this is another benefit of using intermediate TE

64:28

that it has been reported that intermediate TE is

64:32

the most reliable TE when you do comparison, when

64:36

you do follow-up, because you will end up having consistent quality

64:40

every time you do the MRS scan. So,

64:45

yeah, lymphoma. So L lymphoma always

64:49

reminds you with L,

64:51

lipid lactate. So in lymphoma, you would have also

64:55

more lipids. This is what we expect, and also you have

65:00

elevated choline, of course. So what about NAA?

65:03

If the lymphoma is infiltrating the tissue and the lesion is

65:07

really big, so you should expect your NAA to be lower,

65:12

because there's no neuronal element of tissue.

65:14

But most of the time, lymphoma infiltrates the tissue and does not

65:19

destroy the neuronal architecture of the brain, which

65:23

is even more confusing to you. So I would

65:27

definitely, yes, I can do spectroscopy in case of lymphoma.

65:30

At least I would have an idea that I'm dealing

65:34

with an aggressive lesion or a malignant lesion.

65:37

But yes, definitely, I would rely more on

65:40

diffusion because you would expect restricted diffusion and on perfusion

65:44

studies, DSC studies, especially when you look at the PSR or the

65:48

percentageThe

65:54

signal

65:56

recovery percentage, the percentage signal recovery, I remember the name.

66:00

PSR, which is the

66:02

behavior of the curve after the first shoot.

66:05

So does it recover, or does it overshoot?

66:07

That differentiates lymphoma from GBM and also from mets.

66:12

In the report, do we document only ratios or also integrals?

66:16

And so we never document integrals. I never do that.

66:19

And I never document amplitudes as well, because

66:22

when it comes to

66:25

quantification, so the best

66:28

strategy is to use ratios. And this would avoid

66:32

many of the

66:34

pitfalls, because if you have something that might

66:38

have affected the quality of the spectrum, probably if you have some sort of

66:42

fake elevation of choline, the NA could

66:46

have undergone the

66:50

same thing. So

66:52

always using ratios compensates for

66:56

most of these technical challenges, but

67:00

also be careful, again, because using just ratios and relying

67:04

on ratios to make the diagnosis is risky and is very

67:07

challenging. So these are the Q&As.

67:10

Oh, there are many others in the

67:15

chat, but not in the Q. I think we finished the Q&A.

67:20

Ben, what do you think?

67:22

Sounds like you did a great job.

67:28

So we still have a long list here under the chat,

67:32

but I can just pick two or

67:36

three if time permits. Is that okay?

67:37

Yeah. If you'd like to pick a few from the chat, go ahead.

67:40

Yeah. Perfect. So, yeah, there are some good comments.

67:44

Thank you for that.

67:45

Okay.

67:51

Advice for building an MRS program for institutions that do not

67:55

have a robust program.

67:59

Well,

68:00

yeah, I see that.

68:03

I would say that

68:05

building the team, so it's strategies, you have a clear vision first,

68:09

and you should also talk to all partners, all

68:14

stakeholders, we can call them. And

68:18

then training, because if you have a good team, and if you

68:21

have the

68:23

personnel,

68:26

I mean, the technologist and also the

68:30

radiologist, the dear radiologists who are well-trained, that will make a big

68:34

difference when it comes to the time of interpretation.

68:38

And again, one of the major challenges or limitations in

68:42

using spectroscopy, especially when it comes to the reading time or

68:45

exhausting your resources, is getting a poor quality

68:49

scan. Because once you have a poor quality scan, you will end up having

68:54

longer times for interpretation, and you'll struggle with

68:58

that. So I think training is also another factor that is very

69:01

important.

69:03

And finally, I would say,

69:06

also the integration of AI

69:09

into the workflow somehow, at least in the

69:14

acquisition, in the denoising part, and maybe in picking the

69:18

peaks and helping you minimize the time you...

69:22

This is for the time and for the reimbursement and the time constraint and so

69:26

on. But basically, you need more training

69:30

of the staff so that they can...

69:33

This is the key factor behind success of any

69:37

program, including MRS. Well, let's do one more thing

69:42

here.

69:45

So, yeah, I think we've answered many of these

69:49

questions before.

69:52

Yes.

69:54

So,

69:56

yeah. Lipids and lactate both always go hand in

70:01

hand and indicate high-grade features.

70:04

Not always true. Because

70:08

from the histology or from the histopathology

70:13

perspective,

70:15

so you see lactate because of the anaerobic

70:20

metabolism, okay? And this happens because of

70:23

when you have neangiogenesis in these tumors, but this blood supply

70:27

does not support the demand of the growing

70:31

or the expanding tissue. So cells are increasing or proliferating.

70:35

They keep proliferating every single hour, and then they

70:39

end up having less blood supply. And here

70:42

they go into a process of hypoxia, a

70:45

cascade of hypoxia, even before necrosis.

70:48

So you don't have to see necrotic features in the lesion by

70:53

macro imaging, to get a high elevated lactate.

70:56

And this is one of the very useful, valuable

71:00

parts of spectroscopy, that you detect tissue

71:03

hypoxia or necrosis before seeing gross

71:07

necrosis.

71:08

Lipids is

71:11

actually a marker of... So they come from cell membranes.

71:14

And you need to have cell membrane

71:17

destruction to see lipids. So it's a different

71:21

pathology. They are correlated with each other because you will end up having

71:25

both of them in high-grade tumors, but they can precede

71:30

each other. So you can start by necrosis, and once you have cell death,

71:34

the cell membrane will be destroyed, and you will see

71:37

more lipids. And this is an interesting area,

71:41

but I think when it comes to tumors and high-grade tumors, you

71:45

should rely on lactate. Why? Because

71:48

from the pathology perspective, pure pathology perspective, what are the three

71:53

cardinal features of grade four gliomas?

71:57

NeoangiogenesisNecrosis.

72:01

So one of them is neoangiogenesis.

72:02

Neoangiogenesis, necrosis, and hemorrhage. Right?

72:06

So for neoangiogenesis, this is one of the cardinal features, one of the

72:09

hallmarks. So once you get

72:13

neoangiogenesis,

72:15

this is a sign that your tissue

72:19

will evolve into hypoxia very soon.

72:23

Because necrosis, again, is a gross feature.

72:25

Even if you see it microscopically, it's still a gross feature.

72:29

So in between neoangiogenesis and, because those vessels are aberrant,

72:33

as you know, they are very weak, they are fragile, they disrupt very

72:37

easily. And once that happens, you begin the

72:41

cascade of hypoxia and necrosis. And this is why you need to

72:45

rely on lactate, because it translates what's happening in the

72:48

pathology. It correlates very well the grade for

72:53

nature of the

72:56

aggressive

72:58

tumor. So I would say

73:01

lactate is much more better, or much more

73:04

reliable,

73:06

than just using lipids in high-grade

73:10

tumors.

73:11

Yeah, I think we exceeded the time for maybe 15

73:15

minutes, right?

73:18

Hey, great job answering all those questions, though.

73:20

Yeah. Thank you very much.

73:23

Yeah, thank you for that great lecture today, and thanks to everyone for

73:27

participating in our Noon Conference and asking so many great questions.

73:31

You can access a recording of today's conference and all our previous Noon

73:34

Conferences by creating a free account.

73:36

We'll also be emailing out a link to the replay later today.

73:39

Be sure to join us next week on Thursday, April 2nd at 1:00 PM,

73:44

where Dr. Prashant Nagpal will deliver an lecture

73:47

entitled "Cardiac CT: A Journey from Origin to 2026."

73:52

You can register for that at medallity.com and follow us on social media for

73:55

updates on future Noon Conferences.

73:57

Thanks again, and have a great day.