Biomedical Imaging Research (Split off)

[durfen]

I'm a mathematics major, interested in dynamic systems modeling (if I met any of you in the interviews and you didn't make the connection yet, you probably know who I am now! ). I too am not so interested in a wet lab necessarily.

Arguments can be made that UofWashington is the absolute best place to go for mathematical biology. They have ~20 faculty members studying in this niche area, far more than the next best (probably UMichigan, with about 8). A few of these guys are bioinformatics.

I really don't think that UofW is a powerhouse in medical imaging, though I'm not certain. I know that both Yale and WashU are very strong in this area, though I'm sure there are more.

WashU has the best radiology department. Unfortunately they are not really associated with the MSTP or PhD programs, and I was discouraged from going that route when interviewing there.....

---

Members don't see this ad.

 

[Neuronix]

WashU has the best radiology department. Unfortunately they are not really associated with the MSTP or PhD programs, and I was discouraged from going that route when interviewing there.....

(Getting my info from: http://www.acadrad.org/showpage.aspx?page=nihrank05inclrd)

That's very sad actually. I wouldn't say WashU has the best Rads department in the country (MGH alone has higher NIH funding), but they're definately a top program. It would be sad however to discourage people from persuing an MD/PhD in Radiology (usually through Bioengineering or Biophysics) at such a solid basic science imaging place. You can certainly do that kind of work at Penn (I do!) and you won't be discouraged.

I bring all this up not because I'm trying to hijack the UWash thread (sorry!), but because I'm often asked about MSTPs in Biomedical Imaging and there's little info and few programs where you go to do such things.

That all being said, IMO if you're interested in doing imaging work, you want to get the basics of whatever modality you end up working in. That means finding a lab that's building hardware, working on pulse sequences, or otherwise doing cutting edge stuff with new contrast agents or such. Beware of labs where all you get to do is go over to a clinical magnet equipped with standard hardware and push the button on sequences that your vendor or collaborators provided you.

 

[durfen]

(Getting my info from: http://www.acadrad.org/showpage.aspx?page=nihrank05inclrd)

That's very sad actually. I wouldn't say WashU has the best Rads department in the country (MGH alone has higher NIH funding), but they're definately a top program. It would be sad however to discourage people from persuing an MD/PhD in Radiology (usually through Bioengineering or Biophysics) at such a solid basic science imaging place. You can certainly do that kind of work at Penn (I do!) and you won't be discouraged.

I bring all this up not because I'm trying to hijack the UWash thread (sorry!), but because I'm often asked about MSTPs in Biomedical Imaging and there's little info and few programs where you go to do such things.

That all being said, IMO if you're interested in doing imaging work, you want to get the basics of whatever modality you end up working in. That means finding a lab that's building hardware, working on pulse sequences, or otherwise doing cutting edge stuff with new contrast agents or such. Beware of labs where all you get to do is go over to a clinical magnet equipped with standard hardware and push the button on sequences that your vendor or collaborators provided you.

Yup, this is the kind of thing they have at UWash. I'm on the software side however (i.e getting more information from an already reconstructed image, making imaging more clinically useful etc.). I feel that UWash are willing to let me take a crack at that. Penn also.
WashU/UCSF did not, probably because of my software focus. Harvard was iffy, the medical image computing lab was in a dark corner of a dark floor of BWH.

Hey man, but it's Mallinckrodt!

 

[Neuronix]

Yup, this is the kind of thing they have at UWash. I'm on the software side however (i.e getting more information from an already reconstructed image, making imaging more clinically useful etc.).

I have nothing against the image processors, just make sure in your education you learn how to acquire some images yourself and learn the basics of the modalities you're working on. I've worked with a number of faculty and post-docs who have never used the imaging systems in their lives. To me it seems like you're shoeboxed at that point. You're handed data, you get to process it. You can't go and say "Try this or that parameter to make the images look better/more useful" because if you don't understand what the parameters mean you're stuck in one little corner.

If you come from a mathematics background and you know how to program it becomes almost trivial to reconstruct an image from raw data, whether that's projection reconstruction or k-space FT. Though, you do need someone at your site who knows who to get the raw data for you. I think you owe it to yourself in your education to learn and do things like this. Make sure you take courses and preferably acquire some images yourself so that if you work for example on MR you know what things like TR, TE, Bandwidth, etc etc etc etc do to your images.

I'd be happy to talk to you about the labs and courses if you do come here.

 

[durfen]

I have nothing against the image processors, just make sure in your education you learn how to acquire some images yourself and learn the basics of the modalities you're working on. I've worked with a number of faculty and post-docs who have never used the imaging systems in their lives. To me it seems like you're shoeboxed at that point. You're handed data, you get to process it. You can't go and say "Try this or that parameter to make the images look better/more useful" because if you don't understand what the parameters mean you're stuck in one little corner.

If you come from a mathematics background and you know how to program it becomes almost trivial to reconstruct an image from raw data, whether that's projection reconstruction or k-space FT. Though, you do need someone at your site who knows who to get the raw data for you. I think you owe it to yourself in your education to learn and do things like this. Make sure you take courses and preferably acquire some images yourself so that if you work for example on MR you know what things like TR, TE, Bandwidth, etc etc etc etc do to your images.

I'd be happy to talk to you about the labs and courses if you do come here.

Yeah of course you have to learn about the modality. Throwing algorithms at an image does not work a lot of the time and you need to know the basics of the modality to know why it doesn't work and how it could be made better. I have done protocol-optimization studies before in order to improve post-processing (took over a CT scanner for a day ). This is especially important in MR and the various molecular and optical imaging modalities. In these cases, I'm going to try and acquire the images myself.
'Come here' meaning Penn?

 

[Neuronix]

Yep, I pulled these posts out of the UWash thread.

It may seem obvious that you need to know these sorts of things, but the people I've met at the image processing labs at Penn really have little idea how to use the scanners or what the imaging parameters mean. I don't know if that's the way it is everywhere, however. It may just be the people I've met too.

 

[durfen]

Yep, I pulled these posts out of the UWash thread.

It may seem obvious that you need to know these sorts of things, but the people I've met at the image processing labs at Penn really have little idea how to use the scanners or what the imaging parameters mean. I don't know if that's the way it is everywhere, however. It may just be the people I've met too.

Good that you split it off .

To some extent, it is true. Straight CS guys tend to abstract the problem out, which in turn tends to abstract out the clinical aspects as well as the 'character' of the imaging modality. It also restricts the problem space, which is good to reduce cognitive load, but bad since you reduce the solution space as well.
I generally take a clinical approach to things, i.e solve a clinical problem, not an signal processing problem.

What kind of research do you do?

 

[Neuronix]

I'm not sure I want to get too specific because it would be easy to figure out who I am (was it that hard anyways?), but I work on MR detection of a rarely used contrast agent in vivo. As a part of this I work on pulse sequence, MR hardware, and image processing development and this is done on phantoms (solutions), autopsy specimens, pigs, and people. I dabble in some other projects as well, trying to detect other rarely used nucleii in animals and people and explore other relatively rarely used MR contrast mechanisms.

 

[durfen]

I'm not sure I want to get too specific because it would be easy to figure out who I am (was it that hard anyways?), but I work on MR detection of a rarely used contrast agent in vivo. As a part of this I work on pulse sequence, MR hardware, and image processing development and this is done on phantoms (solutions), autopsy specimens, pigs, and people. I dabble in some other projects as well, trying to detect other rarely used nucleii in animals and people and explore other relatively rarely used MR contrast mechanisms.

You could have just PM'ed me, that's what I was expecting heh. There was one guy I know here who worked on MR detection of Xenon for lung characterization (i.e 3D visualization of the lung).

Ah. processing done. time to do some work.

 

[Neuronix]

It's okay I have little reason to hide who I am. As a moderator I like to be open and honest with everyone involved. I can only hope that there are lurkers or future readers who will benefit from the open discussion and contribute.

We also have people here who do hyperpolarized noble gas MRI (He and Xe), but that's not what I work on.

 

[durfen]

It's okay I have little reason to hide who I am. As a moderator I like to be open and honest with everyone involved. I can only hope that there are lurkers or future readers who will benefit from the open discussion and contribute.

We also have people here who do hyperpolarized noble gas MRI (He and Xe), but that's not what I work on.

Yeah. During the interview process I had a tough time finding faculty in this area. Very few institutions have established programs and professors, the rest have newly minted assisstant professors (who were probably grad students in one of these institutions. Needless to say, I was interviewed by a lot of genomics people. Not that I'm not interested in genomics, I will probably do one rotation in it to see if I'm interested, but it was *very* lopsided. Imagine trying to talk about kV vs. mA modulation of dose to a person who works on evolutionary trees.

 

[LaDandouna]

I am in the exact same boat....and I was actually very happy to read this thread...it is very rare to find people interested in doing imaging research...in most of my interviews, I was the only one....

So, Neuronix, I just got into Penn's MSTP today and although I loved Penn during my interview, I was starting to have second thoughts since the imaging people that I talked to in the BIOE department did not seem to have very strong research...which department r u doing ur research in?

 

[durfen]

I am in the exact same boat....and I was actually very happy to read this thread...it is very rare to find people interested in doing imaging research...in most of my interviews, I was the only one....

So, Neuronix, I just got into Penn's MSTP today and although I loved Penn during my interview, I was starting to have second thoughts since the imaging people that I talked to in the BIOE department did not seem to have very strong research...which department r u doing ur research in?

Same problem here. Hey maybe we can be research buddies!

 

[Neuronix]

Well the thing is my mentor has students from BMB (the department I'm in) and Bioengineering. At Penn you don't have to be locked into one department to work with a faculty member. Any faculty member can more or less be in any grad group as long as they can justify it somehow. My mentor and I both think BMB is better than BE for imaging research here for a number of reasons, but that's up to the student to decide. You can PM me if you want more details...

 

[LaDandouna]

So, Neuronix, whats BMB? Also, I know that Penn was trying to hire Wang from Stanford to improve their imaging stuff...do u know anything about that?

Hey, Durfen: what kinda imaging r u interested in?

 

[Neuronix]

BMB is the Department of Biochemistry and Molecular Biophysics. I like them because they're very flexible (only 1 required class) and from there you can make your own cirriculum (from classes in almost anything). They'll let you do 2 rotations if you want, you'll defend your pre-lim after one year, and they give $1500 towards a laptop when you start grad school. Otherwise I just found it a very inviting, friendly, and collaborative department. My boss's primary graduate appointment is in BMB, and while as I mentioned at Penn that doesn't really matter, he's just found that BMB is more flexible about things like coursework and thesis requirements and such.

I have no idea who Wang from Stanford is, but there's a number of completely separate areas who do lots of things that can be called imaging.

 

[LaDandouna]

BMB sounds like a good plan then

 

[Neuronix]

Actually, after thinking about what I said earlier, I should rephrase. I think BMB is better for MR and maybe Optical Imaging at Penn. Many of the faculty in other modalities are not well represented in the department (but potentially could be).

 

[durfen]

I'm in clinical CT and MR, and that's probably where I will be lurking when doing research.

 

[durfen]

Actually, after thinking about what I said earlier, I should rephrase. I think BMB is better for MR and maybe Optical Imaging at Penn. Many of the faculty in other modalities are not well represented in the department (but potentially could be).

Yeah BMB may be rather low-level for me , sort of like processor design vs. operating systems vs. compilers/languages vs. applications vs. software engineering.

I'm sort of between application and software engineering level, heh.

A prof. at Penn thought I would be a good fit for the HHMI program:
http://www.uphs.upenn.edu/ime/phdprogram/home.html
What do you guys think?

And Wang:
http://med.stanford.edu/profiles/frdActionServlet?choiceId=printerprofile&fid=6894

Optical guy, that's why I don't already know him.

 

[LaDandouna]

I was strongly encouraged to apply for the HHMI program during one of my interviews too....I think it is an excellent program too...I already did a HHMI internship with a HHMI investigator this past summer....their programs are just awesome!

 

[durfen]

I was strongly encouraged to apply for the HHMI program during one of my interviews too....I think it is an excellent program too...I already did a HHMI internship with a HHMI investigator this past summer....their programs are just awesome!

Then you know more than me.
Know anything about UW?

 

[Neuronix]

The HHMI program is brand new. They're likely trying to recruit as I think their first class is this fall. I'm not saying it's a bad idea to want to join that department, but talk to Maggie about it first before you get set on doing it. Since it's totally new there may be hurdles for MD/PhDs there.

 

[durfen]

The HHMI program is brand new. They're likely trying to recruit as I think their first class is this fall. I'm not saying it's a bad idea to want to join that department, but talk to Maggie about it first before you get set on doing it. Since it's totally new there may be hurdles for MD/PhDs there.

Right, thought so. My reply to Skip's "Interest in Penn" email contained my interest in the HHMI program....given that I still got accepted means the hurdles can't be that high. There's also a medical image processing group isn't there?

 

[Neuronix]

Right, thought so. My reply to Skip's "Interest in Penn" email contained my interest in the HHMI program....given that I still got accepted means the hurdles can't be that high. There's also a medical image processing group isn't there?

They never put 100% stock in that what you join wanting to do is what you'll end up doing. Even if you interview saying you really want to work with X professor who the program wouldn't let you work with, they'd still accept you if they liked you. Trust me, you should clarify this at second look if it's a big issue for you. I may be making a mountain out of a molehill, but some things are not as simple as "Well I mentioned it to Skip and he didn't object..."

The medical image processing group I know of is:
http://www.cis.upenn.edu/~gee/

A pretty non-descript website, but he's the head of a number of faculty, post-docs, and students up on 3600 Market St.

 

[durfen]

They never put 100% stock in that what you join wanting to do is what you'll end up doing. Even if you interview saying you really want to work with X professor who the program wouldn't let you work with, they'd still accept you if they liked you. Trust me, you should clarify this at second look if it's a big issue for you. I may be making a mountain out of a molehill, but some things are not as simple as "Well I mentioned it to Skip and he didn't object..."

The medical image processing group I know of is:
http://www.cis.upenn.edu/~gee/

A pretty non-descript website, but he's the head of a number of faculty, post-docs, and students up on 3600 Market St.

Yes this is the guy I have been in contact with (and recommended the HHMI and was mentioned). As you say, I will clear up what the hurdles are. Thanks.

 

[LaDandouna]

Then you know more than me.
Know anything about UW?

I am actually going to be at HHMI headqurters for a couple of days at the end of May...I am gonna make an extra effort to find out more things about the program especially since it is brand new...

Unfortunately, I dont know much about UW