Profile cover photo
Profile photo
Chad Haney
69,648 followers -
I'm the medical imaging, Red Wings, Formula One, Tech guy.
I'm the medical imaging, Red Wings, Formula One, Tech guy.

69,648 followers
About
Chad's posts

Post is pinned.Post has attachment
Medical visualization, it's what I see and do
I was going to summarize an article about the history of medical visualization that was discussed in MIT Technology Review in 2012.
The Future of Medical Visualisation
http://goo.gl/HnpAQq
However, I think that the #OpenAccess  article that they reference does a good job on its own. It's kind of strange to think of a review of a review of a review article.
From individual to population: Challenges in Medical Visualization
http://arxiv.org/pdf/1206.1148v2

Rather than review the review of the review, I'll add a few comments and answer your questions. So read either the MIT Tech Rev article or the journal article and ask questions. This is an opportunity to talk to a scientist that works in the medical imaging field.

The article mentions multi-modal volume visualization. If you have been following my  #CHMedicalImagingSeries then you know that each imaging technique (modality) has strengths and weaknesses. Combining imaging modalities, like the MRI and CT below of my head, allow you to take advantages of the strengths and minimize the weaknesses. To get the best out of multi-modality imaging you need to be able to fuse the images. The technical term is image registration or some like image co-registration. There is a lot of research in just this technique to make it more automated. One common technique is called mutual information. Our brains can easily tell that the dark material in MRI is bone (e.g. my skull) and it should match the bright material in CT. Mutual information tells the computer to consider that bright could equal dark by normalizing the images first. The principal axes of the objects are also used to register (align/fuse) the two images. For therapy planning, both surgery (cosmetic) and tumor resection/radiation, multi-modality imaging can have a huge benefit. They mention that in 1993 Altobelli used multi-modality imaging to visualize the possible outcome of complicated craniofacial surgery.

Another use of medical imaging visualization is virtual colonoscopy. Visualization tools that you need are surface/volume rendering, skeletonization, and segmentation.
UCSF Radiology: What Virtual Colonoscopy CT Scans Look Like

Surface/volume rendering is just what it sounds like. The data is analyzed and the surface can be identified and displayed with shading and lighting such that it looks 3D. You can make it true 3D with the right equipment (3D glasses, 3D monitor, and software to split the data into left and right views) but that's not essential. It can't be emphasized enough that modern GPUs have made these difficult calculations become trivial. Some of the early animation work and medical image visualization required high end UNIX workstations. Now that same level of visualization can be done with a low-end gaming PC.

Segmentation is also, just as it sounds. There are automated and manual segmentation tools. For example in the heart and skeleton images below, the tissue of interest has been segmented out of the "background" tissue, e.g., the internal organs, muscle, etc. Again, there is research in this technique alone. Our brain can look at a medical image and identify parts of the brain or organs quickly. "Teaching" a computer program to do that automagically is very difficult, especially if there is motion due to breathing. In that case, you may have to use image registration to get rid of the motion blurring first.

Skeletonization is a process of identifying paths. For colonoscopy, that would be teaching the program to traverse the path of the colon. I've done work where we were measuring blood vessel diameters in a pulmonary hypertension model. Skeletonization was used to automatically identify each part of the vascular tree. From there, it was easy for the software to measure each diameter.

The first three images are fused images of a CT and MRI of me. The yellow surface rendered part is from CT as it shows bone (skull) very well. The grey-scale part of the image is MRI which shows soft tissue very well. The rest of the images are from a Toshiba 320 slice CT. In CT technology, a ring of detectors is used capture the signal from the x-ray source. Each ring is called a slice in clinical CT machines. For a while 64 slice was considered the best. Now 256 and 320 slice machines are becoming available. More slices means you can cover a larger area in a shorter amount of time. So highly detailed images of the heart can be acquired without motion artifacts from the beating heart. Likewise for the lungs.

Here's a few older posts that will hopefully help you understand the article.

Medical Imaging 101 pt 1 (http://goo.gl/LTWUf)
Medical Imaging 101 pt 2: CT (http://goo.gl/IHaFw)
Medical Imaging 101 pt 3: MRI (http://goo.gl/UVbiU)
Functional vs. anatomic image (http://goo.gl/UTPK7)
Visible Human project (http://goo.gl/cv2xU)
Eye of Horus post (http://goo.gl/qpxyh)

Image sources other than the above article:
Lung and brain CT images (http://goo.gl/HHhqSJ)

CT Heart (http://goo.gl/eqbFde)
#ScienceSunday  
PhotoPhotoPhotoPhotoPhoto
Medical Imaging Visualization
8 Photos - View album

Post has shared content
One of my favorite photos from one of my trips to the Dominican Republic. Today's #joinindaily theme is piers. 
Water as a treat?
I missed #fidofriday but I won't fight over it. It's easy to forget how essential water is. This couple isn't fighting over water but they were more interested in water than food. The male, the black fur guy, chased away 2 other strays.

Edit:
I was on vacation and this photo is from the Dominican Republic. It reminded me of a conversation with +Lex Barron about my dog's trainer and Raymond & Lorna Coppinger. Here's part of that conversation.

Our trainer showed us that visual cues are better than verbal ones because dogs look for patterns when hunting.

I actually think a lot of dog trainers go a bit heavy on the wolf/pack thing... in that Coppinger book I linked to, he talks about the rapid social and biological evolution dogs underwent during their 'domestication' and I have to say I am inclined to agree.
I was lucky enough to meet a guy who had a wolf dog (and not a husky mix this dog was absolutely wolf, at least 70% wolf) and he confirmed the idea that a wolf is an entirely different animal; behaviourally dogs and wolves don't have that much in common. His wolf-dog was like nothing I have encountered before. A totally different set of rules.
I would say dogs have evolved to use visual cues in communication with humans, along with sounds (tones and intonation, not words- I had Mark in hysterics a few nights ago when to prove this point I whipped our dog, Sprocket up into a state of frenzy by saying "paper towels" in the same tone we use for 'squirrels') and that this is just as much a factor as their 'hunting' strategy. In fact I doubt a lot of dogs are visual hunters, I think smell may well play a much bigger role than vision.

If I remember the story correctly, my dog's trainer, Jennifer from Call of the Wild (http://goo.gl/W4EyL), said she went on a trip with her mentor, Coppinger. They were observing "stray" dogs in a dump in either Mexico or somewhere in South America. Food was more or less in abundance at the dump but not water. So the humans that would scavenge at the dump developed a symbiotic relationship with the stray dogs. They would provide water and the dogs would guard their belongings. The dogs would "team up" with particular people not because they bonded with them but because they recognized those particular people as valuable resources, i.e., water providers. Kind of first come first served. Jennifer discounts a lot of discussion of packs with stray dogs. They are scavengers and work together when it's convenient.

Dogs: A Startling New Understanding of Canine Origin, Behavior & Evolution
by Raymond Coppinger and Lorna Coppinger
http://goo.gl/eLL7g

You might want to check out my other post about  another dog book.
Made for Each Other
http://goo.gl/QbCZ8

#ScienceEveryday  and #FidoFriday  combined.
Photo

Post has attachment
Pete Souza
Well played sir.

Post has attachment
More Spring Flowers
My lilac bush smells amazing. I think I missed the peak bloom of the allium while I was in Nashville. The irises look kind of sad this year. 
PhotoPhotoPhotoPhotoPhoto
5/22/17
7 Photos - View album

Post has attachment
Half Acre Double Daisy Cutter, nice double IPA with lunch and a little hockey. Now it's time to mow the lawn. How's your weekend going? 
Photo

Post has attachment
Check out the #BIllMeetScienceTwitter hashtag on Twitter. Unfortunately I don't know the story behind the hashtag but I like seeing so many scientist on Twitter. It's also reassuring to see high visibility of female scientists.

ETA
Here's the story behind the hashtag.
This is a better link for background
https://storify.com/may_gun/billmeetsciencetwitter-in-context

The headline for the Forbes article is misleading.
https://goo.gl/RVxcaA
A few scientists on Twitter decided Bill wasn't letting enough experts talk about science on his new Netflix show.

Post has attachment
No comment

Post has shared content
Removing Photos as a separate app = galactic mistake
One of the dumbest moves by Google was to castrate Photos in G+. I echo everything +Kee Hinckley said. Additionally, have you tried changing your profile picture using an existing profile photo? Good luck.
Google has been touting their photo product, but they still make it impossible to actually use by anyone with more than a few hundred photos.

1. Every single product that can add photos gives you a simple list of photos in chronological order. The great search capabilities of Google Photos? Sorry, can't use them. Albums? Sorry, can't use them. Nope, I have to scroll through 20,000 photos from the past two decades trying to find that picture I want.

2. Let's say I've actually used search. I found a bunch of nifty photos of stone walls. I've saved the ones I care about in tabs (because there's no bloody "favorite"). I select the first one, "Add to Album". "Create New Album". I name it. I save it. I go to the next photo. "Add to Album". Now where's the damn album I just created? Its not there. Oh yes it is. The albums are sorted chronologically. By the date of the photo I put in it. I don't know what the date was! So I scroll. And I scroll. And I scroll. Can I search for the album name? Nope. Can I go to Google Photos and search for the album name? Nope. If I find it in Google Photos, can I drop photos on it? Nope. I've got to find it in that scrolling list. There is no other way to go from searching for a photo to adding it to an album.

This is a toy. A lousy toy. Gorgeous AI. Horrible UX.

#plusonly

Post has attachment
Photo

Post has attachment
Goodbye Nashville
Great conference for preclinical imaging hosted by Vanderbilt University. Nevertheless, it feels good to be going home. 
PhotoPhotoPhoto
5/16/17
3 Photos - View album
Wait while more posts are being loaded