Fighting Breast Cancer Through Mobile Phones?

There are many examples of how society can use The Extraordinaries software to do social good. Below I propose a way to tackle breast cancer.

While visiting my family for the Thanksgiving holiday, we got into a discussion about the medical sciences, and in particular about mammography.

Apparently, there is a shortage of radiologists, and my mom has to wait 11-months to have her mammogram looked at. For many people, especially those genetically prone to breast cancer, 11-months may mean the difference between “caught just in time” or “a few weeks too late”.

This got me thinking. What can we do about it?

Each year, tens of thousands of people participate in Susan G. Komen Race for the Cure walkathons. Thousands more around the world face difficult times each year, as their sisters, mothers, daughters, and wives are told that they have breast cancer. That is a ton of people who have a direct vested interest in prevention, and a ton of potential human energy just waiting to be tapped.

Before I offer a potential solution, let me first show you a picture of two mammograms.

On the left, is a normal healthy image. On the right, is a lump showing cancer. It isn’t blatantly obvious, but it isn’t that hard to see. In a sense, we are looking for shapes, for things that stand out as abnormalities. With brief training, I believe that thousands of people could volunteer to help identify problems. They can view the images on mobile phones,  draw circles around areas with potential problems, all delivered through our software.

I propose, that we crowdsource the scanned imagery. I am not suggesting that we replace doctors or radiologists with everyday people. I am saying that we reinforce them, give them 500 extra sets of eyes. After all, doctors are human, and they can miss things. Medicine is one of those fields where science is often surpassed by the artistry of a doctor’s talents. A doctor is a human making a judgment call, based on a specific set of past experiences, and their opinions vary. As a result, many of us get second, third, and fourth opinions.

If we had 500 people look at each x-ray, and the average of those people point to problem in one quadrant of screen, it might raise attention of a doctor to look again. Like Google page rankings, those people with the highest number of hits regarding a potential problem, could get moved to top of list for an appointment.

My business partner Ben Rigby offered his take on this idea:

A problem could be false positives - or cases that are less obvious. Then what do you do after the crowd identifies a problem? Also, why cant you look at your own xray and see the problem, if you had training? Maybe it's more a matter of training, than a case for crowdsourcing. The case becomes very interesting when trained doctors are volunteering... then it's a real match up of high-expertise and people in need.

Ben makes some interesting points. However, I would actually disagree that looking at your own xray is the best approach. I would take one look at my xray, and freak about every little dot on the screen. Contrastingly, the aggregate intuition of several hundred people would be a little more selective in their judgment of what is, or is not, a potential problem.

Nevertheless, in a triage approach, I think this would work.

Having the crowd sift through tens of thousands of backlogged mammography scans to identify those that need the most attention would add tremendous value to the medical community and save thousands of lives each year. When 70 out of 100 people point out a potential problem in one quadrant of a screen, maybe a doctor should look at it a little faster than 11-months.

What do you think?

Tags: breastcancer, komen, breast, susan g. komen, mobile phones, mobile volunteering, online volunteering, nptech, cancer, medicine, doctors, radiology, radiologist, fight, race for the cure, fighting

Comments

[Sumit Chachra]

This is similar to a hot or not idea. Not sure why mobile is the best way to help out, since given broadband speeds on a desktop/laptop I or anyone else could do 5-10 mammograms in a minute.

Accuracy doesn't matter as much since you're going to average out results.

All this said, there are literally 1000's of image processing algorithms developed by 100's of Universities that can automagically detect such shapes/cancers etc. with 99%+ accuracy and no human intervention. I would go that route, since making non-doctors do this, will definitely raise eyebrows (Even if you say you're not trying to "replace doctors").

MY 2 cents!

[JacobColker]

Hi Sumit,

Thanks so much for your comments! We really appreciate your thoughts.

I have two counterpoints for you.

1) Why Mobile? Our research has taught us that time is the single biggest factor in someone’s ability to engage for social good. Without several hours free, few opportunities exist to get involved. In a hectic day, people do have free moments, but often in blocks of only a few minutes. Linked together, their aggregate spare time can have a massive impact.

To harness people’s ad-hoc spare time, we have to reach them during those few moments throughout the day when they are free. When riding on the bus, standing in line, waiting for a meeting, eating lunch, or sitting in an airport. In these moments, people are bound to have a mobile phone within reach. That is really what this whole project is about. We want to harness that ad-hoc human energy that exists throughout someone's average day, and convert it to social good.

By removing the constraints of a physical location, four walls, and a hard line broadband cable, we increase the opportunities throughout the day when someone can engage with our system. You aren't always in front of your computer, but you almost certainly have your phone within reach throughout most of the day. That is why mobile.

2) There really aren't that many image processing algorithms worth their salt. That is why Google still needs help with image tagging (http://images.google.com/imagelabeler/) and why NASA still needs help with identifying craters on Mars (http://clickworkers.arc.nasa.gov/), and why reCAPTCHA still needs help digitizing the world's books (http://recaptcha.net/), and many more. If there was such technology, then why is there still an 11-month backlog of mammography scans? I think the biggest hindrance, is that there is a lack of people. We are missing critical mass to make a system like this work. That is why so many companies have tried to replace humans with algorithms in the first place. But with the powerful technology available in smartphones, we can actually reach that critical mass. And frankly, the human eye is backed by intuition and judgment, two qualities that computers don't have. When it comes to saving lives, I think that humans will be more accurate in the long run.

Any reaction? Thanks so much for debating with us! This is what it's all about!

[Sumit Chachra]

Hey Jacob,

I agree with the first counter-point, although my thought is there is no reason to limit oneself to mobile, it needs to be part of the strategy. Its a good starting point though, since I u/s an idea needs somewhere to start with, and you guys are focusing on delivering volunteering solutions via mobile, which is great (mobile internet is huge in Asia/developing world)

As for the latter I don't agree. Image labeler was just an experiment, an extension of a similar program at CMU if I remember correctly. Computer Vision is not a solved problem (just like AI), and probably never will be. But specific domains have made significant advances, specially in the medical field.

For example picasa can now automatically detect and tag faces in pictures http://picasa.google.com/features-nametags.html You have to thing cancer detection is way easier than recognizing (and learning) faces. Here's a 1993 link to cancer detection: http://portal.acm.org/citation.cfm?id=164155.164199&coll=GUIDE&dl=GUIDE Not sure why there is a 11-month wait, I do know its a huge outsourcing business too. I guess computers can't replace doctors, but I do know these softwares are definitely being used for guidance by doctors!

In any case, I think you guys are working on an awesome idea. Anything to help non-profits has to be helped and commended!! Although I am sceptical of this particular idea, I think you'll have to throw tons of these out there, and hopefully a few will stick and be viral, and ultimately help the segment you are trying to help.

Best of luck!!