Saturday, June 22, 2013

Fine arteries and damage from hypertension

Centerlines of arteries from 7T time of flight MRI image of a hypertensive patient

The 7T MRI let us see arteries that normally don't appear in MRI I had to learn new artery names this is the associated paper: I was limited to what images I could put in the paper.
Centerlines of arteries from a 7T time of flight MRI image of a normal blood pressure patient
You can see that the lower second healthy patient has more visible small lenticulostriate (LCA) arteries and they are straighter than the upper high blood pressure patient. So we are actually seeing and able to measure damage from hypertension. The centerlines are there to give us something e could measure.

Cancer tumors may starve

I am amazed how large cancer tumors can get and the patient still lives. A tumor is a lump but if it is not next to some vital structure it's actually just a large mass and we can live with even huge tumors. Fast growing tumors are growing uncontrolled and very metabolically active so they are sucking up energy or blood sugar from the body. Also we treat cancer with chemotherapy which makes the patient sick and causes the patient to lose their appetite while they are dealing with this growth that is sucking up the patient's energy possibly driving the patient to starvation. This makes it so important for cancer patients to eat well during treatment. This may also explain why sudden weightloss is a sign of cancer and also why cancer patients' immune system are reduced, the tumor is sucking up the energy that would make gone to the immune  system.

Friday, June 21, 2013

MRI gradients

I have a large pool of research images I generated and I think they're really interesting so I will start posting them.
This image shows segmentation of arteries from a Time of Flight-MRI. The arteries are about the brightest thing in the image and we used that to segment them. But the magnetic field of the MRI is not consistent so the other tissue in the back of the brain came out bright.
Arterial segmentation from Time of Flight MRI
To remove this we ran a median filter over the image and created a blurred image.
Median filtered image
We subtracted the blurry median filtered image from the original image which got rid of the intensity gradient. Then reran the segmentation and isolated the arteries without the background.
Segmented arteries after subtracting median filter blurred image
Then we could make quantitative measurements on the isolated arteries. I used this technique to extract the small arteries from a 7T time-of-flight MRI image of hypertensive subject in this paper

Computer everywhere but on the way out of the way

Computers spreading everywhere and they next step is for them to get out of the way. Physicians are spending more time with electronic medical records (EMR) and less time with patients but the real value will be when the physicians can spend more time with patients because they spend less time digging around trying to find things in the medical records.

We are getting the computing power now we need to focus of developing the intelligent software that predicts and provides what is needed. Even if it's right only part of time it can speed things up. First the desktop computer will disappear, then the laptop, and then even handheld computers will disappear because the computer brains will either be embedded in the the device and the device will be connected to hidden computer servers if it needs more power or more importantly integration with data. We will start to see thermometers, stethoscopes, pressure gauges, microscopes, X-rays, CTs, MRIs all hooked up to the networks of information that can tell when a medical measurement is out of the ordinary and what that may signify, but we won't even see the computers anymore.

The cost of walking and riding

We make walking, riding public transit, and bicycling expensive and driving cheap. Walkable, bicycle friendly, transit oriented neighborhood have been restricted there aren't enough to meet demand driving the price up. Walkable neighborhoods tend to be more expensive than car dependent neighborhoods this should encourage builders to want to build in this kind of place because they would actually get better prices for their buildings. We don't have more because we actually restrict this kind of denser neighborhood with parking minimums, height restrictions and it does take quite a bit of planning to build a bike route. In places where there are bike routes they do get used extensively.

The reason this concerns me is that we make it so hard, dangerous, and expensive to get any exercise which will of course lead to more diseases. We are not doing the simple things on the level of society to be healthy and then trying to rely on high tech medical research which I work on. The high tech medicine should be the last resort when things go wrong and things will go wrong often enough that we don't need to add to health problems intentionally and I think we are by subsidizing driving cars so much and making it hard and even dangerous to walk, bike, or go outside without a car around us.  

Don't be bad faster

Clinical biocomputing tends to reuse well tested algorithms and procedures over and over again as opposed to research biocomputing which is always trying something new. In clinical biocomputing we have the opportunity to guess what the user is trying because clinical medicine tends to repeat the same fewer techniques over and over again. If we can guess what the user will do we can precalculate the solution. This will allow us to use more accurate even if slower algorithms because we can calculate it ahead of time.

In computing we are obsessed with speed, faster is better, but we actually have to do something accurate enough to be useful or we are just bad faster. 

Wednesday, March 13, 2013

Gene expression over mutation

The main utility of genetics in health care may be in gene expression not mutation. We have this idea that by knowing our genetic profile we know our genetic disease destiny and can do something special about it. But disease causing mutations tend to either be rare variants with a strong ability to cause a disease or common variants that only slightly increase your risk of disease. If one of this rare strongly disease causing mutations runs in your family you probably know it and you could get tested to see if you inherited it, if you want to know and if there is something you can do about it. But most of us will have the other kind that increase risk slightly.

For example about 4-5% of people will develop an intra-cranial aneurysm, an artery balloons in the brain and has a risk of rupture, during their lives. About 1% of aneurysms rupture a year so most people will live a long life and die of something else, more likely heart disease, than die from the aneurysm. If you come from a high risk family that have a genetic predisposition to aneurysms you have double, 8-10%, life-time risk of developing an aneurysm. And you may also live out your life with the aneurysm still intact. Smoking may quadruple the risk of developing an aneurysm to 16-20% life-time risk, an even stronger risk factor than the genetics.

So what should you do if you have a high genetic risk factor for intra-cranial aneurysms? Don't smoke, control your blood pressure by exercising, eating more fruits and vegetables and less salt, and get regular check-ups. What should you do if you have a normal genetic risk for aneurysms and don't want a 16-20% chance of developing an intra-cranial aneurysm? "Don't smoke, control your blood pressure by exercising, eating more fruits and vegetables and less salt, and get regular check-ups." Sound familiar?

If you have the common mutations that slightly increase disease risk, which is more likely than the rare strong effect mutations, and want to avoid disease (and you actually do want to avoid these diseases) you will do the same things if you have normal risk and don't want these diseases. Prescreening for genetic risks may not make much difference to healthcare.

Many mutations that cause disease happen after we are born. Cancers often start from DNA damage, from the sun, smoking, pollution, diet, viruses, aging, and some may appear to be spontaneous. The immune system may stop many of these cancers early on before we even know we have a tumor. But once they get established tumors often have distinct gene expression profiles.

In a disease like the metastatic cancer glioblastoma certain genes are commonly over expressed and corresponding microRNAs are under expressed. MicroRNAs are short nucleic acids that suppress  genes. Many genes and microRNAs work in balance with the gene up and the corresponding microRNA down. By measuring the gene over-expression profile of a disease and corresponding under expression of microRNAs we could develop treatments by giving the patients small interfering (si)RNAs that mimic the under expressed microRNAs to down regulate the over expressed genes. These patterns of over-expressed (or under-expressed) genes and corresponding microRNA expression patterns could be a key avenue for use to develop highly targeted disease treatments.

Sunday, March 3, 2013

automated software testing

Medical or really any other software should be built with automated testing from the ground up. Unit tests are small test for each unit/class of software. Writing unit test from the beginning of a software project forces the developers to make code more modular and loosely coupled because the unit test runs one class of code in isolation. The unit test also acts as "living" documentation. The unit test shows an example of  how to use the class, and if the unit test works then we know the "living" document is up to date. When developers document source code they often go back and change the source code but not the document leaving a misleading document.

Unit tests are only the first layer of automated testing. Automated tests of the classes working together are needed. These automated tests are more difficult to write but a necessary addition to manual testing.

A manual tester has to run the test the first time to discover the potential problem and then test automation engineers are needed to automate the test because every test has to be reran frequently during development because source code changes can break code that was working before. If we rely only on manual testing we would keep adding and adding new manual tests and we would have to keep adding manual testers who would quickly get bored of rerunning the old tests. Without automated testing the old tests probably just don't get reran because it's too time consuming and dull.

As software developers we really can't say we cannot automate software tests because software development is is all about automation of processes and with medical software it actually could be for brain surgery.

Food work subsidies not food security subsidies

So sugar is looking bad lately, recent studies on diet are showing sugar (cane sugar or corn syrup, it really doesn't matter which) and simple carbohydrates cause diabetes even more than equal calories from other foods. Studies in the New England Journal of medicine are showing the Mediterreanian diet high in fruits, vegetables, nuts, sea food, olive oil and wine protect from heart disease. The last two are, of course, the most popular. 8% of Americans are diabetic and 16% our pre-diabetic so almost 1/4 of Americans cannot process sugar well. 

In response, local governments are banning and taxing sugary drinks and trans fat and they are accused of being a "nanny state". But the first step should be to simply stop subsidizing sugary, carbohydrate laden foods and stop being the treat giving (maybe we should say "granny's cookie") state. Out farm bill provides most of its subsidies to large farms growing corn, corn and more corn, wheat, soy beans and cotton - cotton isn't even food. These programs started in the 1930's to keep farmers employed so it is more of a work program than a food program. Around 98% of America's farmland is used for growing cereals. With so much cereal grown the only way to store so much of so few types of crops is to turn corn into corn syrup and corn oil, wheat into processed white flour, which stores longer than whole grain, soy beans into oil and cotton into cloth. And these subsidies actually reduced the farmland devoted to fruits and vegetables because if farmers take the subsidies they plant fruits or vegetables on the same land. They was to protect the fruit and vegetable farmers from subsidized competition. The result is that our fruits and vegetables are mostly grow in San Joanquin Valley in California and a smaller amount in southern Florida, especially in winter,and trucked all over the continent. This reduces our food security because our nutrition is all grow in a few places. Bakersfield California where the trucks collect has the worst air in America from the truck exhaust and we get our "fresh vegetables" trucked long distance from such a "clean" environment. By simply ending these subsidies farmers would grow more diverse crops as they each try to find a niche to grow crops for the best prices and we would have healthier, more diverse food grown over a wider area making our food healthier and more secure and this actually wouldn't cost us anything except votes from farm communities. 

Saturday, February 2, 2013

Bicycles and walkable neighborhoods

I ride bikes, I used to race bikes, They are the greatest local transportation. They can get you out 10 miles or even more, they're great exercise, reduce air pollution. They can help us solve our diabetes crisis and air pollution. When there are paths for bicycles they are used from what I have seen. Bicycle commuting makes up less than one percent of commuting so it really is trivial, but when there are trails people really do ride to get places on them so we need to build more. And people must want these because houses are more expensive in areas with bicycle trails, so there aren't enough. I find it ironic that when there is a bicycle trail the house prices go up so much because then only wealthier people can afford to live close enough to a trail to save money riding a bicycle instead of driving. Actually neighborhoods with trails for bicycling or walking can be so much more expensive than other neighborhoods it negates the cost savings so we really need to make more so these neighborhoods become affordable.

Look whats going on in Salt Lake City where I used to live:

Or worse in Beijing:

This bicycle even gets power from the brakes to use for going up a hill: Since it's not the Tour de France we are allowed performance enhancing technology.

How to pay for roads

We actually subsidize roads greatly. You don't pay for how far you drive. There are gas taxes and car license taxes but they don't nearly pay for the roads. Much of the cost of roads comes from payroll taxes. We could increase the gas taxes to pay for roads. I have seen that might cost $4.00 per gallon and raise the cost of gas to close to $7.00 per gallon. That would probably get people buying electric cars which wouldn't pay for the roads so we would have to switch to a system that tracks distance, which is a little creepy because that is probably some kind of GPS track how far and then also where your car went. If driving a car cost what it truly costs trains, subways, and buses (which are also usually subsidized) could raise their prices to pay for themselves. Transportation would probably be more expensive but would also be more efficient since we are bearing the full cost. And walking and bicycling would get more popular benefiting our health, air quality (since we aren't driving as much), and the environment. And all because we stopped subsidies for transportation.

Building for health

I work developing medical information technology. But some of the most important health issues could be addressed by building our physical living spaces to encourage health. Most of our "built environment" encourages driving and not walking for bicycling. We have a serious and growing problem with diabetes which is encourage by our collective lack of movement. We are actually paying and regulating for this problem in many ways. We subsidize driving by using payroll tax money to make free public roads, the cost of roads should be placed on driving itself so that roads get built where the most people use them. Many places limit building height which makes housing, work and shopping too spread out to use anything but a car. We regulate single use zones for housing, offices and working putting everywhere we want to go far apart and only accessible by car. Then we pay again with poor health and increased health insurance premiums.

Often the only way to get any exercise is too pay for a health club membership, drive your car through the outdoors at more expense for gas, then get on an exercise machine with a video simulation of the outdoors you just drove through. This takes time, money, and motivation that few of us have to exercise.

I know people want more walkable neighborhoods because the most walkable places to live are always the most expensive. That means we are competing to live there. So let's build more of these places. It should be easy because all we have to do is stop regulating against denser, higher floor buildings and mixed use development. Developers should want to build these places because rents and sales prices of the buildings will be higher than car dependent development. This would make getting a walk in easier for everyone and save us all the pain and costs of poorer health basically for free.

When we make taller buildings we could actually give ourselves more green space. Make the buildings taller but then put larger parks, playgrounds, gardens, and sports fields between them. This would put us closer to nature even though we are living in taller buildings, provide sunlight, encourage walking and bicycling more, improving our health and environment.