DiagnoX – A New Way to Diagnose Rare Diseases

DiagnoX is an open-source project dedicated to diagnosing diseases. The original code is designed and geared towards Aortitis, a rare cardiovascular disease characterized by inflammation in the aorta. Whether you are searching for inspiration for a personal project, a new hobby, or you just want to give back to the community, DiagnoX is a great way to start.

To store and compare the CT scans, the python program uses HOG (histogram of oriented gradients) descriptors. They store a histogram of gradients (consisting of x and y derivatives which have direction and magnitude), which is more efficient than storing the entire image because the “useful” data consists of abrupt changes in the derivatives.

For classification, Linear-SVC, an algorithm which establishes a hyperplane between clusters of data, is used. LinearSVC uses the parameters which the HOG descriptor provides to train the program and draw the hyperplane, effectively classifying each image as either having or not having an inflamed aorta.

With an overall accuracy rate of 94% and a type II (false negative) error rate of only 1.4%, the algorithm proves to be effective.

If you would like to join this community of developers by creating efficient algorithms to diagnose rare diseases, please click here.

 

 

Advertisements

Polarization of Light

Light is an electromagnetic wave, composed of electric fields. A single wave of light only oscillates in one direction, while a ray of light, which contains a myriad of light waves oscillates in many directions. When light, whether it be a singular wave or multiple, only oscillates in one direction, it is polarized.

Screen Shot 2019-01-14 at 5.02.28 PM.png

Also known as “polarization by use of a polaroid filter,” the most common method of polarization is polarization by transmission. A polaroid filter is composed of a thin plastic material, and as evident by its name, “polarizes” light.

Screen Shot 2019-01-14 at 5.02.37 PM.png

If two polarizing filters are placed perpendicular to each other, then no light would pass through. This happens because the first filter would only let light polarized in one direction through, while the second would only let light oscillating in the perpendicular direction through. Thus, no light would pass through, leaving the area of intersection between the two filters pitch black.

 

Polarizing filters have many surprisingly interesting applications, ranging from sunglasses to making 3D movies possible.

 

When light reflects off a surface, it is polarized. This phenomenon is known as “polarization by reflection.” A combination of polarization by reflection and transmission come into play regarding polarized sunglasses. Polarized sunglasses are meant to reduce glare consist of the lens being a vertically oriented polaroid filter. When light bounces off a horizontal surface, it is horizontally polarized. Thus, if the sunglasses only let vertically polarized light through, then the glare doesn’t reach the user’s eyes. Using this same principle, what if one of the filters in polarized sunglasses is rotated 90 degrees. Thus, each lens lets light polarized in completely different directions through. This is how 3D glasses work. In real life, our left and right eye see slightly different images, creating the perception of 3D. 3D movies, likewise, display two images, one polarized vertically, the other horizontally. When generally looking at the screen with special glasses, it may appear blurry, but with 3D glasses, the characters seem to jump out!

 

LIGO: A shocking scientific discovery

Posted by Pranav Kakhandiki| Nov 12, 2018, 12:00:00 PM   

 

      A gravity wave is defined as a disturbance in the fabric of space time. First postulated by Einstein, gravity waves were nothing but a theory – until 2015. Thanks to the LIGO detector, a group of scientists at CalTech were able to confirm their physical existence in our universe.

Screen Shot 2018-11-12 at 11.26.33 AM
A visualization of gravity waves

     Gravity is known to warp space time. To better comprehend this effect, imagine a 2-dimensional cloth, which represents the fabric of space-time. When a mass is placed onto the cloth, the cloth sags, thus demonstrating how the mass stretches space into the 3rd dimension. This sag in the cloth causes other objects to roll toward the center of mass, demonstrating how gravity bends space time. Similarly, masses in the 3rd dimension change the structure of space time. Gravity waves, likewise, are produced when accelerating masses disturb space-time. The masses cause ripples in the fabric, and often happen when two high-density masses, such as black holes or neutron stars, form a twin-system.

Screen Shot 2018-11-12 at 11.27.13 AM
The LIGO detector

     Although the existence of gravity waves were predicted to exist in 1916, and proved mathematically in 1974, they were not physically detected until 2015. The LIGO detector, which stands for Laser Interferometer Gravitational-Wave Observatory, consists of two 4 kilometer buildings at a 90 degree angle with respect to each other. These buildings, often referred to as the “arms”, are narrow, perfectly straight, the exact same length, and transmit light waves. A light wave is produced, split using a mirror, then is sent down both arms. The two waves are then reflected back by mirrors at the ends of each arm and converge at the same point where they were split. Normally, the waves “cancel” each other out, as the troughs of one wave align with the peaks of the other.

Screen Shot 2018-11-12 at 11.28.41 AM
How two waves cancel each other out

     When gravity waves travel through the LIGO detector, one arm slightly lengthens. As the wave continues to pass through, the arms alternate between being longer and shorter. When one arm is longer than the other, the wave peaks and troughs do not cancel out, so the detector senses a signal. As the arms continually shift through being longer and shorter, the strength of the light signal detected varies. This variation detected is a gravity wave!

     From general relativity, to black holes, to Hawking radiation, the discovery of gravity waves was one of the biggest recent breakthroughs in physics. What will the next one be?

Superfluids

Posted by Pranav Kakhandiki| Oct 21, 2018, 5:30:00 PM

 

     Viscosity is defined as the resistance of a liquid to flow, or in other words, a liquid’s “thickness”. Some liquids have greater viscosities than others, which leads to different applications of these very different materials. Oil, for instance, has quite a low viscosity, which is why it is often used as a lubricant. Honey, contrarily, has a relatively high viscosity. However large or small this value of viscosity may be, it is positive, meaning that some constant force is required to move the liquid. But what if there was a liquid whose value of viscosity was equal to zero? These are called superfluids.

Screen Shot 2018-10-21 at 5.22.46 PM.png
Viscosity of Water v. Syrup

     Superfluids exhibit many interesting properties which appear like they defy the laws of physics. No viscosity results in no cohesion force between molecules, which allows atoms of a superfluid to “flow” through almost any medium. For instance, if a superfluid were to be held in a cup, it would be able to pour out of the bottom, because the lack of friction between molecules. As long as the molecules are small enough to fit through the empty space between bonds of the cup, a superfluid would drain out completely. The never-ending fountain is another prime example of an interesting property of superfluids Since they exhibit no viscosity, if a fountain consisted of a superfluid, no kinetic energy would be lost in the process, so the fountain would keep flowing forever.

fountain_1.jpg
The never-ending fountain

    Superfluidity is most commonly shown in Helium-4 (2 neutrons, 2 protons), and only occurs at temperatures below 2.65˚ K. So although the never-ending fountain shown above may seems like a perpetual motion machine, the energy required to keep a superfluid at such a low temperature far exceeds the kinetic energy of the actual system. Although a lot about why superfluidity occurs is still unknown, the commonly accepted theory is that the total spin(angular momentum) adds up to zero, causing the net force between molecules to be zero. The low temperature is required to slow the general speed at which the molecules rotate at. In other liquids, such as water, the net angular momentum does not add up to zero, so forces between molecules create cohesion, thus giving the liquid a viscosity. There is still a lot to be learned about superfluids, from proving exactly how a liquid with zero viscosity can possibly exist to discovering real world applications.

How Statistics can be Deceiving – Simpson’s Paradox

Posted by Pranav Kakhandiki, Edited by Anika Asthana| Sep 23, 2018, 6:00:00 PM

     Statistics is often thought of as straightforward. Data always leans one way, supports one idea, and displays one conclusion. But what if there was a way for data to trick you; to appear simple, but completely change if viewed from another angle. This is the case with Simpson’s paradox.

Screen Shot 2018-09-23 at 3.26.49 PM.png
Political Orientations for Citizens supporting the Landmark Civil Rights Act of 1964.

     The table above provides data for the amounts of democrats and republicans senators who support the Landmark Civil Rights Act of 1964. When analyzing the total votes democrats and republicans, it appears that Republicans favored the Act more than democrats. Overall, 80% of Republicans supported it while only 61% of Democrats approved of it. When dividing each category into northern and southern senators, however, the data tells a different story. In each category, the percentage of democrats who voted in favor of the Act is higher. In the north, 9% more Democrats favored the Act than Republicans. While in the South, 7% more Democrats supported it. So what’s going on? How can the total percentage of Republicans voting for the Act be higher, but the percentage of Democrats is higher in each category? More importantly, what data should we trust? Let’s take a look at another example.

 

Student A Student B
Test #1 99/100 = 99% 1/1 = 100%
Test #2 0/1 = 0% 1/100 = 1%
Total percentage 99/101 = 98% 2/101 = 1.9%

Students A and B both take two tests, but comparing them can be tricky

 

     Take a look at the table above. Two students, A and B, have different teachers, who give two tests in the first semester. While student A has a 98% overall test average, student B only has a 1.9% overall test average. Student B, however achieved a higher grade on each individual test. This vast difference in overall percentages versus individual test percentages occurs because of differences in sample sizes. In this case, although student A scored 0% on test #2, their 0/1 is insignificant compared to their 99/100. Student B, contrarily, scored 1/1 on test #1, which doesn’t influence their grade significantly compared to their 1/100.

     Sadly, there is no single way to escape Simpson’s paradox. In the example above, the total percentages should be trusted over the individual test scores, as the tests were worth different amounts of points. In other cases, the analysis of each individual category is more helpful than looking at the total percentages. For instance, when trying to decide the best hospital to treat cancer, it is better to examine the recovery rate of cancer patients at each hospital, not patient recovery of all diseases. Furthermore, many companies manipulate Simpson’s paradox to their advantage, which makes analyzing the data tremendously harder. The trick is to search for “lurking” variables which hide more information that is not inherently obvious.

 

Image result for statistics magnifying glass
Search for “lurking” variables to get a full picture of the data

     Data can be misleading. Sometimes, trusting overall percentages of data is better than analyzing arbitrary categories, and other times, the very opposite is true. To fully comprehend data and make a conclusion, one must objectively analyze the status quo, searching for hidden features or variables that could provide insightful results. Otherwise, we are susceptible to those who transfigure data, manipulating it to tell a different story that promotes their opinion.

How Artificial Intelligence is Shaping our Future

Posted by Pranav Kakhandiki, Edited by Anika Asthana| Aug 14, 2018, 6:00:00 PM

 

          AI, which stands for for Artificial Intelligence, is a fascinating new field of technology. A sub-field of AI known as machine learning, can automate mundane processes, which gives us more time to focus on more important tasks. Because AI is so new, however, it is hard to predict any benefits or potential drawbacks it might present.

download-1.jpg

          To properly understand AI’s impact, it is important to understand how it works. It is important to distinguish between the two types of machine learning: supervised and unsupervised. Supervised learning uses input-output pairs to train itself on how to perform a specific task. In other words, supervised learning programs train themselves on data provided by the user. For example, if someone wants to predict the cost of a 1500 square-foot house, we would use other data points relating size and cost. If we know the cost of houses which are 500, 750, and 1000 square feet big, then we can use a technique called linear regression, which draws a best fit line, to estimate the cost of a 1500 square foot house. To calculate how close a program gets to predicting values using linear regression, we use the cost function, defined below:

Screen Shot 2018-08-03 at 4.38.19 PM.png
The Cost Function

         

          HØ(x(i))  is the hypothesis function, defined as: Ø1(x)+Ø0, our linear regression model, where Ø1 and Ø0 are the parameters(in this case Ø1 defines the slope of the line, and Ø0 is the y-intercept). y(i) is the actual y value, and m is the total number of (x,y) data-points. Essentially, we are adding the squares of the differences between our hypothesis and the real values, then dividing by the total number of values. The goal of any supervised machine learning program is to minimize the cost function, thus minimizing our overall error in our approximation. The ½ at the beginning is only there to cancel out when the derivative of the cost function is taken, and provides no purpose other than convenience of calculation.

          Unsupervised learning trains a machine without data. But how can any machine learning program predict the size of a house with no other data to estimate with? Unsupervised learning doesn’t predict or approximate with numbers, but rather is used for clustering data. For example, an unsupervised learning program could be able to group articles based on similarity and topic. Artificial Intelligence robots use one or both of these techniques to complete a given task. Semi-supervised learning is the middle ground between the two tactics.

sup-vs-unsup.png
Supervised v.s Unsupervised Learning

          Now that we know the basics of how a simple machine learning algorithm works, we can address its benefits and drawbacks. AI could shape product manufacturing, increasing efficiency and throughput in production. Moreover, AI frees up humans to do what robots can’t do: be creative. Automating simple processes allows employees to think about how they can improve products instead of sorting through spreadsheets.  There are truly endless options with how far AI can go, ranging from automatizing simple tasks like grading tests to performing complex procedures like surgeries. It can even be used to create bots that are similar to humans. Although efficiency and product production can be improved through Artificial Intelligence, there are drawbacks to automating simple tasks, such as the jobs of industry workers. Automating minimum-wage jobs can possibly harm the economy, as it takes crucial sources of income away from a population afflicted with poverty. Though more job positions will open up at AI companies, the total number of jobs might decrease, possibly harming the economy.

          Progress in machine learning has experienced a massive boom in the last few years. Companies are scrambling to create the next bot and automating tasks to make our lives more convenient. Al can also be used in the medical field, assisting doctors diagnose patients. Almost all technologies, however, come with both positive and negative effects. With so much power vested in AI, it’s our responsibility as a human race to decide what we do with it.

CRISPR-Cas9

Posted by Pranav Kakhandiki, Edited by Anika Asthana| Jul 18, 2018, 9:00:00 PM

 

        Gene editing, a field in synthetic biology, is a rapidly emerging subject in today’s world. There are a whole world of possibilities, ranging from creating new organisms, to curing diseases previously thought to be un-curable. CRISPR-Cas9 is a riveting new gene editing technology, standing for Clustered Regularly Interspaced Short Palindromic Repeats. The “palindromic repeats” are small pieces of viruses typically found in bacteria. Cas9 is an enzyme which can cut apart DNA allowing scientists to edit the human genome.

        To truly understand when and why CRISPR can be used, it is important to understand how it works. With so many other gene editing tools, such as ZFNs and TALENs, CRISPR is popular for a reason. Unlike other methods, CRISPR does not need to be paired with separate “cleaving” enzymes. In other words, the Cas9 protein uses small RNAs to cleave DNA, as opposed to using separate enzymes. CRISPR is a “knife” that can cut DNA by matching up a guide RNA to the DNA and then cutting it. After the DNA is cut, the cell loosely “glues” the strands of DNA back together.

crispr-cas9-at-work-data-672x361.jpg

        So how does Cas9 enzyme know which part of the DNA to cut out? The CRISPR-Cas9 complex scans the DNA for Protospacer Adjacent Motifs (PAMs), which are short DNA sequences. CRISPR, then attached to the PAM, unzips the double helix. The Cas9 protein then cuts the DNA in two, successfully eliminating the specific piece of DNA.

Slide1.jpg

        Now we know how CRISPR cuts out DNA, but what can we do with it? CRISPR can be used to make tiny genomes in the base pairs that make up the genome, which may end up being the difference between life and death. Many major diseases are caused by one just one mutant base pair in the genome, such as sickle cell anemia, cystic fibrosis, and muscular dystrophy. CRISPR can cut out the “bad” DNA, and insert the correct sequence of base pairs. Other uses for CRISPR include mutating animals, potentially making them stronger and faster than biologically standard.  

        So if CRISPR is so great why don’t all doctors use it? CRISPR is fairly accurate, but can sometimes cut out the wrong strand of DNA, creating its own mutation. So as of present day, CRISPR isn’t quite reliable to be tested on humans. Many years into the future, however, technology will not be the factor holding us back from using synthetic biology.

        With so many possibilities and so much new technology, many ethical dilemmas arise. Should people be informed whether they will inherit a disease fifty years later in their life? Can we create new animals to do work for us? Can we give ordinary humans superpowers with one simple operation? Although CRISPR is a new and exciting technology, one must consider the ethics behind such an enormous jump in scientific advancement.

6a409921-dc87-42c9-add8-c3e4a73bfc80-feature3.jpg

 

        Ever since 1979, synthetic biology has been an emerging field in science,  allowing us to sequence entire genomes and change genes. The ethics of this field, however, can be questioned, as it allows us to potentially create new organisms, or change them beyond what is natural. This subject can be viewed both positively and negatively. One one hand, this practice includes removing environmental contaminants, creating safer and cleaner air, diagnosing and monitoring disease in eukarya, creating enzymes for biofuels, and developing new drugs and vaccines.

        However, a multitude of negative effects present themselves. One harmful effect of the use of synthetic biology is the creation of new organisms. New organisms could potentially destroy the ecosystem, killing off native species and ruining the environment. Another is bioterrorism, the intentional release of biological agents such as viruses, bacteria, or toxins using CRISPR to intentionally create harmful viruses would have an adverse impact on our world, as it provides weapons stronger than ever before. Although both of these detrimental uses of CRISPR can negatively affect our world, the questionable ethics of gene editing doesn’t stop there. New technologies are allowing scientists to map the entire genome of a human, letting them predict what diseases one can inherit. Although it sounds phenomenal on the surface, the potential action creates a personal conundrum. The question is: do we want to know if we will inherit a disease later in life? If we know we will inherit a non-curable disease such as Parkinson’s, will knowing this information cause us to live our lives differently? So despite the numerous benefits of CRISPR, the massive negative potential of such a technology forces humans to decide what to do with the power vested in gene editing technology.

        CRISPR is a fascinating technology with incredible potential. Using it, we can achieve feats thought to be impossible, making significant advances in the field of synthetic biology. With so much potential, however, much harmful potential presents itself. We can cure diseases, yet will the same technology, create a disease even worse. We can create a safer environment, but can it destroy in a heartbeat with the creation new organisms. With so much potential in gene editing, one question stands out: What will we do with it?