Content:

Dr. Liang's lecture on Image Processing was the basic overview of how we can enhance an image  and obtain important information from the image. I am also enrolled to Digital Image Processing (DIP) course and we have to study different techniques to process an image. Filtering, noise reduction, edge detection, segementation are few of them. The interesting thing about image processing, as mentioned by Dr. Kahol) is that it is the applied domain of mathematics. We can apply different mathematical functions and see the results in the image. When we consider an image as a matrix of pixel values, then we can visualize different mathematical models like masking, filters, transforms in an image. In medical field, we can enhance CT, MRI, and/or XRay images using various techniques that would help a doctor to diagnose disease(s). In fact, the way CT, MRI, X-Ray are generated, they are also application of image processing.


Posted by
Prabal

Digital Image Processing

Content:

This week Dr. Liang gave us an overview of the basic knowledge of digital image processing. I take Jimmy’s course: Intro to Digital Image Processing as my selective course. Digital image processing is widely used in researches and our daily life. The digital camera is the result of digital image processing. There is a chip in the camera which can convert the light intensity into voltage and store the information in matrix. Actually, digital image processing deals with data in matrix, so besides the devices for obtain the image information, it is concerned with mathematics. Also, it needs the knowledge of programming, because it is the computer which does the mathematics for image processing. Usually digital processing is used to find the parts of images we are interested in such as the abnormal darkness in the CT image. The project we do for Jimmy’s course is to detect the contours in the image using an active contour model. It is interesting however we do it for a whole semester. The algorithm sounds amazing for me. It is sure that people who describe the algorithms must be intelligent and with high IQ.


Anyway, happy holiday to everybody~


Posted by

Xiaoxiao

digital imgage processing

Content.  Image processing, I think really cool.  I also have great respect for those people in this field, because it is very hard, not only do you have to know a bit about computer programming, you also have to have a great knowledge in math, and probability.  During the classes this week we talked about point processing, filters, and edge detectiontechniques.  Point processing, is applying some kind of algorithm to each point in the image. While filters are algorithms that are applied to a mask.  Edge detection techniques that we Sobel's method, and the Otsu method. 

Posted by P. Ortiz

Image Processing

Content:

Image Processing is very interesting and it brings the math that is performed by imaging programs that we take for granted to the surface. I find that the lectures that was given was also very helpful in that there was plenty of examples as well as equations that helps to illustrate what the processing is doing. In addition, I find that image processing is very similar to signal processing with the filters, manipulations, etc.

Here's a link with basic information on images and how images work in 2D and 3D that I ran across during my research for my undergraduate project.

http://www.ncsu.edu/scivis/lessons/understandingimages/images1.html

Posted by Eric

Good Overview of Digital Image Processing

Content:
Being totally unfamiliar with imaging, Dr.Liang's lectures gave me a very good idea of what exactly digital image processing deals with. The intimate intervention of imaging with numbers and various enumerations to modulate the images are really exciting to study. There is a lot of math underlying this field and a profound mathematical understanding would really help in excelling and gaining good grip in image processing. Dr.Liang gave a good basic introduction to handling images in terms of image manipulation, image reconstruction and image segmentation. Significant research efforts in this field would surely create wonders in Biomedicine.

Posted by
Harsha Undapalli

Image Processing

This lecture was a continutation of past lectures in which we learned about the different types of resolution.  Here, we looked at some specific mathematical and statistical properties of image data.  In image processing, an image can be represented as matrix values.  Because of this, it is easy to pinpoint certain data values.  Various subsets of image processing were discussed, such as data interpolation, image restoration, and image segmentation.  The types of digital images include binary, grayscale, true color, and indexed color.  Also, basic statistical and manipulation methods were introduced that could help describe and change imaging properties.  Histogram equalization was one statistical method.  Histograms are a way of displaying the distribution of image data.  Also, various functions were described that could change the appearance of the image.

Listed below is a very interesting, informative, yet easy to understand series of lectures on digital image processing.  Images were manipulated using MATLAB for class assignments.  It made me want to take pictures and do assignments! :)

http://eeweb.poly.edu/~onur/lectures/lectures.html

Posted by Annie

Imaging

Content: Hi Guys. Hope everyone had a nice Thanksgiving. I know that this is a new and confusing holiday for some, but I am sure that you have unique holidays that I would not understand. Jimmy really knows his stuff, but this is complex. For images, you acquire, preprocess, enhance, restore, segment, represent descriptively and recognize and interpret. He has focused on image enhancement, restoration and segmentation. The enhancement makes the image more suitable for applications and improves sharpness and contrast. The components here are point processing and neighborhood processing. Point processing utilizes arithmetic operations. Averaging and non-linear median filters can be used for the specific function needed. Geometric functions can also be used for enhancement and rotation is one type of this method. Image restoration involves the removal or reduction of degradation. For Noise, includes gaussian, salt and pepper, speckle, and periodic and is these must are ways to address this. The median filter is advantageous for removal of the salt and pepper noise. Gaussian noise approaches utilize a composite image from a sequence of images. Image segmentation is focused on the separation of the image into component parts. Thresholding is used here. Edge detection is another aspect of the segmentation process. Specific filters including Sobel, Canny edge, and Roberts may be used for this. Edges may be sharp or ramped, requiring different approaches. I will continue to evaluate the mathematical functions that are involved here. So much of brain imaging depends on these approaches.

Posted by Stuart

Image Processing

Content: The lecture was quite intense going into the methods used in digital image processing. Surprising it is simple matrix algebra which can do such amazing manipulations. Without the slides its difficult to write a lot, but of what I remember were point and neighborhood processing and use of various filters to correct the images. With my very limited interest in this field I can only contribute a tutorial site for all those interested. I think this is a good site which introduces to the basics of the field.
http://www.library.cornell.edu/preservation/tutorial/toc.html
Well that was my 2 cents as I don't have and don't think I ever will have a very good understanding of this field.

Posted by

Sheetal Shetty

Digital Imaging Processing

Happy Thanksgiving!
Didn't Debbie say it all?  For those of us not enrolled in the Imaging Class, the information presented, was challenging. I think Jimmy gave a good introduction, stressing that it was only introductory, but the concepts were difficult.  I have used Jasc and Photo Shop so now have a better understanding that a mask is a "piece" of the entire image, and that we manipulate for imagery, using consequtive masks. I won't pretend to interpret more. I'm waiting for the slides!
Also, as with the review, Kanav has a way of mandating learning, so his summary of Monday's lecture was appreciated.
Lee

Week of 11/23/09

Content:  This week Dr. Liang gave us lectures on imaging.  Also, Dr. Kahol gave us a lecture that included image filtering.  Filtering is commonly applied using masks in an image.  One of the topics included in the imaging lectures was image interpolation.  Interpolation can be used for functions such as scaling or rotating an image.  Another topic covered was image restoration.  Image restoration can be preformed by applying filters that adjust image intensity histograms or generally lighten or darken an image.  Image restoration can also be applied by using filters that sharpen or smooth an image.  Additionally, segmentation was a topic that was covered.  Segmentation can be preformed using filters such as edge detection filters.

Dr. Liang did some nice work organizing and summarizing the topics that he covered.  I also thought Dr. Kahol did a nice job expaining some filtering topics.  I liked getting to learn about the imaging topics that were covered this week.  The filters covered are so widely used that it is interesting to know more about how they work.  It was interesting to learn about the segmentation filters.  I can imagine how those filters are practical for a large number of purposes like object detection.

Here is a link to a open source program that is like photoshop.  If anyone doesn't have photoshop and wants to try image filters they could use the program at the link below.
http://www.gimp.org/
It seems like GIMP has a lot of functionality that is similar to photoshop.

Posted by:  Nate

Image Processing

Content: The image processing lectures by Jimmy were very interesting to me because, 1st of all, it involves math (rather than boring slides with only text!) and 2nd of all, it shows the math implementations in image form. As Dr. Kahol says, image processing is the best implementation of math as the result of the math can be observed in the image. We learnt the implementation of different filter on images to make it sharp or blur. Image processing is a very promising field in biomedical informatics. In fact, being a signal processing guy, the 1st motivation that I have got for switching to BMI is to do research on image/video processing.

Posted by Gazi

Research Validation

Content: Dr. Petiti’s lecture was about validating any research method. The outcomes from any research can come in different shape and terms. It can be measured in terms of mean, risk/odd ratio or hazard ratio. A research study is valid as long as it finds the truths. There are 2 types of validity – internal and external. Internal validity researches with the design method of the study and the measurement analysis, where the latter validates the difference in applicability to person to person or place to place. Also the lecture discusses about different types of errors in research method, like insufficient or not properly randomized samples.

Posted by Gazi

Surgical Simulators

Content: The lecture on surgical simulator was the most interesting lecture so far, as it relates to the research work we do in the lab. We are developing “active world” which is a whole new animated parallel world where different patients have been created with different scenarios. Residents and doctors can hone their cognitive skills by going through these scenarios in the animated world. There are at home simulators like drilling simulator, FLS or virtual peg transfer game. Surgeons can practice it offline even from home to improve their surgical skill. The peg transfer simulator or wii games like marble mania are actually being used at the patient’s bedside for warming-up before the surgery. We are also developing simulator that can measure effect of fatigue at different time of the shift.

Posted by Gazi

Imaging Manipulation by Dr. Jianming Liang

The imaging manipulation presented in both lectures this week were interesting but rather hard to follow, especially the functions and equations for the various types of filters and masks.  They all seemed to blend together.  Although I love math, trying to follow the math examples for each type of manipulation was near impossible.  I understanding the masking, average, minimum and maximum filters, etc but fail to see when these would be used in clinical practice.  Obviously the filters and masks are needed to help clarify an image but I don't understand how you determine which one to use and when.  I appreciated the simplistic approach of using a photograph familiar to everyone instead of using actual medical imaging pictures. We could clearly see the results of each type of manipulation which was helpful but I don't understand when you would use a 3x3 versus a 5x5 mask, etc.. 

In addition, the slides are not posted for review.  Will these be posted soon?

Posted by Debbie Carter

Dr. Petiti's lecture on designing for validity was another interesting topic on research design methods. She first described outcomes of a research design and then explained the measures that affects the outcome(s). The main focus of the was on understanding different types of study design and the factors that affect the study. As an example, she mentioned the study on breast cancer self examination, which was pretty interesting. The other examples that she mentioned were 'cellphones as the cause of brain cancer', and 'drinking as the cause of coronary heart disease'. The first example can't be done until the participants are forced to do BSE, and different law enforcement should be required. In all these cases randomization is required. But, we should also consider the risk of technology being changed in the future. That would definitely affect the study.

Prabal

Research Validity

Content:

Dr. Petitti gave a great lecture on research validity where she focused on things that are major threats to research validity. These threats are as follows:

• bias in design
• measurement error
• type 2 statistical error
• poor study conduct

These can all lead to studies that are not valid or credible. Also, it was mentioned that typically study sizes are very important yet we see studies being published with very low numbers that have questionable conclusions.

On a side note, a recent hack attack on Climate Change specialists' emails have showed that results may not be what it seemed when it was presented. As well, I believe one of the articles covering the story mentioned that there was some plotting by these groups to prevent the release of data/Information as required the Freedom of Information Act. I found these articles interesting since it was published just days after our lecture on Research Validity. Also on a side note, the summary above for the articles may not be accurate since the authenticity of the documents have not been verified as authentic, but here are the articles:

• Wall Street Journal:
  http://online.wsj.com/article/SB125883405294859215.html?mod=googlenews_wsj
• New York Times:
  http://www.nytimes.com/2009/11/21/science/earth/21climate.html?_r=1

Posted by Eric

Research Validity

Dr. Petitti brings some real world experiences into the classroom.  We had the opportunity to discuss the news interview on the latest mammogram recommendations coming from the task force where she is the vice chair.  It's very interesting to hear the different perspectives and how people hear things differently and the media tries to twist the truth or make you stumble.  Nonetheless, she did a good job on the news interview I saw.  Then, the next day, she shows up in our class to talk about research validity.  Of all people to question if research is valid or not, I think Dr. Petitti has a lot of experience with evidence based medicine and has certainlhy shared her knowledge on the validity of research.  This lecture was good but I felt it needed a lot more explanation, especially when talking about the odds ratio and the risk ratio.  It looks very similar to the sensitivity and the specificity tables but not quite the same.  I guess in order to better understand, we should take her biostatistics course.  The confusion for me was determining the sample size and coming up with the 4 factors to plug into a power analysis to recommend a valid sample size.  I'm not sure if the 4 factors are made up or if these are coming from some other type of calculation.  I'd like to see a little more in regards to the points she reviewed as the information shared felt a little rushed and we were unable to practice with examples.


Posted by Debbie Carter

Study Design

Content:
 Dr. Petiti's class on Reaserch Methods was another very useful and  insightful  lecture. This time she discussed about experimental studies and also introduced the measurements that are used for experimental validity like the odds  ratio, hazard ratio and risk  ratio. She mentioned a very good point that the basic step towards desinging a study is to understand the types of study desingns and know about the factors and biasness that can influence the study. She ta lked about study of  occurence of breast cancer with people taught about  BSE  and how sample size is very important factor in a study and how can a small sample size can get to a wrong conjecture.In another study of Coronary Heart Disease with drinking or non-drinking the experimental validity measurements were considered like the risk ratio and how the occurence of risk ratio of one event can be the  inverse of risk ratio of occurence of another event.In the study of occurence of cancer in people using cellphones she told about the factors which could bring a change in  the final outcome of results like the time taken to develope cancer and also the technological advancements during that time which could make the whole study unimportant. I would like to thank Nate and Sheetal for providing link to the articles. They are really good .

Posted by
Ashutosh Singraur

AMIA 09

Content: I was very fortunate to also be able to attend AMIA in San Fransisco.  There were many presentation and talks about a wide range of subjects in Biomedical Informatics.  I was most struck by the a presentation that Dr. Shotliffe gave when he was assessing the past, present and future of BMIE.  He basically gave the same presentation that he gave to us in the beginning of the semester in 501.  I thought it was incredible that we were able to hear this lecture before in a more private setting.

I tried to attend as many panel sessions as possible.  Topics in these panel sessions included, global health, BMI in the industry,  entreperneuship, and standards. One of the key note speakers was Mark D. Smith.  He is the CEO and president of the California Healthcare Foundation.  He have a really nice presentation about infomatics in his company.  He re-emphasis that he was struck by high levels of today's technology, yet the healthcare industry has not taken full advantage.  An example he gave was the fact that when nurse takes the patient's temperature with a digital termometer, and the nurse still has to write down the temperature either on paper, or in the computer.  Instead the thermoter should be connected to the computer and it should be automatically recorded.

The some panels were very interested and some needed a bit more structure.  The panel about global health was amazing, they talked about how biomedical informatics is being implemented in more poverty stricken countries.  The panel about Informatics roles in industry was lacking a bit, or I just had different expectations from it.  The panel was composed of people from industry and they just talked about their company and what the company was doing.  The description for the this panel mentioned that it was going to be about how to get into industry.  This is the first year AMIA had a panel like this one, maybe in the upcoming years, they will improve.  I also attented a panel on standards, yes standards.  The panel consisted of C. Chute from the Mayo Clinic.  I was told by various of my peers, especially Mithra, that he gave a lecture about standards in past years and he made it very exiting. I was a bit dissapointed on this panel too, because Chute only talked about BRIDG infastructure and not really about standards.  It was good expusure to BRIDG though, I can't complain too much.  I also went to a panel on enterprenuaship in BMI, again, I think I had different expectations about this panel, though it was nice to get a few names of upcoming companies in BMI. 

I also attended the student paper competition, which I thought was the best part of the whole conference, I really learned a lot from these presentations. I also attented most of the paper presentation from our department and I would like to congratulate Mithra for a great presentation!

Dr. Shortliffe mentioned this article in one of his presentations, I thought it might be interesting to read

http://www.time.com/time/health/article/0,8599,1883002,00.html


Posted by p. ortiz

Valid Research

Dr. Petitti gave us more insight into conducting meaningful research.  This lecture focused more on the areas that could prove to be problematic throughout conducting research and interpreting results.  One area that could pose problems was the initial choice of the right study design.  This is probably the most difficult and time consuming part of the research process.  Once the right design is chosen, there are still many areas that could threaten the validity of the design, such as incorrect measurement of outcomes and lack of application to a wider variety of situations.  If the study is only "perfect" within its own design, its serves no value for the rest of the research community.  With these issues in mind, it is more likely that you are prone to conduct more valid and valuable research.

Posted by Annie