Persistent Problems

07:10 am

Yesterday, at around 2:45 pm, Nilkath called up with a news that I was dreading. He said, the audio  from the ultrasound Doppler is not recordable. He tried it, but only images can be stored. Not even video recording is possible.

Solutions to this:

1. I remember reading somewhere that 15 to 45 secs of .avi clips can be stored in the ultrasound machine. Possibility is that, may be the model that I read about is different. So, I messaged to ask Nilkanth the model number. He said it was Phlips HD11XE. Read throgh all the documentation possible about the data management capabilities of this model. Couldn't find any mention of audio or video storage capability.

2. At this instant I remember my External guide telling me something to the effect, "What if, by using some kind of recording device, you could capture the sound coming from the ultrasound machine, and process it to diagnose Placental Dysfunction?" Now I realise and understand why he said that. So, going by his way, I could buy a digital recorder to record the audio output from the ultrasound machine. Two ways it could be done,

• Hold the recorder in front of the speaker. Need a digital recorder not a tape recorder. I donno how that works. Gotta find out.
• If audio output ports exists, gotta find out if any digital audio recorder can be connected to these ports.

In any case, I will have to be there in the ultrasound lab at all times to record the output by myself. Can't ask Nilkanth to do it.

3. If the above two solutions get striked off, I will have to go for a Doppler ultrasound machine, and ask Nilkanth to do the same procedure that he does with the ultrasound machine with my hand-held machine. For this both the thing I dislike will have to be done. 1. Ask him to do it, 2. I have to be there.

First step, straight into the puddle. :(

ATB.

A Visit in Vain

12:45 pm

Returned a half an hour back from what I may call a futile visit to Goa Medical College. Shouldn't have had such high hopes about today. :( Was hoping to stay there, see patients, and whichever patient does the Doppler that I need, tell Nilkanth to copy only that data, and burn my CD. Turned out, I went at the wrong time. It was very crowded and everyone was very busy. I didn't realize that I was imposing, untill Nilkanth told me to leave the CDs with him. So, will be going there next to collect the CDs when he informs me that it's done. Have informed him that I need the audio clip of MCA and UAD for IUGR. Hope he remembers to  do it. Not having a good feeling about this :( Let's hope for the best..

The only good thing that happened is that I got time to prepare for the remaining work. Cousin sister Sneha Nargundkar informed that the reviews will surely be on this weekend, which gives me only 2 days to be prepared with the ppt. Have prepared the outline. Will be doing the first two bullets today. I think this top isn't lucky for me.. 2nd instance of something bad happening when I've worn this top. Hmm..

Anyway, ATB!

06:30 pm

Editing this post.. Urgently.. Because I just stumbled upon a very important fact and a consideration that I should make right at the beginning of project. That of gestational age! The placental or fetal conditions vary during the period of maternity. The diagnosis too, thus, varies. In my Signal Extraction system, there exists a neural network component which learns the data fed to it to get trained to interpret.

Firstly, I will have to decide on a particular range of gestational age for which my system will work. Secondly, due to this fact, the system will have to be trained with the data belonging to this gestational age! Now, depending on this, to come up with a range, I have to either go with gestational age of majority data  that Nilkanth gives me, OR I can find a paper on the net which will mention which is the best gestational age at which Placental Insufficiency can be detected, OR I can find a paper on the net which will mention which is the most common gestational age at which either Umbilical Artery or Middle Cerebral Artery Doppler is performed. Freeks.. Gotta make this very important decision, right away!

ATB!

The Data Declaration

05:20 pm

Just returned after a visit to GMC. Dr. Jeevan Vernekar is so kind. He is the best of the rare breed doctors that I was talking about yesterday. And Nilkath, the Ultrasound Expert was just awesome! Totally willing to help, and there, in case I need any help. Got his number! He showed me the entire 'Umbilical Artery Doppler' and 'Middle Cerebral Artery Doppler' procedure, the audio data of which I need. He's called me tomorrow to the hospital, to take the first bits of data home! :) Yeay!! Everything that I wanted is here now!!

Nilkanth told me that the audio data is recordable, and can be written on to a CD. I will have to see exactly what format the data is recorded, and whether it is compatible to be used with MATLAB. He also told me that it wasn't necessary to get an official letter or anything, but, just to be careful, I'm thinking, I'll talk to the our ETC HOD about this, get a letter ready, show it to Dr. Mahesh Sardessai, the HOD of that GMC Radiology Department and Get Going!! Truly happy!!

Am thinking of giving the data acquisition a maximum of 2 weeks from now. So, the day that I have to start processing the data is 24th of September.

• Gotta go to GMC tomorrow by 10:30 am with a CD. Yeay!!

ATB!

Recognizing Patterns, Thank you Dr. Robi Polikar

12:20 pm

Started the 2- way database search. The one that happens sooner, shall be pursued. Compiled the Workflow study for Placental Insufficiency, mailed our External Guide and asked him for help regarding the database, which he had initially promised us. But looking at his long distance interest in the subject matter, I highly doubt his involvement. In an attempt in desperation to search for his alternate email ID, found that he has a Facebook account, and, surprise, surprise, a Twitter account! Followed him on Twitter, and sent him a friends request on Facebook. He accepted it. So he knows I exist, and doesn't forget my name.

The other direction that I'm desperately pursuing is my own search for a good Ultrasound specialist in and around Panaji or Ponda, who might be gentle and understanding and encouraging. A difficult species to find indeed. The two ultrasound specialists I've seen till now are:

1. Dr. Sanjay Sardesai: He's been really really helpful in what I asked him at the time. Didn't get an idea of asking him for the database at the time. He's been really encouraging,.. May God Bless him!

2. Dr. Rajaram Mahambrey : The worst ultrasound specialist,.. I don't think he had any idea of what I was saying, nor was he ready to listen to my idea. All that I heard him say was "No,.. I don't think that can be done...", "No,.. That's not possible..", "No, I don't think that's a good idea". When he heard me for just 1 minutes, he said oh,.. is it? That's great! For every "No" of his, I had to prove him wrong, and then he would change his word to yes. How can you trust or even feel like working with such a doctor? Horrible experience all in all. No help whatsoever.

3. Dr. Jeevan Vernekar: Looking forward to meeting him. Got his name from my uncle, Dr. Sharatchandra Nargundkar, and his contact with the help from my, batch-mate Rohit Naik. The great coincidence which connects all the three meetings with the ultrasound specialists is that, Dr. Sanjay Sardessai works in GMC, so does Dr. Jeevan Vernekar, and GMC is the place Dr. Rajaram Mahambrey told me to visit in order to get the database. So, really hoping that I'll get help from this good man! ATB.

Now, I'm guessing, that in the worst case scenario, as Dr. Mahambrey told me, I'll have to get a letter from GEC, HOD, Dr. R. B. Lohani, which can allow me to use the GMC database. Have to be prepared for this I guess. Hope our HOD helps in this regard.

• Gotta prepare the letter for permission to access the database after the meeting with Dr. Jeevan Vernekar.

Today I decided to work a bit with MATLAB, in an attempt to stay on track with the first schedule I ever made. Learnt that signal processing toolbox and wavelet toolbox already exists in the MATLAB version I have on my Dell. Then, learnt to import and play a .wav file, and plot its Power Spectral Density, fft. The Wavelet tutorial documentation provided in the MATLAB help files is nice too. Started reading it. 

• Gotta continue.

Thought about the 'Pattern Recognition' pdf that I'd downloaded by Dr. Robi Polikar. This man is a magician really. So nice is his way of explaining things. Since I'm going to be trying to recognize patterns in the abnormal data, and train the neural network and stuff, realized I have to go through this and memntion it as 'Literature survey' in my Review I. Started reading it.

• Gotta continue.

ATB!

Obstetrics with an Open Mind

09:09pm (while watching KBC)

Visited another ultrasound clinic today. That of Dr. Rajaram Mahambrey. Surprised with his lack of knowledge about ultrasound, and beaming negativity... A futile meeting I must say, with respect to the knowledge I was hoping to get. Though, realized a few things, after watching a few project presentations in college, and in retrospect.

1. I need a data base. But since I will be trying to find the frequency pattern for a particular abnormality, and then make sure that, THAT is what I have to look for in a signal to detect the abnormality, I need an abnormality database of size 50 at the least.

2. Since I need to prove that the database is in fact the abnormality database, I will have to have it's image output with it. So, the database should be an audio + a video or image database.

3. Private doctors, don't know anything,.... ANYTHING!!!! So, gotta contact the government doctors, ask if any official letter is necessary to obtain the database. That will be troublesome!! The other alternative is, contact the External Guide for database. He needs the workflow!! Damn it!! Gotta start working both ways. Have the workflow ready. Will mail it to the External Guide, brief him with planned activities, ask for his opinion and ask for the database. Hoping Anisha Cotta's project presentation, for which he is the external is soon, so that we get to meet him, and ask him as much information as we need. But, gotta mail him first. That's the 1st TO DO thing tomorrow. Get the workflow, and mail.

ATB!

Searching and Sorting

09:54 pm

Got my hands on Yadnesh Sanzgiri's Paper. Went through. Pretty cool. After spending at least a week since the beginning of my project, on trying to search for any information regarding any ultrasound doppler signal database, or their classification, I have realised that they do not exist on the net. Not even on physionet. A first concern!! Because a lot  depends on it. Got to think of alternatives.

Have to meet the ultrasound specialist who Rohan Naik from micro was talking about! It seems he's free most of the times. We'll see if he's got anything of my concern. Either information about how diagnosis of Placental Insufficiency is done based on the audio doppler, or the database.Can I do without it? I guess NOT.

Coming to the review Presentation, guess the sequence to progress would be this:

• What is Placenta? It's function and importance. What is Placenta Insufficiency? Its ffects. Diagnosis workflow in short, emphasis on Umbilical Artery Doppler and middle cerebral artery doppler (Image).
• Discussion of audio doppler. Comparison of information content in Audio and Image. Thus importance of Audio diagnosis. But not much research, very less papers available.Thus the aim of the paper.
• Summary of 2 papers that I've found on the subject.
• What is doppler effect? Application in our case of  Biomedical Diagnosis(UAD and MCA). Explanation in short how Image and audio sounds are produced (have a paper + textbook xerox for explanation).
• Planned signal Processing using MATLAB and Neural Network: Looking through Principal Component Analysis, Independent Component Analysis, Wavelet and variations available with combinations of these. Their comparison to find out which would be the appropriate one.
• Basic study of neural network.

Preparation for all this will take at least 7 days + 2days apprx 10 days. Assuming week after Ganesh Chaturthi are the reviews, I have just about 14 days in hand. Gotta make each day count.

1. Gotta catch hold of that Ponda Ultrasound Specialist and find information about audio indications of Placental Insufficiency, and see if he can give me some database.

2. Gotta start getting the presentation ready, because then I can just take excerpts from this and make the report for submission at the time of the review. Considering the number of days at hand and bullets as mentioned, each bullet can afford around 2 to 3 days. Starting from the day after tomorrow because tomorrow is my father's 66th birthday, gotta kinda celebrate type of a thing.

Nothing else on my mind right now. The way our external guide told us to go ahead with the project makes no sense considering the direction of the project, and doing it just for the sake of it is not the luxury I can afford to take considering the time and effort required to do it first according to him, and then redo it according to me. Have to put my foot down! Here we go! Sorted!

ATB

New Possibilities, Independent Component Analysis

08:30 pm

Was away from my laptop for a period of 11 days, a reason due to which I couldn't dive into my project work in full swing. Got my laptop back today, and have promised to take it's good care,.. at least for a year, till my project gets completed. The replaced part are, Keyboard : 3 months warranty, Motherboard: 6 months warranty, battery: 1 year warranty!

Getting back right on. Was struck with 2 new possibilities of DSP stage of the project, the Principle Component Analysis (PCA), and Independent Component Analysis (ICA). Nothing tough, only have to find papers comparing all of these DSP methods, and have to resolve to the best one. If I find these papers, will be able to add to the literature survey section too.

Both ICA and PCA have a lot of papers written on them, both work in MATLAB. Toolboxes for both are available. This one is for ICA.

Deadlines and checkpoints mentioned in previous posts have been achieved. The new to do:

1. Study the photo material Chapter 4 brought from Dr. Sanjay Sardesai.
2. Form the mail to be sent to the External Guide which will have the Project flow as he had mentioned. Mail and ask   for database.
3. Form the Project Review document.
4. Dive into MATLAB signal processing, try out ICALAB, Wavelet Transforms.
5. Search for papers that compare Digital Signal Processing Techniques to choose the best one and to be used as literature survey papers.

Here is a good page that will give you the source for anything related to ICA.

Diving into MATLAB now! Yeay! ATB