Brain MRI Tumor questions

In the project Brain MRI Tumor Detection and Classification i cannot understand what the values in the meas(xdata) represend.I mean why a set (x,y) is classified as BENIGN for example.

I cannot also understand how classification here works..in this line of code

    species = svmclassify(svmStruct1,feat,'showplot',false)

why dont we put img2 insted of 'feat'??

Answer 1

[Walter Roberson]

The values in meas(xdata) represent some kind of measured data. SVM does not know or care how the measurements were taken or what they actually are. As far as SVM is concerned, it does not matter whether some parts are (say) temperature and other parts are (say) x-ray pixel intensity, and other parts are (say) acceleration along a particular axis.

"I mean why a set (x,y) is classified as BENIGN for example."

It just is. SVM does not care about cause and effect; at most it cares about correlation. If, for example, one of the readings were I2Sb2F11 concentration and another of the readings were H2S03, and high readings of I2Sb2F11 occurred in the samples marked class Tumor Grade 1, then SVM does not care whether I2Sb2F11 causes Tumor, or Tumor causes I2Sb2F11, or whether the H2S03 was reacting with the probe leads and leading to contamination that the healthy cells rejected but the Tumor cells were not able to flush.

The data you input for SVM does not in itself have an explanatory power: you just take a bunch of measurements of things that might be relevant, somehow, and you have some kind of external judgement about the class of the training samples, and SVM tries to figure out what the important correlations are. Once correlations are identified, then the researchers can go back and write up another grant to try to model causation.

[DIMITRIOS THEODOROPOULOS]

Thank you for your time.I just wanted an example for the kind of data which lead to classification of tumors.I was thinking,perhaps x would be intensity or y perhaps diameter or a tumor??

The second part i think i did not explain well....We put an image an the SVM classifies it as BENIGN or not...Then why we just put this image on SVM and instead we put its analysis?? Sorry if ididn.t understand well your explanation in second part

[Walter Roberson]

For SVM, you can use either raw data or features.

You should be asking yourself questions:

• are there significant differences in the time it takes to program and debug using raw data compare to extracting features?
• are there significant differences in the execution time?
• are there significant differences in the ability to use the classification to potentially explain the results?
• after training and testing and arriving at a classification system, are there significant differences in the potential for simplifying the processing for a second version of the program? For example it could turn out that 90% of the correlation came from the center 1/3 x 1/3 block and that you could perhaps extract just that center part of the input and then execute more quickly because you had less data to work with