Any image from a CT that is converted to JPEG format cannot be used reliably to read out the attenuation in Houndsfield units ever!

That image format uses image compression and other image proceeing NOT DESIGNED to preserve such properties. DICOM lets you save JPG files through a form of encapsulation, so it can administratively be handled correctly and safely. That is, the image data will be registered with meta data such as what patient is involved, when and where was the study performed and so on.

JPG files are only intended for screen-captures / medical photographs, not to store CT's.

Any modern CT can, at least for the past 15 years, produce a DICOM file format that preserves all image properties directly. Only those images preserve the correct values.

The DICOM standard defines a lot of different file formats for different purposes. Dicom encapsulated JPG is totally unsuitable for the storage of native CT data.

But even if you have the original  DICOM CT data available, you must wonder whether the unmodified CT data actually are suitable. The Houndsfield unit has serious limitations. Therefore in radiotherapy al lot of effort must be put into calibration and transformation into electron-density maps.

Houndsfield units are NOT equivalent to optical density .

Therefore: Please state the purpose of your project?

image J can be fed any garbage and be made to produce output.

Any use of JPG as an intermediate format between a calibrated medical image format renders the result useless for purposes of measurement of attenuation.

In addition the resulting JPG image is not suitable [while not designed for] further post-processing, as even [small] feature sizes are not guaranteed to remain conserverd.

It is not that difficult to load DICOM CT data directly and several such software packages are available in the public domain.

If you are confused by some of my statements, ask any clinical physicist on a radiotherapy department for advice. They take such details very, very seriously for good reasons.

mr Judocus Borm thank you for  a great information.Actually, now i know why the results not acceptable . my research in dose reduction in abdominal ct protocol by change parameters such as KVP and mas . my supervisor asked me to collect image dicom format with different parameters and do measurements by image j program . now how can collect image of abdominal ct in dicom format (not jpg) from internet ???

I tested this at home: Downloaded imageJ for debian Linux. It ran straight from the box and the examples were downloaded from the NIH website. See the NIH imageJ documentation at http://imagej.nih.gov there choose imageJ.

What you need is files with the extension .dcm or dcm.zip .

The only other formats that would preserve all required metadata is .tif or .tif.zip

Any patient disk with CT data is nowadays likely to contain a file dicomdir.dir summing up all dicom files on that disk [all with extension .dcm . 

A few years back such disks were loaded with a custom image viewer, but nowadays most come with clean dicom files.

If you have a friend working at a radiology department he/she can supply you with clean anonymized data in the same format,

With regard to the warning on lossy formats such as JPG, the imagej on-line documentation at http://imagej.nih.gov/ij/docs/guide/146-7.html#toc-Section-7 explicitely states:  Lossy compression Transcoding an image into a format that uses lossy compression will alter the original data, introducing artifacts (see 11: Warning on JPEG Compression↓). This is the case, e.g., for JPEG formats (with the exception of some JPEG2000 images that use lossless compression). As such, these types of data are intended for human interpretation only and are not suitable for quantitative analyses

My other warning stands:

On different CT scanners the spectral content of the CT beam IS different even when the tube voltage is the same, say 100 kV. So 30 Houndsfield units on CT brand A does not equal 30 Houndsfield units on brand B. They come close, but not that close.

Full details of this latter effect can be supplied by any clinical physicist working in radiation therapy. Do not ask physicians: this is far too subtle for > 98% of them. Here even radiation therapists take it as a fact of life and "that is what we have a physicist for...." and I doubt it is any better elsewhere.

If you only have to measure physical sizes you can directly use the data. If you actually want to give meaning to the attenuation data please consider my remark carefully.

With dose reduction you can probably get away ignoring such details, but in radiotherapy you can not.

There each CT [and specific settings] is calibrated and Houndsfield units are converted to calibrated electron densities.

Do not ask me more, as I am not a physicist in radiation therapy.

If you really want to know more abouthtese things, the IAEA has at least two suitable publications on their website on these matters [www.iaea.org].

I hope this will get you going.

you can watch this video and learn how to set scale in microscopic photos

https://m.youtube.com/watch?v=M8_mdtXocHE