CSV files

CSV files, meaning Comma Separated Values, is a standard format for inputting data to a spread sheet program as in Open Office, Libre Office, Microsoft Office and Google writer.

Brown fat and other ROIs are normally used over patient population and statistically compared. Thus the spread sheet offers a convenient method to manipulate data.

In many cases several physicians will be working on the population so it is very desirable to have the results stored on a shared directory over a LAN, some place which is central and accessible to everyone. When the Save button is pressed, you may browse to the chosen location and save the file in that location. Henceforth both the Save and Load buttons will default to the defined location. (Obviously if you change the location, the default will be changed.)

There are 2 distinct parts to the csv file: the results and the ROI definitions. The results are used for the statistical comparisons. The ROI definitions are used by the Load button. For example it is possible to use the previous ROIs on a follow up study of the same patient.

The ROIs may be loaded automatically using the Auto start in Options. These csv files are sufficiently important that their existence is color coded in Read BI studies and Read from CD. The color coding allows for immediate recognition of which patients have been processed and which not.

Results

A typical set of results is shown in Libre Office Calc

Column A has the label of the ROI. It is highly advised to use a label from the combo box list of labels as it will help to identify the ROI when you are looking at the spreadsheet when the ROI is no longer visible.

Column B is the ROI number which in our example runs from 1 to 13. Column C is the type, which in this case is 2 for a ROI drawn on a coronal view. You list of ROIs can be a mixture of ROI drawn on the axial, coronal and sagittal views.

Column D may be the most important column. On each line is the volume of the given ROI in ml. However on line 15 you can see coronal as the label and sum as the ROI type. This line contains the sum of all values. It is not a simple sum but rather a sum where overlapping pixels are counted only once. Thus the volume on line 15 is less than or equal to the sum of all the numbers above it. This same avoidance of dual counting of overlapping pixels is applied to all the following results as well.

Column E is the Volume*Mean, which is column D * column F. If you take a calculator and multiply the values in these columns you may see a slightly different value. This is because the values in all columns are shown rounded to a given accuracy whereas the value in column E was calculated before rounding.

Column F and G are the SUVMean and its standard deviation. Columns H to M are various measurements of the max. The most common is SUVMax in column L. Next to it in column M is the MeanMax, which gives an idea of how accurate the SUVMax might be. The MeanMax is done by calculating the SUVMax in each slice separately and then taking the mean of those maximum values. This is the value reported in parentheses in the Brown fat results table.

Columns H and I are the SUVPeak calculate via Wahl and the mean of its slice values. Columns J and K are SUVqPeak and its mean slice values calculated by qPET - a quantitative extension of the Deauville scale....

Note: The most significant line may be line 15. Column A says coronal which tells which way the data is being displayed. For manual ROIs a set of ROIs is specific to a given orientation on which the ROIs were drawn. Column B says "sum" which is the result which is shown if Show All is checked. Column D shows the numeric value of the sum which is not necessarily the result of summing all the values in Column D above this value. Each individual value is the result for that particular ROI, whereas in line 15 any overlap has been eliminated giving the result documented in the TMTV result.

ROIs

On line 18 is reported Number of ROIs = 13. These are the binary values of the ROI defined points. This section is not meant to be used by the human. It is meant to be read directly from the csv file, to read back in the previously defined ROIs.

While these values are visible you probably just want to ignore them, especially when you are doing copy-paste from many spreadsheet files into an overall concentration of all results.

CT values

if the CT condition is used, Hounsfield Units (HU) and their standard deviations will be shown. The CT values are used as part of the filter to determine if a given point is to be accepted or ignored. The point must pass both the SUV and CT tests to be accepted. What is shown in the table is the average value with its standard deviation of those CT values which have passed both tests. This is not the same value which will be shown on the image when you click on some particular point. The click on the image will include all points in the given radius without any regard to any SUV or CT limitations.

ROI details

What is the difference between Nifti ROIs and manual ROIs? The manual ROIs are a list of vertices in a given orientation (axial, coronal or sagittal). Since the orientation specifies a plane, X,Y pairs define the vertices. The accepted points are not stored.

or, without SUV being checked

Nifti ROIs are a list of X,Y,Z coordinates which satisfy the internal conditions of the auto segmentation protocol. Checking SUV will add an addition condition to be satisfied, thus possibly eliminating some of the points.

Return to help.