Part 2 - complex programs
1. Introduction - basic properties and structure of complex programs
1.1. Concept of complex programs
Complex programs are used for automatic evaluation of scintigraphic data, especially dynamic and functional studies. Complex programs consist of a sequence of algorithms for image and curve processing according to the relevant mathematical models of the studied processes. They contain commands for all operations and manipulations resulting from the methodology, including the presentation of graphical and alphanumeric results on the display and on the printer.
The basic concept of
complex programs for the evaluation of scintigraphic studies is
based on the principles already mentioned in the description of
the basal system OSTNUCLINE. These principles are the maximum complexity , exactness , generality
and flexibility of evaluation of scintigraphic studies, with
the greatest possible elegance, unpretentiousness and ease for
the operator. *)
*) The mentioned concept and the specific
algorithms used for mathematical processing are based on many
years of research and development work in the field of functional
scintigraphy and physical-mathematical methods at the Department
of Nuclear Medicine, University Hospital in Ostrava. They are
tested both in experimental and phantom measurements, as well as
in the intensive clinical practice of the author's workplace and
a number of other workplaces.
One of the characteristic features of the complex programs of the OSTNUCLINE system is, on the one hand, maximum automation , but on the other hand, the possibility of control and manual intervention of the operator (these are not "black box" type programs). We follow the principle: everything that is possible, the program does it using mathematical algorithms automatically, but under control - the result is offered for approval or modification.The mentioned automation mainly concerns the automatic marking of areas of interest on scintigraphic images and the automatic delimitation of significant points and sections on curves. If we have a well-tuned algorithms of this kind are being evaluated for the majority of tests automatically, but in the anomalous cases, we can modify the automatic procedure for the purpose of m achieve the correct result. This approach respects the variable nature of biological data .
The typical activity of our programs for complex mathematical evaluation of functional scintigraphic studies is roughly as follows:
In the first phase , the display shows appropriately summed sequential images representing a kind of "inspection" of the entire study. They are offered for preliminary visual evaluation (either automatically generated or manually entered text) and for photographic or printed documentation . Then follows the definition of areas of interest in the images (either manual or automatic) and the construction of time curves of radioactivity in these areas.
The mathematical analysis of these curves then calculates the quantitative parameters of the dynamics of the investigated process (such as ejection fraction, various volumes, velocities, flows, time intervals, indices, etc.). Some important parameters are also provided with verbal evaluation. Based on mathematical analysis of curves, images of significant stages of dynamics , including parametric images, are also constructed . These images are (in addition to possibly quantitative analysis) also offered for visual evaluation and printing or photographic documentation.
Computer-generated and processed images, quantitative parameters and verbal descriptions are continuously stored in memory, from where they are selected at the end of the evaluation and summarized on the display for final editing . The program then ends with the printing of the results protocol , which is continuously formed and finally coordinated , and which is sent by post or electronically to the relevant clinical partner.
1.2. Computer processing, display and documentation of images
The basic units that make up any scintigraphic study are the scintigraphic images of the distribution of the radio indicator in the field of view of the scintillation camera. The basic methods of scintigraphic image processing are mentioned in the basic OSTNUCLINE manual, part 1, §3.2. In complex programs, these procedures are included (along with other mathematical procedures) in appropriate sequences resulting from the methodology. The aim of these processes is to achieve optimal display for the purposes of visual evaluation and marking of areas of interest.
In most complex programs, a series of suitably summed sequence images is initially displayed, which represents a certain cross-section of the dynamic study . These images are offered for preliminary visual evaluation with standard or non-standard text (see §1.6). In addition, this series of images (including a text description) can be printed for documentation purposes.
For more accurate and detailed evaluation, images of significant phases of the dynamics of the observed process are created during mathematical processing of curves - eg images of heart in end-diastole and end-systole, images of kidneys in secretion and excretion phase, images of heart chambers and lungs during bolus and under. These images are optimized both in terms of time sequence (designed on the basis of mathematical analyzes relevant ntime-activity curves), as well as in terms of spatial arrangement and modulation of the image (images are suitably filtered, have a suitable contrast and cut-off level, a suitable color scale is used, etc.). These images can be photographed or printed for documentation, and visual evaluation is supplemented and specified on the basis of them. Pictures of prominent phase dynamics are part of the Final Protocol and in addition is stored in SAVE AREA, where they can be after the program or a bottom of the recall and manipulate them using standard apparatus OSTNUCLINE system.
1.3. Marking of areas of interest
Defining areas of interest (ROI) in complex programs is done both automatically and manually . If the delimitation is performed automatically according to suitable mathematical algorithms, the possibility of visual inspection and manual adjustment of the ROI always follows . The " ROI " procedures described in the OSTNUCLINE base system manual, which is modified accordingly, are used to manually plot ROIs as well as to check and edit automatically defined ROIs . E.g. if it is necessary to draw or check only one ROI, the relevant submenu does not contain commands for selecting the serial number of the ROI, in some programs the ROI is drawn simu lon two or more paintings, etc. When plotting ROI, tools are available for fast scrolling in the study (over time for a dynamic study) and for adjusting image modulation (LT / UT).
1.4. Mathematical processing of curves
The most important part of a comlex evaluation of dynamic scintigraphic studies is usually the creation and processing of time curves of radioactivity in defined areas of interest. The mathematical analysis of these curves calculates important quantitative parameters of the dynamics of the investigated process.
In the appropriate part of the complex evaluation (usually as soon as the ROIs are defined manually or automatically), a procedure for creating curves from the ROI is included . This procedure gradually calculates the integrals of the accumulated pulses in the individual ROIs for individual images of the dynamic study and stores them in the memory (in the SAVE AREA). During the actual processing, these curves are triggered and displayed in Cartesian coordinates, where on the horizontal axis is the time and on the vertical axis is the frequency in pulses per second (this value is proportional to the instantaneous activity of the radio indicator in the given ROI). The horizontal axis is usually provided with a scalein seconds or minutes, the vertical axis has a scale only if the curve is already calibrated in absolute units of some quantity (eg in milliliters). The curves are displayed on a linear scale , which takes into account the possible grouping of the scintigraphic study into groups of images at different frequencies. In studies where fast and slow dynamics are captured, at the beginning of the processing of the curves there is the possibility of their expansion and compression in the time axis so that the part that is important in the given stage of processing is captured in sufficient detail.
A typical initial step in the processing of curves is the definition of certain significant points and sections on the curves - eg the point of arrival of the activity, the point of maximum, ascending or setup sections, etc. The program automatically determines these points using suitable mathematical algorithms, marks them on the displayed curve with crosses and offers them for approval or manual modification. At the same time, the name of the landmark is shown on the display. In most cases, the point is defined correctly, to the question "Do you agree with the automatically defined point?" we will answer in the affirmative and the program pok rfurther. During the manual modification, we can move the cursor along the curve arbitrarily using either the mouse or the arrow keys on the keyboard; in the table on the display, the coordinates of the respective points are shown for better orientation. By pressing the left mouse button (or the "M" key on the keyboard) we end the manual delimitation and select the point where the cursor was located; the program continues.
This is followed by our own mathematical analysis of curves - for significant points and sections, time intervals, ratios, integrals and other quantities are determined, appropriate functions are interpolated by least squares , various velocity coefficients are calculated, curves are derived, integrated, filtered, deconvolution and so on. . - according to the used mathematical model of the investigated process. Significant points and sections , such as points of maximum rate of increase or leakage of radioactivity, are sometimes also marked (automatically with the possibility of manual modification) on mathematically processed curves .
In some cases, certain external data and parameter values are entered at the request of the program - eg applied activity, heart rate, activity of samples taken, height and weight of the patient. These data are then taken into account in the calculations.
The results of individual stages of mathematical processing of curves are continuously displayed in a graphically coordinated form, along with curves, axis descriptions and other information (for possibly photo documentation), including asking whether we agree with the calculation or want to repeat it (perhaps with a different definition significant points and sections of curves).
1.5. Construction and use of parametric images
Calculations of the most important parameters of the dynamics of the investigated process are performed mostly from the curves of the time course of the activity in suitably defined areas of interest. Sometimes, however, we are interested in the regional distribution of dynamic parameters of the investigated process, which is displayed using so-called parametric images . The construction of a parametric image consists in principle in the following steps:
The resulting image then captures not a simple distribution of activity, but a detailed regional distribution of the dynamics of the monitored process, mapped by the relevant dynamic parameter. In the parametric images, we can clearly and clearly see the functional anomalies of those places of the examined organ whose average activity does not visibly differ from the surroundings. In addition to visual evaluation, we can also perform regional quantification of the function of any part of the organ on these images .
A typical example of a parametric image may be the image of the kidneys in the excretion half-lives of a radiolabel, in which the structure of the renal cortex and pelvis clearly appears, including any sites of delayed excretion or sites of retention. However, the most important example is the parametric images of the phase distribution and amplitude of cardiac pulsation, generated by Fourier analysis of cardiac cycle dynamics in radionuclide ventriculography - see paragraph 3.1.
Parametric images are a very useful complement to a complex evaluation of functional scintigraphic studies. However, there are some pitfalls. Above all, parametric images are usually quite "sensitive" to statistical fluctuations in scintigraphic data. In order for parametric images to be of good quality, it is necessary to achieve good "statistics" (ie a high number of accumulated pulses in the image cells) of the scintigraphic study. Otherwise, parametric images are highly "noisy" or even unevaluable, with the risk of artifacts. Another possible drawback is the somewhat more complex interpretation of parametric images, which requires more experience than simple scintigraphic images. It is always necessary to keep in mind the possibility of interference of overlapping structures with different dynamics in a given projection.
1.6. System of verbal descriptions and evaluation of results
"In the beginning was the word
, and the word was with God, and the word was God ..."
The Gospel of
St. John. 1
Complex computer evaluation of a scintigraphic study does not consist only in mathematical analysis of images and time curves of radioindicator distribution. The completion of the diagnostic process is the formulation and writing of a summary result and finding . The final result of the scintigraphic study is therefore a verbal statement , resp. system of statements.
The basic idea of our system of verbal descriptions of scintigraphic studies can be briefly formulated as follows:
Evaluation texts in the computer evaluation of scintigraphic studies are created in basically three ways:
A common feature of these three methods is the possibility of easy and operative editing of all texts, ie their correction, completion, etc. Each user can use the editor to create their own text files , which can then be called in the text frames of the program with the command "Texts". All text files (standard and user) are stored in the VG11 directory.
The conclusion , which usually summarizes and interprets the results of the evaluation of images, curves and parameters, is again created in our programs either automatically or manually . If the visual evaluation of the images was chosen by the standard normal text and at the same time the values ??of the selected calculated parameters are in the normal range, then the appropriate standard formulation of the conclusion is used (which can be modified, deleted, etc.). Otherwise, the conclusion is entered manually.
The described system of verbal evaluation and insertion of descriptions in computer evaluation of scintigraphic studies is very simple, undemanding and flexible for physicians. It increases the complexity, comfort and productivity of work, as well as the level and speed of the entire diagnostic process.
1.7. Final protocol
A complex computer evaluation of each scintigraphic study ends with the printing of the final report on a printer. The final protocol contains (see Fig.1.7.1):
By suitable graphic division and choice of font size and saturation, maximum clarity of the protocol is achieved .
The printing of the report is optional: at the request of the program we enter the required number of copies of the report (if we enter zero, the report will not be printed). By default, the report is printed on A4 office paper . For lecture purposes (projection on an overhead projector) or presentation on a negatoscope, the final reports (or intermediate results or other images, curves and texts) can also be printed on transparencies - see the attached image. For this purpose, before printing, the printer must be set to a higher print density (quality "best", paper "transparent film") and then return the setting to the standard for printing on paper (when printing on membranes, the ink consumption is significantly higher!).
The entire final protocol is stored for the scintigraphic study and can also be archived with it . The evaluated scintigraphic study, which has a final protocol, is marked with the "OK" symbol in the study list table ("Study" command). Each time the evaluated scintigraphic study is called up, we can use the Program results command (in version 3.3 in the Text submenu) to display this protocol again (and possibly print it out) - it is very advantageous eg for retrospective studies and also for repeated examinations of patients.
In the OSTNUCLINE 2000 system, you can use the Set item to select whether you want to use an internal editor (an improved editor similar to Ostnucline 3.3) or a standard WORD editor to present and edit textual information (visual evaluation, conclusion) and the result log . At the same time, we will select a template for the result protocol (Fig.3.1.4).
For routine work, an internal editor can be recommended, which is fast and provides sufficient options for inserting and editing texts in RTF format, including printing and archiving the result report in the same format (improved compared to version 3.3 with the possibility of editing).
If we choose the standard WORD editor, the loading of texts will slow down a bit, but a larger arsenal of editing tools will be available. The result log will then be automatically in the WORD-document format (.doc file) according to the default template and will also be archived in this way. This protocol already has all the standard editing options, according to the default template it can contain a logo or a picture of the workplace and when sent by e-mail it will be readable on every PC (see Telenuclear Medicine). However , the protocol created by the internal editor can also be added to the WORD (.doc) format .
For routine clinical practice, it is recommended to use the internal editor during the evaluation of the scintigraphic study, and only when creating the final protocol by clicking on the WORD item, if necessary, convert it to the WORD (.doc) format.
Some important stages of a complex evaluation of a scintigraphic study may be useful to print in the form of a so-called intermediate result protocol , which has a fundamentally similar structure to the result protocol (see the lower part of Fig. 1.7.1). This can be done at any time during processing by a complex program using the Print. Intermediate protocols can also be archived along with scintigraphic studies. A typical example of an inter-result protocol is the printing of the results of regional quantification of cardiac cycle dynamics using Fourier images of phase and amplitude, or the results of deconvolution analysis in dynamic renal scintigraphy. Mezivýsledkový report usually contains the details and parameters that are not listed in the Results Final Protocol - that may be a reason to print mezivýsl e dkového Protocol, particularly in pathological and complicated cases.
2. Integration of complex programs into the OSTNUCLINE system
2.1. Table of complex programs
The following complex programs are available in the current version of the OSTNUCLINE system :
VENTR - radionuclide ventricucography
FOURI - Fourier phase analysis (independent)
RKG - bolus angiocardiography
RENDYN - dynamic scintigraphy of kidneys
TRNDYN - dynam . study of transplanted kidney
RENSTAT - static scintigraphy of kidneys (with
quantification)
UFM - dynamic uroflowmetry and cystography
HEPDYN - dynamic scinti. liver - cholescintigraphy
PERMO - dynamic scinti.
THYR quatitative scintigraphy of the thyroid gland
PULMSTAT - static scinti. lungs (perfusion +
ventilation)
OSTEOSTAT - static scintigraphy of the
OSTEODYN skeleton - dynamic (3-phase) scinti. skeleton
GENSTAT - general static scintigraphy
SPLPORTO - dynam . splenoportography
MAMOGR - scintimamography
CISTER - cisternography
FLEBOGR - radionuclide phlebography
OESOGAST - dynamic scintigraphy of the esophagus and stomach
2.2. Running complex programs
By running the
Complex programs command in the main menu of the
OSTNUCLINE system, a list of complex programs
will appear on the display , from where we will select the
desired program with the mouse - it will start and work with the
scintigraphic study that was previously opened. Note: If the study has already been evaluated, a preview
of the relevant result protocol will be displayed
for information . If a complex program consists of several parts, a table of the
individual parts of the program is displayed. If we start the
evaluation (and thus run the program for the first time for a
given scintigraphic study), we always choose to run the first
part of the program with the mouse; the next part starts after s
bluntly
automatically. After the end of the whole program, we return to
the basic menu PROCESSING, from where possibly. with the command
"Study" we can choose another scintigraphic study for a
complex evaluation.
2.3. Interrupt and restart complex programs
In some cases, the need arises to interrupt the automatic running of a complex program. It is eg. If we find incorrect hoarded studies (which prevents further evaluation) or processing error occurred or inaccuracies that b y cause incorrect results. In such cases, it is not necessary to run the program "empty" until the end and then run it all over again from the beginning (as is unfortunately the case with many systems) , but the complex program can be interrupted.
To interrupt a complex program, use the " Exit " command , which is displayed in the lower right corner of the screen. If we move the mouse to this command and press the left button, the complex program will be interrupted and a table of individual parts of the given complex program will appear on the display. Using this table, we can either restart the program from the place where the error occurred, or run the entire program from the beginning, or end the evaluation with the Exit command at the end of the table. In the new system Ostnucline 2000 to ave the opportunity to jog run calculations and processing backwards and forwards using the arrow keys " ¬ ", " ® , Which makes it even easier to control the execution of a complex program and to correct errors.
The possibility of interrupting and restarting complex programs is very advantageous for practice and time-saving. If we make a fault somewhere towards the end of the program (eg. The choice of significant points on curves or when inserting values externally measured parameters), it is not necessary again distinguished ROI, construct curves etc., but interrupts email it to run the program up to the part where the error occurred. Also at the launch research activities sometimes different parts of a complex program repeatedly to realize the different variants of calculation, explored the impact of the modified sections elections curves for f ITAC and other calculations etc.
2.4. Creation of other complex programs
The OSTNUCLINE system can be supplemented by other complex programs. The authors of the system will supplement it on the basis of the development of the field of nuclear medicine, their own research and development and according to the needs of user workplaces. Programs are written in "C" and Visual C ++ and use subroutines from the G11LIB library.
In the future, if the need arises: Simpler programs can be created by users. For reasons of continuity and compatibility, we recommend that users adhere to the concepts of existing complex programs when creating their own programs. More detailed instructions for creating programs, including their inclusion in the "Complex programs" menu, will be given in the OSTNUCLINE manual - programmer's manual.
2.5. Auxiliary files for the operation of complex programs
In addition to their own programs and evaluated scintigraphic studies, complex programs also use some other files stored on disk in their operation. Above all, these are SAVE AREA (described in the basic manual, §1.2), in which complex programs store various intermediate and final images and curves. These are then accessible for possible further processing, photography, comparison, etc. from the basic OSTNUCLINE system , or complex programs.
Furthermore, complex programs create and use three types of auxiliary files. Curves, intermediate results and texts of verbal evaluation of images and curves during the processing of a given study are stored in files with the .PAR extension. .TXT files contain implicit formulations of normal ratings for individual complex programs. Specific examples of the wording of implicit texts are given in the instructions for individual complex programs. Using tex t AC editor, the user can modify the wording of the normal assessment and create other texts. Some complex programs (eg VENTR, RENDYN, THYR) use the data contained in the CAMERA.PAR file. These are data on the scintillation camera display scale, the camera-computer dead time value and calibration parameters for geometric calculations. For the correct operation of these programs, it is necessary to calibrate these data on the basis of appropriate measurements and insert them (using an editor) into the file KAMERA.PA R. The text file KAMERA.PAR has the following structure:
d | P1 | Q1 | DT | |
camera0 | 3.57 | .84670 | 7.89 | 6.0 |
camera1 | 4.2 | 1.8467 | 17.89 | 6.0 |
Column "d" shows the display scales in [ millimeters / cell ] image matrix for each camera used (0, 1, ...). The values ??of P1 and Q1 are regression coefficients for the geometric method (now abandoned) of calculating the end-diastolic volume of the ventricle in the VENTR program. DT is the effective dead time (non-paralyzable) of the entire camera-computer system, which is used to correct rapid dynamic studies for dead time (eg in bolus radiocardiography).
The most important
parameter that each user must calibrate is the display scale d .
We recommend the following procedure as the simplest: Place two and 99m Tc point sources with an activity of several MBq
at a distance of 100 mm from each other in the field of view of
the given scintillation camera (two drops of technetium on a
straight strip of paper, eg millimeter) are enough. We make sure
that the image of the sources is exactly horizontal on the
persistent display. We store the scintigraphic image in a 64x64,
16 bit matrix for a few minutes. In the basic menu
"PROCESSING" we start "Slice" and through the
images of point sources we lead a horizontal profile with a width
of about 5 pixels. The command "R / W" then goes
through the individual points of the profile, recording
the x-coordinates of the maxima in the image profiles of both
point sources. The difference "r" of these values
??indicates the number of pixels per 100 mm. The display scale is
then easily calculated as d = 100 / r [mm / pixel]. We save the
measured value in this way using the e editor in the appropriate place
in column "d" of the CAMERA.PAR file.
Note: In some of the future versions of the
system, we plan to create a program that would perform all these
and other calibration operations and automatically save the
resulting parameters to the CAMERA.PAR file and others.
Calibration of regression coefficients P1 and Q1 is performed using phantoms of the heart chamber and is quite complex (it is described in the book Ullmann V., Kuba J., Kuchař O., Mrhač L., Dudzik J.: "Computer processing of dynamic scintigraphic studies" (conclusion Report of research task No. 30-02-03 MZd) , Ostrava 1985). Since the VENTR program implements a proportional geometric method of calculating ventricular volume, which requires only the display scale "d" and not the regression coefficients P1 and Q1, the user will probably not have a reason to calibrate these parameters.
The measurement of the effective dead time of the camera-computer system is described in the book Ullmann V., Kuba J., Dubroka L., Kuchař O., Závada M., afarčík K., Mrhač L.: "Computer data processing in nuclear medicine" (conclusion Research Task Report No. 30-02-01 MZd) , Ostrava 1980. For the Mediso device in conjunction with the MB 9100 or MB 9200 cameras, the value of the effective dead time is between about 5 and 6 microseconds, so the default value is 6 microseconds. can be left in the CAMERA.PAR file again.