ImageJ based segmentation, processing and evaluation of digital radiology phantom images [lazyzero.de]

Diese Seite ist nicht editierbar. Sie können den Quelltext sehen, jedoch nicht verändern. Kontaktieren Sie den Administrator, wenn Sie glauben, dass hier ein Fehler vorliegt.
====== ImageJ based segmentation, processing and evaluation of digital radiology phantom images ======

===== Abstract =====

==== Introduction ====

In digital radiology it is state of the art to perform specialised procedures to estimate the image quality of the complete system on regular bases. This is done using dedicated phantoms. In general the images taken with these phantoms have to be evaluated manually by a visual observer. This procedure is, depending on the test, very time consuming. To reduce the evaluation time the Open Source quality control tool for medical images - Optimage - was developed. The Optimage software package supports various modalities and phantoms [1]. This work describes the used image processing methods for the segmentation and evaluation of the projection radiography module based on the German DIN 6868-13 (Constancy testing of projection radiography systems with digital image receptors) standard [2].

==== Materials and Methods ====

The phantom DIGI-13 from IBA Wellhöfer was selected as test phantom. For the segmentation of the phantom features, an ImageJ plugin was written. This plugin was designed to deliver a polygon ROI containing the phantom features. During the iterative segmentation process the plugin uses as well build-in ImageJ methods and additional ImageJ plugins for image processing. A classification of the intermediate data checks, if the iterative process has finished or not. The image segmentation uses filter methods to reduce high frequency noise in the image and a combination of the variance filter and contrast enhancement to extract the build-in grid. The Particle Analyzer is then used to segment the grid elements. The subsequent classification uses geometric rules to find the embedded features. From these “approximately” determined positions, the inner grip corners are detected by evaluating the grid crossings.

==== Results ====

The developed plugin is tested with a large number of phantom images regarding the stability and correctness of the results. The plugin works correctly for a large range of different doses and rotation angles. Limitations: Results are not valid for images manipulated during post processing.

==== Conclusions ====

The developed ImageJ plugin is now integrated into Optimage to perform the segmentation of the DIGI-13 phantom. Due to the power of the ImageJ provided segmentation and processing methods, the needed functionality was easy to implement.

==== References ====

[1] Optimage central organised Image Quality Control including statistics and reporting: A.Jahnen, C.Schilz, F.Shannoun, A.Schreiner, J.Hermen, C.Moll; Radiation Protection Dosimetry, Oxford University Press (2008).\\
[2] DIN 6868-13 Constancy testing of projection radiography systems with digital image receptors: NAR; Deutsches Institut für Normung e.V. (2003-02).

===== Author =====
Christian Moll¹, Clemens Schilz², Johannes Hermen¹, Andreas Jahnen¹
{{tag>Moll Schilz Hermen Jahnen}}


===== Organisation =====

CRP Henri Tudor¹, 29, Av. John F. Kennedy, 1855 Luxembourg, Luxembourg\\
Brüderkrankenhaus Trier², Nordallee 1, 54290 Trier, Germany\\


Homepage http://santec.tudor.lu

===== Download =====