PhD Scientific Days 2019

Budapest, April 25–26, 2019

An effective image enhancement algorithm for digital variance angiography

Szöllősi, Dávid

Dr. Dávid Szöllősi1,2, Dr. Marcell Gyánó1,3, Dr. Viktor Imre Óriás1,3,4 Dr. Szabolcs Osváth1,2, Dr. Péter Sótonyi3, Dr. Krisztián Szigeti1,2

1: Kinepict Health Ltd, Budapest, Hungary
2: Department of Biophysics and Radiation Biology, Semmelweis University, Budapest, Hungary
3: Heart and Vascular Center, Semmelweis University, Budapest, Hungary
4: Bács-Kiskun County Hospital, Kecskemét, Hungary

Language of the presentation


Text of the abstract

Introduction: Digital variance angiography (DVA) is a method based on applying kinetic imaging to X-ray angiographic image series. In pursuit of better patient safety and higher cost-efficiency, image quality is a limiting factor: better image quality (e.g. higher contrast-to-noise ratio (CNR), less artifacts) provides an opportunity X-ray dose and injected contrast media reduction.

Aims: Our purpose was to develop an image processing algorithm that can suppress the motion artifacts and enhance the contrast of DVA images.

Method: An image processing algorithm based on temporal differentiation (DVA+) was developed and implemented in Matlab 2016a (Mathworks). To compare the contrast-to-noise ratio, abdominal and lower extremity X-ray angiographic image series of 30 patients were used to create DVA and DVA+ images. Pairs of vascular and perivascular background regions of interest (ROIs) were selected by hand (n=3283) using an ImageJ macro. CNR was calculated as the ratio of the vessel-background difference and the background standard deviation.

Results: The median SNR of DVA images was 9.26 (Q1: 4.90, Q3: 15.86) while the median SNR of DVA+ images was 14.44 (Q1: 7.14, Q3: 26.00). The median ratio of the SNR values of the same ROI pairs was 1.49 (Q1: 1.12, Q3: 2.02).

Conclusion: DVA+ is an effective algorithm for CNR improvement in X-ray angiography based on kinetic imaging. This improvement may enable the implementation of substantial X-ray and/or contrast media dose reduction clinical protocols.

Data of the presenter

Doctoral School: Basic and Translational Medicine
Program: Cellular and molecular biophysics
Supervisor: Krisztián Szigeti