Open source processing of ultrasound images with the USTB
The UltraSound ToolBox (USTB), is a collaboration between several international institutions lead by former and current CIUS researchers Alfonso Rodriguez-Molares, (University of Vigo, Spain), Ole Marius Hoel Rindal (UiO and SINTEF) and Stefano Fiorentini (NTNU). It is an open source toolbox for processing of ultrasonic signals, where researchers can share and compare processing methods for ultrasound imaging.
The idea is to make it easier to compare increasingly complex methods introduced in the medical ultrasound research literature. With researchers collaborating and sharing the implementation of their methods, one can scrutinise and build on each other’s methods and research findings – thus bringing the research field forward.
The USTB has so far been used in 47 scientific publications by researchers from institutions such as the Johns Hopkins University (USA), University of Granada (Spain), Delft University of Technology (Netherlands), Iran University of Science and Technology, Federal University of Technology – Parana (Brazil) and the Indian Institute of Technology.
An open-source implementation of all the processing steps when creating an ultrasound image is not just invaluable for researchers, it also offers a unique way to teach ultrasound imaging hands-on. This is exploited in a new course offered at the University of Oslo from the fall of 2021 on Ultrasound Imaging (IN3015/4015). The course is taught module-based, where each module consists of one week of lectures accompanied by a second week to solve a hands-on exercise using the USTB.
The methods currently included in the USTB are: Beamforming including several adaptive beamforming methods, elastography imaging, and some Doppler and image post processing techniques.
The development of the USTB is an ongoing project with improvements and bug fixes added continuously. Recent developments include the addition of and integration with other research software developed in CIUS. This includes FlowLine Ultrasound Simulation Tool (FLUST) developed by CIUS researchers Jørgen Avdal, Ingvild Kinn Ekroll and Hans Torp at NTNU. And the development of a data pipeline between the USTB and Framework for heterogenous medical image computing and visualization (FAST) developed by CIUS researchers Erik Smistad and Andreas Østvik at NTNU.
The figures display images of four cardiac views from a patient heart recorded by CIUS cardiology researchers Erik Andreas Berg and Torvald Espeland. The parasternal short axis (PSAX), the parasternal long axis (PLAX), the apical long axis (ALAX) and the apical four chamber (A4C) view. The raw ultrasound channel data was recorded on a GE Vingmed E95 scanner by CIUS cardiologist researchers, however the images displayed here are processed using the USTB.
