3D medical image processing in cardiology
By: Graylight Imaging MedTeam
The previous post dealt with the topic of medical image segmentation as a method for preparing neural network learning data. However, the 3D medical image processing cardiovascular models prepared via the segmentation process have much wider applications.
The revolution in 3D medical image processing
It can be seen, that in recent years, advances in technology, 3D printing and artificial intelligence have revolutionised the medical world. Undoubtedly, this is welcome and vital news, as it heralds numerous benefits.
Applications of 3D heart modelling and printing
The three-dimensional anatomical models produced by 3D medical image processing are used primarily in medical education. Among other things, they make it possible to present to pupils and students the full spectrum of malformations and variations.
They also give specialists the opportunity to familiarise themselves with the anatomy before performing a procedure – especially in the case of heart defects, and to assess vascular anatomy.
Example of a 3D Coronary computed tomography angiography render
with a stent on the right coronary artery obtained during the project.
In addition, with 3D medical image processing it is possible to perform simulations that use heart models, allowing experts to learn and train without risking the health and lives of patients.
3D medical image processing also makes it possible to practice different scenarios and perform complex treatments and procedures.
Such simulations are particularly employed when training specialists. The use of 3D printing is also extremely helpful in the case of patients with tumours – the use of different printing materials and colours makes it possible to locate the lesion, its size or vascularisation.
Our experience in 3d medical image processing and segmentation in cardiology
We have been segmenting anatomical areas in cardiology for over 3 years. Creating a ground-truth set allowing to achieve desired results here involves creating a set of several hundred segmentations.
The biggest challenge, apart from the quantity, is the accuracy of created segmentations. From our experience in creating various 3D models so far, it has been the segmentation of the coronary arteries that we have found most difficult and labour-intensive.
But let’s not prejudge the facts, because it is mainly the experience of segmenting coronary arteries that the next text will deal with. The difficulties of working with these structures occur due to the following factors: the complex shape of the vessels, the variety of vessels in the patients – especially when anomalies are present, the small diameter of the vessel in relation to the spatial resolution of the examination, the blurring and noise of the image resulting mainly from the movements of the heart.
So, you will read about how our segmentation process looks like and how we managed to improve it in the next post.
Example of Coronary Computed Tomography Angiography image with motion artifacts and noise.
Examples of the use of 3D modelling and 3D printing in the cardiovascular field are numerous and could be mentioned almost endlessly (source). Just a few examples of applications have helped to illustrate the enormous possibilities offered by the combination of technology and medicine under the aegis of 3D medical image processing.