Understanding coordinate systems and DICOM for deep learning medical image analysis

Understanding coordinate systems and DICOM for deep learning medical image analysis

Sometimes you think you understand something, but you fail to explain it. This is the time that you have to look back from a different perspective and start over. When you dig in medical images you will see different concepts to seem vague and non-intuitive , at least in the beginning. You will see people discussing DICOM and coordinate systems you have never heard before. As a result, a lot of misconceptions and confusions are born. If you are in this position, or if you would like to know about AI in medical imaging this article is for you.

Back in 2017, when I applied for my master’s degree in biomedical engineering everybody asked me why, as I was already obsessed with deep learning. Now, every multidisciplinary deep learning research project requires domain knowledge such as medical imaging. Interestingly, the funding in the AI healthcare domain is continuously increasing. As an quantitative example of first google search that one can find out:

The market for machine learning in diagnostic imaging will top 2 billion $ by 2023.

So, the reason that I decided to write this article is to help ML people dive into medical imaging .

In a previous article , I talked about a common deep learning pipeline applied to multi-modal magnetic resonance datasets. All of that of course with our under development open source pytorch library called medicalzoo-pytorch . However, I didn’t dive into the particularities of the medical world too much. In the end, I used already processed data from an ML competition (and not from a messy hospital), so somebody else did the dirty work for me. This tutorial is partly based in the nipy [1] and 3D Slicer.org [2] documentations for medical images and Dicom files.

However, I decided to adapt and revisit the concepts and make them more familiar to Machine and deep learning engineers . There are a lot of assumptions that ML engineers have no idea about. Other multi-disciplinary projects have this kind of terminology problem. To this end, I considered it of great value to bridge this gap between medical imaging concepts and deep learning that no one talks about , in this humble post. At least, I’ll try my best, one concept at a time!

For a more comprehensive approach on AI in medicine, check out the awesome brand-new coursera course on AI for medicine.

Contents

The coordinate systems in medical imaging
World coordinate system
Anatomical coordinate system
Medical Image coordinate system (Voxel space)
Moving between worlds
Moving from one modality to another
Introduction to DICOM for machine learning engineers

Notation : medical image tutorials often call the MRI and CT exams as ‘model’. For convenience, and to avoid misconceptions we will use the...