Annotating Medical Image Data

Abstract

This chapter describes the annotation of the medical image data that were used in the VISCERAL project. Annotation of regions in the 3D images is nontrivial, and tools need to be chosen to limit the manual work and have semi-automated annotation available. For this, several tools that were available free of charge or with limited costs were tested and compared. The GeoS tool was finally chosen for the annotation based on the detailed analysis, allowing for efficient and effective annotations. 3D slice was chosen for smaller structures with low contrast to complement the annotations. A detailed quality control was also installed, including an automatic tool that attributes organs to annotate and volumes to specific annotators, and then compares results. This allowed to judge the confidence in specific annotators and also to iteratively refine the annotation instructions to limit the subjectivity of the task as much as possible. For several structures, some subjectivity remains and this was measured via double annotations of the structure. This allows the judgement of the quality of automatic segmentations.

An Artificial Intelligence Method for Phenotyping of OCT Scans Using Unsupervised and Self-supervised Deep Learning

“…Technological advances in AI and their applications to medical imaging have been fueled by the advent of large-scale biobank studies with multimodal data (e.g., the UK Biobank) (32). However, to date, image analysis has largely been limited by the lack of data annotation by experts (33) and limited transferability from one dataset to another. These limitations have consequences in settings where datasets lack the information needed for specific studies.…”

Distributed contrastive learning for medical image segmentation

“…Deep learning has demonstrated outstanding achievements in 3D medical image analysis [52,21,39,33], yet is heavily hampered by the expensive cost of the required expert annotations [49,23]. To address this problem, Self-Supervised Learning (SSL) has received significant attention due to its promising ability to learn representations without annotations [10,11,6,28,20], which has become an important label-efficient solution in 3D medical image analysis [69,50,32,2,34,36].…”

ContIG: Self-supervised Multimodal Contrastive Learning for Medical Imaging with Genetics

“…It aids in disease prevention, early detection, diagnosis, and treatment. However, efforts to employ machine learning algorithms to support in clinical settings are often hampered by the high costs of required expert annotations [41]. At the same time, large-scale biobank studies have recently started to aggregate unprecedented scales of multimodal data on human health.…”