Artificial Intelligence through supervised machine learning remains an attractive and popular research area in medical image processing. The objective of such research is often tied to the development of an intelligent computer aided diagnostic system whose aim is to assist physicians in their task of diagnosing diseases. The quality of the resulting system depends largely on the availability of good data for the machine learning algorithm to train on. Training data of a supervised learning process needs to include ground truth, i.e., data that have been correctly annotated by experts. Due to the complex nature of most medical images, human error, experience, and perception play a strong role in the quality of the ground truth. In this paper, we present the results of annotating lumbar spine Magnetic Resonance Imaging images for automatic image segmentation and propose confidence and consistency metrics to measure the quality and variability of the resulting ground truth data, respectively.
Natalia, F., Meidia, H., Afriliana, N., Al-Kafri, A. S., Sudirman, S., Simpson, A., … Bashtawi, M. (2019). Development of Ground Truth Data for Automatic Lumbar Spine MRI Image Segmentation. In Proceedings - 20th International Conference on High Performance Computing and Communications, 16th International Conference on Smart City and 4th International Conference on Data Science and Systems, HPCC/SmartCity/DSS 2018 (pp. 1449–1454). Institute of Electrical and Electronics Engineers Inc. https://doi.org/10.1109/HPCC/SmartCity/DSS.2018.00239