- Journal
- Année
- 2024
- DOI
- 10.1109/ISBI56570.2024.10635117
Abstract
Primary hyperparathyroidism usually involves surgery. In this situation, pre-operative localization of aberrant Parathyroid Glands (PGs) is crucial. One of the existing approaches, Nuclear Medicine is the widely used procedure. It provides PINHOLE and Low Energy High-Resolution (LEHR) images through the injection of two radio-elements, technetium ((TC)-T-99m-MIBI) and iodine (I-123). After that, the physicians perform a manual normalization and subtraction of both images for PGs detection which is time-consuming. This paper proposes an automatic detection of the PGs by the combination of statistical normalization and artificial intelligence for the subtraction of (TC)-T-99m-MIBI and iodine I-123 images. The proposed methodology is applied to PINHOLE images from 88 retrospective, single-center studies. The obtained results achieved a mean correlation of 0.95 compared to the physician results.