NUclear MEdicine DIagnostic and Artificial Intelligence

Overview

Artificial intelligence algorithms for parathyroid adenoma detection in parathyroid scans The aim of the study is to develop programs for the automatic subtraction of planar and tomographic parathyroid images and for the localization of parathyroid adenomas using artificial intelligence tools. Inclusion criteria Patients having undergone a double isotope parathyroid diagnostic examination using the standard Besançon University Hospital protocol (double injection of 123I and 99mTc, pinhole image, low energy high resolution (LEHR) and (single-photon emission tomography / Computed tomography) SPECT/CT +/- (positron emission tomography (PET) Choline). Secondary endpoints Concordance between diagnoses made by nuclear physicians using both methods (current method and new treatment method).

State of the art : Planar scintigraphy of the parathyroid glands offers good diagnostic performance. The introduction of SPECT coupled with CT ten years ago has improved anatomical location, facilitating surgery. With dual-isotope scans, the physician has to subtract the images to highlight the parathyroid adenomas, and use the SPECT/CT slices to pinpoint the exact location of the adenomas. Several software packages, supplied by nuclear medicine image processing software publishers, are available on the market, but they require numerous actions on the part of doctors, which reduce reproducibility and necessitate a learning curve for the physician. Research Hypothesis : The images were taken as part of the patient's diagnosis, without any additional acquisition compared with clinical routine. Acquisitions were of good quality, with good centering of the thyroid on the images. The 3 Pinhole, LEHR and SPECT/CT acquisitions were performed after injection of Iodine 123 and 3 hours after Technetium 99m. Expected results : Creation of a program for automatic subtraction of planar and tomographic images. Use artificial intelligence algorithms to help doctors with their diagnosis. Main objective : Development of an artificial intelligence-based diagnostic assistance program. Secondary objectives : Creation of an automatic program for subtracting double-isotope parathyroid images. This image processing is a prerequisite for artificial intelligence analysis methods. Inclusion Criteria : Men and women over 18 years of age who have undergone double-isotope parathyroid scintigraphy. No refusal to use data for research purposes Retrospective (after 13/04/2022: Date of posting of patient information in the department) and prospective data Non-inclusion criteria : Patient refusal to use data for research purposes Patients who did not benefit from the standard acquisition protocol (123I injection followed by 99mTc 3 hours after LEHR, PINHOLE and SPECT/CT acquisitions). Description of research methodology : Retrospective and prospective study using data from the medical records of patients meeting the eligibility criteria. No additional examinations were carried out on the patient.