Near-Infrared Autofluorescence Signature and Deep Learning Enhance Assessment of Parathyroid Glands

Akgun E, Ibrahimli A, Berber E. Near-Infrared Autofluorescence Signature: A New Parameter for Intraoperative Assessment of Parathyroid Glands in Primary Hyperparathyroidism. J Am Coll Surg. 2024, in press.

Editorial: Brown TC. Near-Infrared Autofluorescence Is an Important Adjunct in Parathyroid Operation. J Am Coll Surg. 2024, in press.

Successful parathyroidectomy in patients with hyperparathyroidism depends on accurately identifying diseased glands. This prospective study evaluated the experience of a single center with a deep learning model that used near-infrared autofluorescence (NIRAF) signatures to determine, intraoperatively, whether parathyroid glands were diseased or normal.

More than 2,000 glands (597 diseased glands) from 797 patients were evaluated. Normal glands had higher NIRAF intensity and lower heterogeneity than diseased glands; use of NIRAF data with the deep learning model yielded an accuracy of 83.3%.

The authors concluded that use of NIRAF combined with a deep learning model had potential value for intraoperative identification of diseased parathyroid glands.

In the editorial that accompanied the article, Brown presented data from his personal experience that confirmed the potential value of NIRAF in identification of abnormal parathyroid glands.