Development of an algorithm for intraoperative autofluorescence assessment of parathyroid glands in primary hyperparathyroidism using artificial intelligence

Akbulut, Serkan; Erten, Ozgun; Kim, Yoo Seok; Gökçeimam, Mehmet; Berber, Eren

doi:10.1016/j.surg.2021.01.033

Cited by 14 publications

(17 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In our previous AI model that used manually calculated parathyroid AF signal intensity and heterogeneity values, the PPV to predict a normal gland was 94%. 4 In that study, a first-generation NIFI camera with a lower resolution and frame rate was used, 3 but the similar PPV rates in the two studies support a potential utility of the AI model to differentiate normal from abnormal PGs during parathyroidectomy.…”

Section: Discussionmentioning

confidence: 87%

“…IRBapproved data forms were filled out intraoperatively and entered into a Redcap database. As described before, 4,6 a PG was considered abnormal if hypersecreting per Miami criteria and/or hypercellular on histopathology. The remaining PGs in parathyroidectomy procedures and those identified during thyroidectomy procedures in eucalcemic patients were scored as normal.…”

Section: Surgical Techniquementioning

confidence: 99%

“…The fact that PGs emit a unique autofluorescence (AF) signal when stimulated with a 785-nm wavelength laser 3,4 provides the surgeon an additional tool for parathyroid identification. 3,4 In addition to the ability of this technology to detect PGs, 5 our group has also had an interest in the utility of AF signals to provide an additional intraoperative parameter for differentiating normal from abnormal PGs in primary hyperparathyroidism. We initially reported that abnormal versus normal glands exhibited a darker and more heterogeneous AF pattern on NIFI.…”

Section: Introductionmentioning

confidence: 99%

“…3 Furthermore, we reported that using machine learning, decision trees could be created that could be used to help differentiate normal from abnormal glands. 4 Nevertheless, AF intensity and heterogeneity values used to create these algorithms were manually calculated.…”

Section: Introductionmentioning

confidence: 99%

“…Subsequently, we demonstrated that these differences could be quantified by measuring the intensity and heterogeneity of the AF signal 3 . Furthermore, we reported that using machine learning, decision trees could be created that could be used to help differentiate normal from abnormal glands 4 . Nevertheless, AF intensity and heterogeneity values used to create these algorithms were manually calculated.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

A visual deep learning model to predict abnormal versus normal parathyroid glands using intraoperative autofluorescence signals

Avci

Isiktas

Ergun

et al. 2022

Journal of Surgical Oncology

Self Cite

View full text Add to dashboard Cite

Background: Previous work demonstrated that abnormal versus normal parathyroid glands (PGs) exhibit different patterns of autofluorescence, with former appearing darker and more heterogenous. Our objective was to develop a visual artificial intelligence model using intraoperative autofluorescence signals to predict whether a PG is abnormal (hypersecreting and/or hypercellular) or normal before excision during surgical exploration for primary hyperparathyroidism.Methods: A total of 906 intraoperative parathyroid autofluorescence images of 303 patients undergoing parathyroidectomy/thyroidectomy were used to develop model. Autofluorescence image of each PG was uploaded into the visual artificial intelligence platform as abnormal or normal. For deep learning, randomly chosen 80% of data was used for training, 10% for testing, 10% for validation. The area under the receiver operating characteristic (AUROC), area under the precision-recall curve (AUPRC), recall (sensitivity), and precision (positive predictive value) of the model were calculated.Results: AUROC and AUPRC of the model to predict normal and abnormal PGs were 0.90 and 0.93, respectively. Recall and precision of the model were 89% each. Conclusion:Visual artificial intelligence platforms may be used to compare the autofluorescence signal of a given parathyroid gland against a large database. This may be a new adjunctive tool for intraoperative assessment of parathyroid glands during surgical exploration for primary hyperparathyroidism.

show abstract

Section: Discussionmentioning

confidence: 87%

Section: Surgical Techniquementioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

A visual deep learning model to predict abnormal versus normal parathyroid glands using intraoperative autofluorescence signals

Avci

Isiktas

Ergun

et al. 2022

Journal of Surgical Oncology

Self Cite

View full text Add to dashboard Cite

show abstract

Near-Infrared Autofluorescence Signatures of Single- vs Multigland Disease in Primary Hyperparathyroidism

Akgun,

Berber

2024

JAMA Otolaryngol Head Neck Surg

View full text Add to dashboard Cite

ImportanceThe success of parathyroidectomy depends on accurate intraoperative localization and identification of all diseased glands in parathyroid exploration based on surgeon expertise to prevent persistent hyperparathyroidism. Near-infrared autofluorescence (NIRAF) imaging has recently emerged as a promising adjunctive intraoperative tool for localizing parathyroid glands; however, its potential utility in the assessment of parathyroid glands has yet to be established.ObjectiveTo analyze the differences in NIRAF signatures of parathyroid glands in single vs multiple glands in primary hyperparathyroidism (pHPT).Design, Setting, and ParticipantsThis prospective diagnostic study analyzed in vivo NIRAF images of parathyroid glands obtained during parathyroidectomies between November 18, 2019, and December 31, 2023, at a single tertiary referral center. Pixel intensities of the images were measured using third-party software. Patients who underwent parathyroidectomy for sporadic pHPT using a second-generation NIRAF imaging device were included. Patients with multiple endocrine neoplasm disorders were excluded. In vivo NIRAF images obtained during the procedures were analyzed.ExposureNear-infrared autofluorescence imaging during parathyroidectomy.Main Outcomes and MeasuresThe primary outcomes were the autofluorescence intensity and heterogeneity of single adenomas and multigland disease (ie, double adenomas and 3- or 4-gland hyperplasia) in sporadic pHPT. Normalized autofluorescence intensity was calculated by dividing the mean pixel intensity of the parathyroid gland by the background tissue. A heterogeneity index was calculated by dividing the standard deviation by the mean pixel intensity of the gland. The secondary outcome was the visibility of each parathyroid gland on NIRAF imaging before it became apparent to the naked eye during exploration.ResultsA total of 1287 in vivo NIRAF images obtained from 377 patients (median [IQR] age, 66 [56-73] years; 299 female [79.3%]) were analyzed. Of all patients, 230 (61.0%) had a single adenoma, 91 (24.1%) had double adenomas, and 56 (14.9%) had 3- or 4-gland hyperplasia. A mean (SD) of 3.4 (1.1) parathyroid glands were identified in the procedures. A comparison of 581 diseased glands (45.1%) and 706 normal glands (54.9%) showed a lower median normalized autofluorescence intensity of 2.09 (95% CI, 1.07-4.01) vs 2.66 (95% CI, 1.43-4.20; effect size = 0.36) and higher heterogeneity index of 0.18 (95% CI, 0.07-0.41) vs 0.11 (95% CI, 0.01-0.27; effect size = 0.45), respectively. Of diseased glands, single adenomas (233 [40.1%]) vs double adenomas (187 [32.2%]) and 3- or 4-gland hyperplasia (161 [27.7%]) had a lower median autofluorescence intensity of 1.92 (95% CI, 1.02-4.44) vs 2.22 (95% CI, 1.10-3.97; effect size = 0.21), respectively. On receiver operating characteristic analysis, the optimal autofluorescence intensity threshold to differentiate between single adenomas vs multigland disease was 2.14, with a sensitivity of 64.4%, specificity of 58.1%, and area under the curve of 0.626.Conclusions and RelevanceThese findings suggest that parathyroid glands in single- vs multigland disease may exhibit different autofluorescence characteristics. Although the effect size was modest, the differences identified should be kept in mind when assessing the parathyroid glands during surgical exploration.

show abstract

A coaxial excitation, dual‐red‐green‐blue/near‐infrared paired imaging system toward computer‐aided detection of parathyroid glands in situ and ex vivo

Kim

Lee

Kim

et al. 2022

Journal of Biophotonics

View full text Add to dashboard Cite

Early and precise detection of parathyroid glands (PGs) is a challenging problem in thyroidectomy due to their small size and similar appearance to surrounding tissues. Near‐infrared autofluorescence (NIRAF) has stimulated interest as a method to localize PGs. However, high incidence of false positives for PGs has been reported with this technique. We introduce a prototype equipped with a coaxial excitation light (785 nm) and a dual‐sensor to address the issue of false positives with the NIRAF technique. We test the clinical feasibility of our prototype in situ and ex vivo using sterile drapes on 10 human subjects. Video data (1287 images) of detected PGs were collected to train, validate and compare the performance for PG detection. We achieved a mean average precision of 94.7% and a 19.5‐millisecond processing time/detection. This feasibility study supports the effectiveness of the optical design and may open new doors for a deep learning‐based PG detection method.

show abstract

Development of an algorithm for intraoperative autofluorescence assessment of parathyroid glands in primary hyperparathyroidism using artificial intelligence

Cited by 14 publications

References 25 publications

A visual deep learning model to predict abnormal versus normal parathyroid glands using intraoperative autofluorescence signals

A visual deep learning model to predict abnormal versus normal parathyroid glands using intraoperative autofluorescence signals

Near-Infrared Autofluorescence Signatures of Single- vs Multigland Disease in Primary Hyperparathyroidism

A coaxial excitation, dual‐red‐green‐blue/near‐infrared paired imaging system toward computer‐aided detection of parathyroid glands in situ and ex vivo

Contact Info

Product

Resources

About