Towards automated spectroscopic tissue classification in thyroid and parathyroid surgery

Area of Science:

• Surgical innovation
• Medical imaging
• Spectroscopy

Background:

• Preventing iatrogenic parathyroid injury during surgery is crucial.
• Parathyroid-specific image enhancement could aid surgeons.
• Hyperspectral imaging requires understanding tissue spectral signatures.

Purpose of the Study:

• To investigate the feasibility of diffuse reflectance spectroscopy (DRS) for automated tissue classification.
• To assess silicon (Si) and indium-gallium-arsenide (InGaAs) sensors for distinguishing parathyroid tissue.

Main Methods:

• DRS (350-1830 nm) was performed during (para-)thyroid resections.
• 36 spectral features were extracted for classification.
• Binary logistic regression and leave-one-out cross-validation were used.

Main Results:

• Classification accuracy for parathyroid vs. adipose was 79% (Si), 82% (InGaAs), and 97% (combined).
• Classification accuracy for parathyroid vs. thyroid was 80% (Si), 75% (InGaAs), and 82% (combined).
• Combined Si and InGaAs sensors improved accuracy.

Conclusions:

• Si and InGaAs sensors are accurate for automated spectroscopic classification of parathyroid, adipose, and thyroid tissues.
• Combining sensor technologies enhances classification accuracy.
• Further research towards hyperspectral imaging is warranted.