Validation of Artificial Intelligence-enhanced Stimulated Raman Histopathology for Intraoperative Margin Assessment...

Area of Science:

• Pathology
• Oncology
• Medical Imaging

Background:

• Stimulated Raman histology (SRH) offers near-real-time tissue visualization for intraoperative pathology.
• The ROBOSPEC study investigates the accuracy of an integrated artificial intelligence (AI) tool with SRH.

Purpose of the Study:

• To evaluate the accuracy of an AI tool (NYU-AI) for detecting positive surgical margins during robot-assisted radical prostatectomy (RARP).
• To compare the diagnostic performance of AI-enhanced SRH with conventional frozen section (cryo-HE) pathology.

Main Methods:

• Prospective pilot study of 18 patients with prostate cancer undergoing RARP.
• SRH imaging and cryo-HE staining of prostatectomy specimens.
• Analysis using a validated AI algorithm (NYU-AI) and review by blinded urologists.

Main Results:

• NYU-AI demonstrated strong diagnostic agreement with cryo-HE for identifying positive surgical margins.
• Patient-based analysis showed high sensitivity (1.0) and specificity (0.93).
• Sample-based analysis reported high specificity (0.97) and similar sensitivity.

Conclusions:

• Preliminary findings support SRH with NYU-AI for intraoperative detection of positive surgical margins in RARP.
• Further studies are needed to validate this technique for clinical implementation.