Dados do Trabalho

Título

INITIAL EXPERIENCE WITH ARTIFICIAL INTELLIGENCE-ENABLED MULTISPECTRAL IMAGING TO IDENTIFY CANCEROUS MARGINS IN ROBOTIC-ASSISTED RADICAL PROSTATECTOMY

Resumo

Background:
Radical prostatectomy (RP) is the most common surgical treatment for localized prostate cancer, and positive surgical margins (PSM) play a crucial role in cancer control. MarginAssure Imaging System developed by CytoVeris uses a Multi-Spectral Tissue Auto-Fluorescence Imaging (MS-TAFI) concept and employs AI-based machine learning algorithms in tumor detection. This platform is capable of analyzing the unique "optical fingerprint" of tissue based on its intrinsic biomolecular and morphological characteristics without requiring the use of dyes or imaging agents.The system includes an imaging chamber where a surgically excised specimen is placed and imaged. The AI-generated output produced by the system displays the presence or absence of cancerous tissue.
Objective
To evaluate the effectiveness of the MarginAssure Imaging System in real-time identification of ECE and PSM in specimens of robotic-assisted radical prostatectomy (RARP).
Methodology:
This is a prospective study performed from Jan 2023 to April 2023 in our center. We evaluated 511 margins from 179 patients and divided the machine learning process into two phases:
Phase 1 (training) consists of using MarginAssure to image excised prostate specimens and correlate the results with clinical data including preoperative MRI images and biopsy results as well as postoperative histopathology diagnosis to develop the “labeled” data.
Phase 2 (testing) will be the validation of the device on immediately excised prostate specimens before fixation in the intraoperative setting. Postoperatively, the histopathology “gold standard” will be the determinate of positive or negative margins. The imaging results will be evaluated against the pathology.
Results:
The median age 65, Median AUA 14, median SHIM 19, Overall PSM 22% and overall EPE 50%. Our analysis described 76.9% accuracy, 74.4% AUC, 76.5% specificity, and 91.6% sensitivity, 95% confidence interval is (76 +/- 0.1568). These early, initial results validate the feasibility of the approach.
Conclusion:
Initial results showed that the MarginAssure Imaging System has the potential to differentiate between cancerous and benign tissue which may enhance the precision of tumor resection during radical prostatectomy. We described acceptable sensitivity and specificity rates of detection. Next steps would include building a more comprehensive database for algorithm training/testing.