Real-time administration of indocyanine green in combination with computer vision and artificial intelligence for the identification and delineation of colorectal liver metastases.

INTRODUCTION

Fluorescence guided surgery for the identification of colorectal liver metastases (CRLM) can be better with low specificity and antecedent dosing impracticalities limiting indocyanine green (ICG) usefulness currently. We investigated the application of artificial intelligence methods (AIM) to demonstrate and characterise CLRMs based on dynamic signalling immediately following intraoperative ICG administration.

METHODS

Twenty-five patients with liver surface lesions (24 CRLM and 1 benign cyst) undergoing open/laparoscopic/robotic procedures were studied. ICG (0.05 mg/kg) was administered with near-infrared recording of fluorescence perfusion. User-selected region-of-interest (ROI) perfusion profiles were generated, milestones relating to ICG inflow/outflow extracted and used to train a machine learning (ML) classifier. 2D heatmaps were constructed in a subset using AIM to depict whole screen imaging based on dynamic tissue-ICG interaction. Fluorescence appearances were also assessed microscopically (using H&E and fresh-frozen preparations) to provide tissue-level explainability of such methods.

RESULTS

The ML algorithm correctly classified 97.2 % of CRLM ROIs (n = 132) and all benign lesion ROIs (n = 6) within 90-s of ICG administration following initial mathematical curve analysis identifying ICG inflow/outflow differentials between healthy liver and CRLMs. Time-fluorescence plots extracted for each pixel in 10 lesions enabled creation of 2D characterising heatmaps using flow parameters and through unsupervised ML. Microscopy confirmed statistically less CLRM fluorescence vs adjacent liver (mean ± std deviation signal/area 2.46 ± 9.56 vs 507.43 ± 160.82 respectively p < 0.001) with H&E diminishing ICG signal (n = 4).

CONCLUSION

ML accurately identifies CRLMs from surrounding liver tissue enabling representative 2D mapping of such lesions from their fluorescence perfusion patterns using AIM. This may assist in reducing positive margin rates at metastatectomy and in identifying unexpected/occult malignancies.