robotic surgery in 339, 29, 75, 47, and 10 cases, respectively. Of these, 11 had Clavien III complication, with only one case of intra-operative open conversion. RAPN implementation based on the concept of trifecta was supervised by MISC. The positive surgical margin rate was 0% for the recent 43 cases. Median follow-up and warm ischemic time was 13 (0.4z38) months and 19.8 (7z53) min within 25 min, respec- tively, except in the initial two cases. Only one complication, an aneu- rysm belonging to Clavien III, was observed. CONCLUSIONS: This is the first report of robotic surgery using 3D reconstructional imaging implemented within a constitutional framework of an academic institution. MISC provides immeasurable benefits in terms of patient safety and education of the robotic surgical team, ultimately providing an efficient system for RAPN. Source of Funding: none MP34-13 BEYOND THE LEARNING CURVE: ROBOTIC PARTIAL NEPHRECTOMY OUTCOMES CONTINUE TO IMPROVE WITH SURGEON EXPERIENCE David J. Paulucci, New York, NY; Louis S. Krane, Ashok K. Hemal, Winston-Salem, NC; Ketan K. Badani*, New York, NY INTRODUCTION AND OBJECTIVES: The learning curve to attain acceptable levels of warm ischemia time (WIT), postoperative complication (PC) rates and estimated blood loss (EBL) during robotic partial nephrectomy (RPN) is reportedly short (~20-30 cases) for high volume surgeons with extensive experience in robotic surgery or laparoscopic partial nephrectomy. Since increasing surgeon experience beyond the initial learning curve has been shown to result in improved perioperative outcomes in radical prostatectomy, we sought to deter- mine whether this phenomenon also exists in RPN by analyzing the influence of surgeon experience on perioperative outcomes in two experienced surgeons who have surpassed the initial learning curve of RPN. METHODS: The present retrospective study identified 573 pa- tients undergoing RPN from 2 surgeons at 2 medical centers. The most recent 175 consecutive RPN cases performed by the first surgeon and 398 recent cases performed by the second surgeon were analyzed. We assessed trends in perioperative outcomes and tumor specific char- acteristics using pearson product moment correlations and univariable logistic regression analysis to assess the association of the number of RPNs on WIT, EBL, PC, body mass index (BMI), tumor size and R.E.N.A.L. nephrometry score. RESULTS: Median R.E.N.A.L. score, tumor size, BMI were 6, 2.6cm and 29.22. Median WIT, EBL, and complication rates were 15 minutes, 75 cc and 17.5% respectively. With increasing surgeon experience, WIT (r¼-.340, p<.001) decreased (Figure 1) while R.E.N.A.L. score (r¼.174, p<.001) (Figure 2), tumor size (r¼.102, p¼.012) and BMI (r¼.082, p¼.049) increased over time. Increasing surgeon experience was not associated with PC (p¼.826) or EBL (p¼.984). CONCLUSIONS: The learning curve for RPN improves steadily even after the initial learning curve is reached. WIT continues to trend downwards despite increasing tumor complexity, size, and higher BMI patients suggesting the true learning curve is longer than previous re- ports have suggested. Source of Funding: None. MP34-14 INVESTIGATING THE FEASIBILITY OF OPTICAL COHERENCE TOMOGRAPHY TO IDENTIFY PROSTATE CANCER e AN EX-VIVO STUDY Amit Patel*, London, United Kingdom; Christoph Otte, Alexander Schlaefer, Hamburg, Germany; Dror Nir, Cambridge, United Kingdom; Sebastian Otte, Hamburg, Germany; Thane Ngo, Tim Loke, Mathias Winkler, London, United Kingdom INTRODUCTION AND OBJECTIVES: Prostate cancer diag- nosis can benefit from cost-reduction, shortening & reduction of observer variability. A targeted real-time biopsy technique would be desirable. Optical Coherence Tomography (OCT) has shown refine- ment in probe design, resolution & imaging depth make it a candidate for a ‘virtual’ biopsy approach. We explore the feasibility of OCT in combination with computerised interpretation of signals & application of Machine-Learning algorithms for tissue diagnosis. We report the results of OCT imaging on ex-vivo prostate tissue to identify cancer without the need for biopsy core processing. METHODS: OCT scans were obtained from 24 patients(pts) who underwent radical prostatectomy. 2 tissue strips 15mm x 8mm x 6mm were prepared & scanned twice from the capsular & the excision surface with an OCT microscope. Bidirectional Dynamic Cortex Memory Network was trained & tested on randomly chosen samples of A-scan data through 10 cycles. Mean classification rate & standard deviation were calculated. e476 THE JOURNAL OF UROLOGY â Vol. 195, No. 4S, Supplement, Saturday, May 7, 2016