Search

Study Objective: To compare sentinel lymph node (SLN) mapping in women with cervical cancer stage >IB1 (tumor size >2 cm) using indocyanine green (ICG) versus the standard technique using radioisotope technetium 99m radiocolloid (Tc99m) radiotracer with or without blue dye.Design: European multicenter, retrospective observational study (Canadian Task Force classification II-2).Setting: Four academic medical centers.Patients: Ninety-five women with stage IB1 cervical cancer (>2 cm) who underwent SLN mapping with Tc99m with or without blue dye or ICG and radical hysterectomy.Intervention: The detection rate and bilateral mapping rate were compared between ICG and standard Tc99m radiotracer with or without blue dye. Lymphadenectomy was performed, and the false-negative rate was assessed.Measurements and Main Results: Forty-seven patients underwent SLN mapping with Tc99m with or without blue dye, and 48 did so with ICG. All patients underwent radical hysterectomy with or without bilateral salpingo-oophorectomy between 2008 and 2016. The overall detection rate of SLN mapping was 91.5% for Tc99m with or without blue dye and 100% for ICG. A 91.7% rate of bilateral migration was achieved for ICG, significantly higher than the 66% obtained with Tc99m with or without blue dye (p = .025). Nine of the 23 SLN-positive patients (39.1%) were diagnosed exclusively as a result of the ultrastaging used to identify micrometastases or isolated tumor cells only.Conclusions: In advanced cervical cancer (stage IB1 >2 cm), the detection rate and bilateral migration rate on real-time fluorescent SLN mapping were higher with ICG than with Tc99m radiotracer with or without blue dye. SLN mapping and ultrastaging can provide additional information for nodal staging in advanced cervical cancer. In this setting, ICG is a promising tool for mapping, appearing less affected by higher disease stage compared with traditional methods. [ABSTRACT FROM AUTHOR]

The aim of this paper is to develop a general framework for training neural networks (NNs) in a distributed environment, where training data is partitioned over a set of agents that communicate with each other through a sparse, possibly time-varying, connectivity pattern. In such distributed scenario, the training problem can be formulated as the (regularized) optimization of a non-convex social cost function, given by the sum of local (non-convex) costs, where each agent contributes with a single error term defined with respect to its local dataset. To devise a flexible and efficient solution, we customize a recently proposed framework for non-convex optimization over networks, which hinges on a (primal) convexification–decomposition technique to handle non-convexity, and a dynamic consensus procedure to diffuse information among the agents. Several typical choices for the training criterion (e.g., squared loss, cross entropy, etc.) and regularization (e.g., ℓ 2 norm, sparsity inducing penalties, etc.) are included in the framework and explored along the paper. Convergence to a stationary solution of the social non-convex problem is guaranteed under mild assumptions. Additionally, we show a principled way allowing each agent to exploit a possible multi-core architecture (e.g., a local cloud) in order to parallelize its local optimization step, resulting in strategies that are both distributed (across the agents) and parallel (inside each agent) in nature. A comprehensive set of experimental results validate the proposed approach. [ABSTRACT FROM AUTHOR]

Indocyanine green (ICG) represents a feasible alternative to the more traditional methods of sentinel lymph node (SLN) mapping, and interest in this promising tracer is growing. This report outlines our experience with ICG in a minimally invasive laparoscopic approach in women with endometrial cancer and cervical cancer using the Storz SPIES ICG near-infrared fluorescence imaging technology. A total of 49 patients with clinical stage I endometrial cancer (n = 40) or stage I cervical cancer (n = 9) were retrospectively reviewed. All patients had undergone simple or radical laparoscopic hysterectomy with pelvic and/or aortic lymphadenectomy and SLN mapping by means of an intracervical injection of ICG dye at the 3 o'clock and 9 o'clock locations after the induction of general anesthesia. The detection rate of ICG was 100% (49 of 49). The rate of bilateral SLN detection was 86% (42 of 49). Positive lymph nodes were found in 6 patients (12%), with at least 1 positive SLN. The sensitivity and negative predictive value of SLN detection were 100%. All procedures were successfully completed without conversion to open laparotomy, and no intraoperative or postoperative complications occurred. In our preliminary experience, ICG showed a high overall detection rate, and bilateral mapping appears to be a feasible alternative to the more traditional methods of SLN mapping in patients with endometrial cancer and cervical cancer. Laparoscopic SLN mapping with ICG appears to be safe, easy, and reproducible, with a positive impact on patient management. [ABSTRACT FROM AUTHOR]

The semi-supervised support vector machine (S 3 VM) is a well-known algorithm for performing semi-supervised inference under the large margin principle. In this paper, we are interested in the problem of training a S 3 VM when the labeled and unlabeled samples are distributed over a network of interconnected agents. In particular, the aim is to design a distributed training protocol over networks, where communication is restricted only to neighboring agents and no coordinating authority is present. Using a standard relaxation of the original S 3 VM, we formulate the training problem as the distributed minimization of a non-convex social cost function. To find a (stationary) solution in a distributed manner, we employ two different strategies: (i) a distributed gradient descent algorithm; (ii) a recently developed framework for In-Network Nonconvex Optimization (NEXT), which is based on successive convexifications of the original problem, interleaved by state diffusion steps. Our experimental results show that the proposed distributed algorithms have comparable performance with respect to a centralized implementation, while highlighting the pros and cons of the proposed solutions. To the date, this is the first work that paves the way toward the broad field of distributed semi-supervised learning over networks. [ABSTRACT FROM AUTHOR]