Departamento de Ingeniería Industrial, Instituto de Innovación en Productividad y Logística CATENA-USFQ, Colegio de Ciencias e Ingeniería, Universidad San Francisco de Quito, Diego de Robles s/n y Vía Interoceánica, Quito 170901, Ecuador.
Department of Industrial and Management Systems Engineering, University of South Florida, 4202 E. Fowler Avenue, Tampa, FL 33620, USA.
Artif Intell Med. 2021 Sep;119:102154. doi: 10.1016/j.artmed.2021.102154. Epub 2021 Aug 24.
Deep learning plays a critical role in medical image segmentation. Nevertheless, manually designing a neural network for a specific segmentation problem is a very difficult and time-consuming task due to the massive hyperparameter search space, long training time and large volumetric data. Therefore, most designed networks are highly complex, task specific and over-parametrized. Recently, multiobjective neural architecture search (NAS) methods have been proposed to automate the design of accurate and efficient segmentation architectures. However, they only search for either the micro- or macro-structure of the architecture, do not use the information produced during the optimization process to increase the efficiency of the search, or do not consider the volumetric nature of medical images. In this work, we present EMONAS-Net, an Efficient MultiObjective NAS framework for 3D medical image segmentation that optimizes both the segmentation accuracy and size of the network. EMONAS-Net has two key components, a novel search space that considers the configuration of the micro- and macro-structure of the architecture and a Surrogate-assisted Multiobjective Evolutionary based Algorithm (SaMEA algorithm) that efficiently searches for the best hyperparameter values. The SaMEA algorithm uses the information collected during the initial generations of the evolutionary process to identify the most promising subproblems and select the best performing hyperparameter values during mutation to improve the convergence speed. Furthermore, a Random Forest surrogate model is incorporated to accelerate the fitness evaluation of the candidate architectures. EMONAS-Net is tested on the tasks of prostate segmentation from the MICCAI PROMISE12 challenge, hippocampus segmentation from the Medical Segmentation Decathlon challenge, and cardiac segmentation from the MICCAI ACDC challenge. In all the benchmarks, the proposed framework finds architectures that perform better or comparable with competing state-of-the-art NAS methods while being considerably smaller and reducing the architecture search time by more than 50%.
深度学习在医学图像分割中起着关键作用。然而，由于大量的超参数搜索空间、长的训练时间和大体积的数据，手动为特定的分割问题设计神经网络是一项非常困难和耗时的任务。因此，大多数设计的网络都非常复杂、特定于任务且参数过多。最近，已经提出了多目标神经架构搜索（NAS）方法来自动设计准确且高效的分割架构。然而，它们只搜索架构的微观或宏观结构，不利用优化过程中产生的信息来提高搜索效率，或者不考虑医学图像的体积性质。在这项工作中，我们提出了 EMONAS-Net，这是一种用于 3D 医学图像分割的高效多目标 NAS 框架，该框架优化了分割准确性和网络大小。EMONAS-Net 有两个关键组成部分，一个新颖的搜索空间，考虑了架构的微观和宏观结构的配置，以及一种基于代理的多目标进化算法（SaMEA 算法），该算法有效地搜索最佳的超参数值。SaMEA 算法利用进化过程初始几代中收集的信息来识别最有前途的子问题，并在突变过程中选择表现最佳的超参数值，以提高收敛速度。此外，还引入了随机森林代理模型来加速候选架构的适应度评估。EMONAS-Net 在 MICCAI PROMISE12 挑战赛的前列腺分割任务、Medical Segmentation Decathlon 挑战赛的海马体分割任务和 MICCAI ACDC 挑战赛的心脏分割任务上进行了测试。在所有基准测试中，所提出的框架都找到了性能优于或可与竞争的最先进的 NAS 方法相媲美的架构，同时体积更小，将架构搜索时间减少了 50%以上。
