Dendritic Learning and Miss Region Detection-Based Deep Network for Multi-scale Medical Segmentation  

作　　者：Lin Zhong Zhipeng Liu Houtian He Zhenyu Lei Shangce Gao 

机构地区：[1]Faculty of Engineering,University of Toyama,Toyama,9300887,Japan

出　　处：《Journal of Bionic Engineering》2024年第4期2073-2085,共13页仿生工程学报（英文版）

基　　金：supported by the Japan Society for the Promotion of Science(JSPS)KAKENHI under Grant JP22H03643;Japan Science and Technology Agency(JST)Support for Pioneering Research Initiated by the Next Generation(SPRING)under Grant JPMJSP2145;JST through the Establishment of University Fellowships towards the Creation of Science Technology Innovation under Grant JPMJFS2115.

摘　　要：Automatic identification and segmentation of lesions in medical images has become a focus area for researchers.Segmentation for medical image provides professionals with a clearer and more detailed view by accurately identifying and isolating specific tissues,organs,or lesions from complex medical images,which is crucial for early diagnosis of diseases,treatment planning,and efficacy tracking.This paper introduces a deep network based on dendritic learning and missing region detection(DMNet),a new approach to medical image segmentation.DMNet combines a dendritic neuron model(DNM)with an improved SegNet framework to improve segmentation accuracy,especially in challenging tasks such as breast lesion and COVID-19 CT scan analysis.This work provides a new approach to medical image segmentation and confirms its effectiveness.Experiments have demonstrated that DMNet outperforms classic and latest methods in various performance metrics,proving its effectiveness and stability in medical image segmentation tasks.

关 键 词：Medical image segmentation Dendritic learning Deep supervision Dynamic focal loss 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] R319[自动化与计算机技术—计算机科学与技术]