Automated High Uptake Regions Recognition and Lymphoma Detection Based on Fully Convolutional Networks on Chest and Abdomen PET Image

CHENG-LONG DENG, HUI-YAN JIANG, HAO-MING LI

Abstract


[18F]-Fluorodeoxyglucose Positron Emission Tomography (PET) is an essential imaging modality for the detection of the lymphomas. Generally, the lesions are detected by modeling the characteristics of abnormal areas via thresholding the degree of FDG uptake. However, it is difficult to detect lesions precisely because of inconsistent shape, discontinuous localization. Besides, the sites of normal physiological FDG uptake and normal FDG excretion (sFEPU) such as the kidneys are always interference with lymphoma detection. To address these issues, we propose a novel framework for the recognition of sFEPU and detection of lymphoma based on fully convolutional networks (FCN). FCN is used to extract high-level semantic information for lymphoma detection and meanwhile multi-scale integration in many times to refine the edge detection of sFEPU. Experimental results demonstrates the satisfactory detection and recognition accuracy compared to existing methods.

Keywords


Feature Selection, Thresholding, FCN.


DOI
10.12783/dtbh/icmsb2017/17985

Full Text:

PDF

Refbacks

  • There are currently no refbacks.