DocumentCode :
3463633
Title :
Application of neural networks to lesion detection in SPECT
Author :
Tourassi, Georgia D. ; Floyd, Carey E., Jr. ; Munley, Michael T. ; Bowsher, James E. ; Coleman, R. Edward
Author_Institution :
Duke Univ., Durham, NC, USA
fYear :
1991
fDate :
2-9 Nov. 1991
Firstpage :
2179
Abstract :
An artificial, single-layer, feedforward neural network was developed for lesion detection in single photon emission computed tomographic (SPECT) images. The network has 121 input nodes and one output node. A backpropagation algorithm with a sigmoid activation function is used for its supervised learning. The diagnostic performance of the neural network is studied at various noise levels for different numbers of training images using receiver operating characteristics analysis. Three noise levels (30000, 50000, and 100000 counts/slice) were used. At each noise level two training data sets (one with 40 and one with 200 simulated SPECT images) and a testing data set (200 SPECT images unknown to the trained network) were generated. The diagnostic task was the detection of a cold lesion 1.0 cm in radius at a known location of the image. The neural network trained quickly and accurately in all cases. Network convergence to a minimum in cumulative squared error was easy to achieve regardless of the noise level or the number of training images. However, the performance of the network as a decision maker for lesion detection on new images was very much affected by the number of training images. When trained with a sufficiently large number of images the performance of the network was significantly improved, particularly in the low count (high noise) case.<>
Keywords :
computerised tomography; feedforward neural nets; medical image processing; radioisotope scanning and imaging; 2 cm; backpropagation algorithm; cold lesion; cumulative squared error; diagnostic performance; input nodes; lesion detection; medical diagnostic imaging; network convergence; nuclear medicine; output node; receiver operating characteristics analysis; sigmoid activation function; single photon emission computerised tomography; single-layer feedforward neural network; supervised learning; training images; Artificial neural networks; Backpropagation algorithms; Computer networks; Feedforward neural networks; Lesions; Neural networks; Noise level; Optical computing; Single photon emission computed tomography; Supervised learning;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Nuclear Science Symposium and Medical Imaging Conference, 1991., Conference Record of the 1991 IEEE
Conference_Location :
Santa Fe, NM, USA
Print_ISBN :
0-7803-0513-2
Type :
conf
DOI :
10.1109/NSSMIC.1991.259305
Filename :
259305
Link To Document :
بازگشت