论文信息 - CT-Image Classification by Threshold Circuits

CT-Image Classification by Threshold Circuits

We present an algorithm that computes a depth-three threshold circuit for the classification of liver tissue. The circuit is calculated from a sample set S of 348 positive (abnormal findings) and 348 negative (normal liver tissue) examples by a local search strategy. The local search is based on simulated annealing with the logarithmic cooling schedule c(k) = Γ/ In (k + 2). The parameter Γ depends on S and the neighbourhood relation is determined by the classical Perceptron algorithm. The examples are fragments of DICOM CT images of size n = 14161 = 119 x 119. On test sets of 50 + 50 examples (disjoint from the learning set) we obtain a correct classification of about 97%.

[1] Alvaro L. Ronco,et al. Use of artificial neural networks in modeling associations of discriminant factors: towards an intelligent selective breast cancer screening , 1999, Artif. Intell. Medicine.

[2] Moshe Sipper,et al. A fuzzy-genetic approach to breast cancer diagnosis , 1999, Artif. Intell. Medicine.

[3] Stephen I. Gallant,et al. Perceptron-based learning algorithms , 1990, IEEE Trans. Neural Networks.

[4] Thomas Roß,et al. Feature selection for optimized skin tumor recognition using genetic algorithms , 1999, Artif. Intell. Medicine.

[5] Andreas Alexander Albrecht,et al. Stochastic Simulations of Two-Dimensional Composite Packings , 1997 .

[6] Chak-Kuen Wong,et al. On Logarithmic Simulated Annealing , 2000, IFIP TCS.

[7] David B. Fogel,et al. Evolving artificial neural networks for screening features from mammograms , 1998, Artif. Intell. Medicine.

[8] Klaus-Uwe Höffgen,et al. Computational Limitations on Training Sigmoid Neural Networks , 1993, Information Processing Letters.

[9] Y. Wu,et al. Potential usefulness of an artificial neural network for differential diagnosis of interstitial lung diseases: pilot study. , 1990, Radiology.