Spatially varying scatter compensation for chest radiographs using a hybrid Madaline artificial neural network

We developed a hybrid artificial neural network for scatter compensation in digital portable chest radiographs. The network inputs an image region of interest (ROl), and outputs the scatter estimate at the ROl's center. We segmented each image into four regions by relative detected exposure, then trained a separate Adaline (adaptive linear element) or adaptive filter for each region. We produced a spatially varying hybrid Madaline (multiple Adaline) by combining outputs from weight matrices of different sizes trained for different durations. The network was trained with 20 patients or 1280 examples, then evaluated with another 5patients or 320 examples. Scatter estimation errors were not very different, ranging from the Adaline's 6.9% to the hybrid Madaline's 5.5%. Primary errors (more relevant to quantitative radiography techniques like dual energy imaging) were 43% for the Adaline, reduced to 27% for the Madaline, and further reduced to 19% for the hybrid Madaline. The trained weight matrices, which act like convolution filters, resembled the shape and magnitude of scatter point spread functions. All networks outperformed conventional convolution-subtraction techniques using analytical kernels. With its spatially varying neural network model, the hybrid Madaline provided the most accurate and robust estimation of scatter and primary exposures.

Duke Scholars

Author Joseph Yuan-Chieh Lo Radiology