Researchers from the University of Calcutta in India credit ArrayFire in a paper published in the Applied Soft Computing Journal. The paper is titled “Chest X-ray enhancement to interpret pneumonia malformation based on fuzzy soft set and Dempster–Shafer theory of evidence” and showcases an algorithm that is qualitatively and quantitatively improved in both accuracy and execution time over other common methods used in X-ray enhancement.
The details of the algorithm development are described in the paper. Figure 1 below shows the basic structure of the algorithm: the separate processing of the original image and its complement, the use of fussy soft sets, the use of Dempster-Shafer theory, and the ultimate creation of the enhanced image.
The results of the algorithm are shown in Figure 6 below. The bottom row is the enhanced version of the top row. Enhancement leads to clarity, enabling medical professionals to better diagnose and treat patients.
In Figure 9, we see the segmentation of the various competing enhancement methods. The method using ArrayFire in this paper is the far right (h) image. Table 3 shows that the proposed method quantitatively outperforms the 5 other methods evaluated in terms of uniformity measure, Jaccards similarity coefficient, and segmentation accuracy.
ArrayFire was used to accelerate the performance of the algorithm leading to the fastest runtime of 9 different methods. The rightmost yellow bar representing XEFSDS (ArrayFire-based) method is shown below. The lower bar indicates the fastest runtime.
Thanks to these researchers for sharing their great work with us!