H-ProMed: Ultrasound image segmentation based on the evolutionary neural network and an improved principal curve

The purpose of this work is to develop a method for accurate and robust prostate segmentation in transrectal ultrasound (TRUS) images. These images are difficult to segment due to missing/ambiguous boundary between the prostate and neighboring structures, the presence of shadow artifacts, as well as the large variability in prostate shapes. This paper develops a novel hybrid method for TRUS prostate segmentation by combining an improved principal curve-based method with an evolutionary neural network; the former for achieving the data sequences while and the latter for improving the smoothness of the prostate contour. Both qualitative and quantitative experimental results showed that our proposed method achieved superior segmentation accuracy and robustness as compared to state-of-the-art methods. The average Dice similarity coefficient (DSC), Jaccard similarity coefficient (Ω), and accuracy (ACC) of prostate contours against ground-truths were 96.8%, 95.7%, and 96.4%, and the DSC of around 92% and 95% for other deep learning and hybrid methods, respectively.