Aim: Breast cancer prognosis is in the spotlight owing to its potentially major clinical importance in effective therapeutic management. Following our recent prognostic establishment of the fractal features calculated on binary breast tumor histopathology images, this study aimed to accomplish the first optimization of this methodology by direct comparison of monofractal, multifractal and co-occurrence algorithms in analysis of binary versus grayscale image formats.
Patients & Methods: The study included 93 patients with invasive breast cancer, without systemic treatment and a long median follow-up of 150 months.
Results: Grayscale images provided a better prognostic source in comparison to binary, while monofractal, multifractal and co-occurrence image analysis algorithms exerted a comparable performance.
Conclusion: The critical prognostic importance of the grayscale texture is revealed.
Download full-text PDF |
Source |
---|---|
http://dx.doi.org/10.2217/bmm-2016-0165 | DOI Listing |
Enter search terms and have AI summaries delivered each week - change queries or unsubscribe any time!