Prognostic Biomarker Value of Binary and Grayscale Breast Tumor Histopathology Images

Abstract

The critical prognostic importance of the grayscale texture is revealed.

The Pan-Cytokeratin Staining Intensity and Fractal Computational Analysis of Breast Tumor Malignant Growth Patterns Prognosticate the Occurrence of Distant Metastasis

“…This result was not expected in view of the fact that multifractal analysis was particularly designed for improved analysis of irregular natural forms with their typically uneven distribution of complexity. Our previous study has indicated a comparable prognostic value of monofractal and multifractal analyses of pan-tissue stained breast tumor tissue sections (28). …”

“…This was striking in view of the fact that intensity quantification is the standard approach to immunostaining evaluation, while fractal analysis is mainly considered as a method of choice for pan-tissue staining. We have previously conducted a fractal analysis of the pan-tissue H&E stained tumor histopathology specimens for the prognostic purpose (24, 28). Fractal analysis thereby performed far better on grayscale then on binary images (28) which directly opposed the results obtained in this study.…”

“…We have previously conducted a fractal analysis of the pan-tissue H&E stained tumor histopathology specimens for the prognostic purpose (24, 28). Fractal analysis thereby performed far better on grayscale then on binary images (28) which directly opposed the results obtained in this study. The prognostic outperformance of binary fractal analysis achieved here was surprising because grayscale image binarization results in a massive loss of information.…”

Effects of different preprocessing algorithms on the prognostic value of breast tumour microscopic images

“…The invasive breast cancer (BC) and inflammatory breast cancer (IBC) patient groups used in this study have been previously described in full detail, together with their power calculations for unprocessed images (Rajkovic et al ., 2016a,b). The targeted effect size by hazard ratio, variability (in standard deviations) and the event rate are essential parameters for the calculation of sample size needed for Cox regression.…”

“…This rationale is based on the hypothesis that such images contain distributed prognostic information which remain unused by either visual or computational evaluations targeting specific structures. There have been several previous reports utilising this approach for prognostic purposes (Tambasco & Magliocco, ; Tambasco et al ., ; Vujasinovic et al ., ; Kolarevic et al ., ; Rajkovic et al ., 2016a,b; Mezheyeuski et al ., ). Such unselective approach to histopathology image analysis has a potential for routine clinical application as it often outperforms the major molecular prognostic tools such as multiplex ELISA (Jia et al ., ) and gene expression microarrays (Buyse et al ., ).…”