Comparison of Monofractal, Multifractal and gray level Co-occurrence matrix algorithms in analysis of Breast tumor microscopic images for prognosis of distant metastasis risk

Abstract

Breast cancer prognosis is a subject undergoing intense study due to its high clinical relevance for effective therapeutic management and a great patient interest in disease progression. Prognostic value of fractal and gray level co-occurrence matrix texture analysis algorithms has been previously established on tumour histology images, but without any direct performance comparison. Therefore, this study was designed to compare the prognostic power of the monofractal, multifractal and co-occurrence algorithms on the same set of images. The investigation was retrospective, with 51 patients selected on account of non-metastatic IBC diagnosis, stage IIIB. Image analysis was performed on digital images of primary tumour tissue sections stained with haematoxylin/eosin. Bootstrap-corrected Cox proportional hazards regression P-values indicated a significant association with metastasis outcome of at least one of the features within each group. AUC values were far better for co-occurrence (0.66-0.77) then for fractal features (0.60-0.64). Correction by the split-sample cross-validation likewise indicated the generalizability only for the co-occurrence features, with their classification accuracies ranging between 67 and 72 %, while accuracies of monofractal and multifractal features were reduced to nearly random 52-55 %. These findings indicate for the first time that the prognostic value of texture analysis of tumour histology is less dependent on the morphological complexity of the image as measured by fractal analysis, but predominantly on the spatial distribution of the gray pixel intensities as calculated by the co-occurrence features.

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

“…Monofractal and multifractal analysis of binary images was performed by use of the regular non-overlapping box counting method (FracLac plugin version 2015Sep090313a9330 for Fiji / ImageJ ) according to formulas previously described in full detail (24). The box-counting method involves overlaying of the binary image with a meshed lattice of square boxes of decreasing size (scale) ε expressed as the box size relative to image size.…”

Prediction of Chemotherapy Response in Primary Osteosarcoma by Use of the Multifractal Analysis of Magnetic Resonance Images

“…The relationship between measured continuous feature values and a binary outcome is usually evaluated by the ROC analysis (8). However, in the present study, the performances of continuous multifractal predictor values were additionally assessed by the single classification tree pattern recognition model.…”

“…These features may arise from different growth patterns of malignant cells depending on whether they are chemoresistant or chemosensitive. The previous study considering the prediction of a chemotherapy response of breast carcinoma based on multifractal analysis of microscopic histopathology images identified f(α) max parameter (13), while in prognostic studies f(α) min and D qmax were the best performers (8). The observed discrepancy can be explained by different tumor and imaging types and by differences in chemotherapy.…”

“…The primary tumors are generally the main source of information for the prediction of their drug resistance and the risk of distant metastasis occurrence (8). The relationship between measured continuous feature values and a binary outcome is usually evaluated by the ROC analysis (8).…”

“…It relies on computational analysis of medical images, often by use of fractal geometry which has been developed as a new geometrical concept with an intent to resolve the shortcomings of traditional geometry in structural analysis of complex and irregular natural objects (6). Methods based on multifractal spectra have been found useful for analysis of a wide range of medical images including ultrasound (7), histopathology (8)(9)(10), positron emission tomography (11) and magnetic resonance (MR) imaging (12). Tumor image analysis has been explored as a predictive tool in other cancer types (13), but never in osteosarcoma.…”