Co-occurring gland tensors in localized cluster graphs: Quantitative histomorphometry for predicting biochemical recurrence for intermediate grade prostate cancer

TY - CHAP

T1 - Co-occurring gland tensors in localized cluster graphs: Quantitative histomorphometry for predicting biochemical recurrence for intermediate grade prostate cancer

AU - Lee, George

AU - Sparks, Rachel

AU - Ali, Sahirzeeshan

AU - Madabhushi, Anant

AU - Feldman, Michael D.

AU - Master, Stephen R.

AU - Shih, Natalie

AU - Tomaszewski, John E.

PY - 2013

Y1 - 2013

N2 - Quantitative histomorphometry (QH), computational tools to analyze tissue histology, has become increasingly important for aiding pathologists in assessing cancer severity. In this study, we introduce a novel set of QH features utilizing co- occurring gland tensors (CGT) in localized cluster graphs to quantitatively evaluate prostate cancer (CaP) histology. CGTs offer three main advantages over previous QH features: 1) CGTs capture local glandular networks via the use of cluster graphs, whereas previous work use fully connected graphs. 2) CGTs utilize gland orientation, a feature which has not been previously used in QH. 3) CGTs provide a novel use for co-occurrence matrices by examining directional similarity in local gland networks. We extract 4 CGT features from 56 re- gions across 40 intermediate grade CaP patients and evaluated the ability of CGT features to predict biochemical recurrence (BCR) following surgery. Intermediate Gleason score 7 can- cers represent the predictive borderline for BCR cases, where 50% of cases develop BCR.We found that CGT features out- performed 5 different sets of QH features in terms of Random Forest classification (66% for CGT features versus 55% for next closest QH feature set). These results were found to be statistically significant, and suggests the use of CGT features for quantification of CaP histology.

AB - Quantitative histomorphometry (QH), computational tools to analyze tissue histology, has become increasingly important for aiding pathologists in assessing cancer severity. In this study, we introduce a novel set of QH features utilizing co- occurring gland tensors (CGT) in localized cluster graphs to quantitatively evaluate prostate cancer (CaP) histology. CGTs offer three main advantages over previous QH features: 1) CGTs capture local glandular networks via the use of cluster graphs, whereas previous work use fully connected graphs. 2) CGTs utilize gland orientation, a feature which has not been previously used in QH. 3) CGTs provide a novel use for co-occurrence matrices by examining directional similarity in local gland networks. We extract 4 CGT features from 56 re- gions across 40 intermediate grade CaP patients and evaluated the ability of CGT features to predict biochemical recurrence (BCR) following surgery. Intermediate Gleason score 7 can- cers represent the predictive borderline for BCR cases, where 50% of cases develop BCR.We found that CGT features out- performed 5 different sets of QH features in terms of Random Forest classification (66% for CGT features versus 55% for next closest QH feature set). These results were found to be statistically significant, and suggests the use of CGT features for quantification of CaP histology.

U2 - 10.1109/ISBI.2013.6556425

DO - 10.1109/ISBI.2013.6556425

M3 - Chapter

SN - 9781467364546

T3 - Proceedings - International Symposium on Biomedical Imaging

SP - 113

EP - 116

BT - Proceedings - International Symposium on Biomedical Imaging

ER -