Automatic Gleason grading of prostate cancer using quantitative phase imaging and machine learning

Abstract

We present an approach for automatic diagnosis of tissue biopsies. Our methodology consists of a quantitative phase imaging tissue scanner and machine learning algorithms to process these data. We illustrate the performance by automatic Gleason grading of prostate specimens. The imaging system operates on the principle of interferometry and, as a result, reports on the nanoscale architecture of the unlabeled specimen. We use these data to train a random forest classifier to learn textural behaviors of prostate samples and classify each pixel in the image into different classes. Automatic diagnosis results were computed from the segmented regions. By combining morphological features with quantitative information from the glands and stroma, logistic regression was used to discriminate regions with Gleason grade 3 versus grade 4 cancer in prostatectomy tissue. The overall accuracy of this classification derived from a receiver operating curve was 82%, which is in the range of human error when interobserver variability is considered. We anticipate that our approach will provide a clinically objective and quantitative metric for Gleason grading, allowing us to corroborate results across instruments and laboratories and feed the computer algorithms for improved accuracy.

Current status of artificial intelligence applications in urology and their potential to influence clinical practice

“…Nguyen et al. trained an ML classifier to automate Gleason grading of prostate cancer. The ML classifier was able to differentiate epithelial tissue and stromal tissue with an AUC of 0.97 in non‐cancer cases and 0.87 in cancer cases.…”

“…The ANNs achieved an area under the curve (AUC) of 0.974 in identifying prostate cancer, outperforming diagnosis approaches utilizing handcrafted nuclear architecture features. Nguyen et al [12] trained an ML classifier to automate Gleason grading of prostate cancer. The ML classifier was able to differentiate epithelial tissue and stromal tissue with an AUC of 0.97 in non-cancer cases and 0.87 in cancer cases.…”

Quantitative Phase Imaging for Label-Free Analysis of Cancer Cells—Focus on Digital Holographic Microscopy

“…Measurement of cell growth has been well documented for both adherent and non-adherent cells [8,9,15,[25][26][27][28]47,49,56,59,61]. By segmentation of cellular outlines, morphology and motility can also be studied [7,50,51,57,80]. Furthermore, differences in cell types can be detected if they display distinctly different morphologies or migration patterns in response to cytotoxic drugs [30,81,82].…”

Machine Learning with Optical Phase Signatures for Phenotypic Profiling of Cell Lines

“…For example, time‐lapse imaging of single cells reveals the effects on phase signals of agents inducing apoptosis , cytoskeletal disruption , or reduced proliferation . In clinically relevant applications, phase features of cells have been used to distinguish cancer from non‐cancer cells in suspension , primary cancer from metastatic cancer cells , and in clinical biopsy specimens fixed on slides . Biomedical applications of quantitative phase imaging were recently reviewed .…”