Biology subjects

Oner, M. U.

Publications and source records attributed to Oner, M. U..

2 recordsLinked to original sources

bioRxiv2022

An AI-assisted Tool For Efficient Prostate Cancer Diagnosis

Pathologists diagnose prostate cancer by core needle biopsy. For low-grade and low-volume cases, the pathologists look for the few malignant glands out of hundreds within a core. They may miss the few malignant glands, resulting in repeat biopsies or missed therapeutic opportunities. This study developed a multi-resolution deep learning pipeline detecting malignant glands in core needle biopsies to help pathologists effectively and accurately diagnose prostate cancer in low-grade and low-volume cases. The pipeline consisted of two stages: the gland segmentation model detected the glands within the sections and the multi-resolution model classified each detected gland into benign vs. malignant. Analyzing a gland at multiple resolutions provided the classification model to exploit both morphology information (of nuclei and glands) and neighborhood information (for architectural patterns), important in prostate gland classification. We developed and tested our pipeline on the slides of a local cohort of 99 patients in Singapore. The images were made publicly available, becoming the first digital histopathology dataset of prostatic carcinoma patients of Asian ancestry. Our pipeline successfully classified the core needle biopsy parts (81 parts: 50 benign and 31 malignant) into benign vs. malignant. It achieved an AUROC value of 0.997 (95% CI: 0.987 - 1.000). Moreover, it produced heatmaps highlighting the malignancy of each gland in core needle biopsies. Hence, our pipeline can effectively assist pathologists in core needle biopsy analysis.

cancer biology↗

bioRxiv2021

Obtaining Spatially Resolved Tumor Purity Maps Using Deep Multiple Instance Learning In A Pan-cancer Study

Tumor purity is the proportion of cancer cells in the tumor tissue. An accurate tumor purity estimation is crucial for accurate pathologic evaluation and for sample selection to minimize normal cell contamination in high throughput genomic analysis. We developed a novel deep multiple instance learning model predicting tumor purity from H&E stained digital histopathology slides. Our model successfully predicted tumor purity from slides of fresh-frozen sections in eight different TCGA cohorts and formalin-fixed paraffin-embedded sections in a local Singapore cohort. The predictions were highly consistent with genomic tumor purity values, which were inferred from genomic data and accepted as the golden standard. Besides, we obtained spatially resolved tumor purity maps and showed that tumor purity varies spatially within a sample. Our analyses on tumor purity maps also suggested that pathologists might have chosen high tumor content regions inside the slides during tumor purity estimation in the TCGA cohorts, which resulted in higher values than genomic tumor purity values. In short, our model can be utilized for high throughput sample selection for genomic analysis, which will help reduce pathologists workload and decrease inter-observer variability. Moreover, spatial tumor purity maps can help better understand the tumor microenvironment as a key determinant in tumor formation and therapeutic response.

cancer biology↗