3D Imaging Rapidly Diagnoses Aggressive Prostate Cancer
By HospiMedica International staff writers Posted on 04 Jan 2022 |
Image: Screenshot of two 3D prostate biopsy samples (Photo courtesy of Xie et al/Cancer Research)
Deep-learning (DL) can enable tissue microstructures to be volumetrically segmented to improve prediction of aggressive prostate cancer, according to a new study.
The new technique, called image-translation-assisted segmentation in 3D (ITAS3D), developed at the University of Washington (UW; Seattle, USA), and Case Western Reserve University (CWRU; Cleveland, OH, USA), enables annotation-free, biomarker-based interpretation of glandular features, without requiring immunolabeling. Instead, it implements non-destructive 3D pathology and computational analysis of whole prostate biopsies, which are labeled with a rapid, inexpensive fluorescent analog of standard hematoxylin and eosin staining.
The computational 3D approach was applied to 300 biopsies extracted from 50 archived radical prostatectomy specimens, of which 118 contained cancer. The results showed that 3D features in the prostate cancer biopsies were superior to 2D features for risk stratification, and provided more information on the complex tree-like structure of the glands, thus increasing the likelihood that the computer would correctly predict a cancer's aggressiveness. The study was published on December 1, 2021, in Cancer Research.
“The 3D glandular features in cancer biopsies were superior to corresponding 2D features for risk stratification of low- to intermediate-risk prostate cancer patients based on their clinical biochemical recurrence outcomes,” concluded lead author Weisi Xie, PhD, of UW, and colleagues. “An end-to-end pipeline for DL-assisted computational 3D histology analysis of whole prostate biopsies shows that nondestructive 3D pathology has the potential to enable superior prognostic stratification of prostate cancer patients.”
The microscopic structure of the prostate gland is the basis for prognostic grading by pathologists; however, prostate cancer treatment planning is largely dependent upon examination of core-needle biopsies. Since the interpretation of the convoluted 3D glandular structure via visual inspection of a limited number of 2D histology sections is often unreliable, both under- and over-treatment of such patients is common.
Related Links:
University of Washington
Case Western Reserve University
The new technique, called image-translation-assisted segmentation in 3D (ITAS3D), developed at the University of Washington (UW; Seattle, USA), and Case Western Reserve University (CWRU; Cleveland, OH, USA), enables annotation-free, biomarker-based interpretation of glandular features, without requiring immunolabeling. Instead, it implements non-destructive 3D pathology and computational analysis of whole prostate biopsies, which are labeled with a rapid, inexpensive fluorescent analog of standard hematoxylin and eosin staining.
The computational 3D approach was applied to 300 biopsies extracted from 50 archived radical prostatectomy specimens, of which 118 contained cancer. The results showed that 3D features in the prostate cancer biopsies were superior to 2D features for risk stratification, and provided more information on the complex tree-like structure of the glands, thus increasing the likelihood that the computer would correctly predict a cancer's aggressiveness. The study was published on December 1, 2021, in Cancer Research.
“The 3D glandular features in cancer biopsies were superior to corresponding 2D features for risk stratification of low- to intermediate-risk prostate cancer patients based on their clinical biochemical recurrence outcomes,” concluded lead author Weisi Xie, PhD, of UW, and colleagues. “An end-to-end pipeline for DL-assisted computational 3D histology analysis of whole prostate biopsies shows that nondestructive 3D pathology has the potential to enable superior prognostic stratification of prostate cancer patients.”
The microscopic structure of the prostate gland is the basis for prognostic grading by pathologists; however, prostate cancer treatment planning is largely dependent upon examination of core-needle biopsies. Since the interpretation of the convoluted 3D glandular structure via visual inspection of a limited number of 2D histology sections is often unreliable, both under- and over-treatment of such patients is common.
Related Links:
University of Washington
Case Western Reserve University
Latest AI News
- AI-Powered Algorithm to Revolutionize Detection of Atrial Fibrillation
- AI Diagnostic Tool Accurately Detects Valvular Disorders Often Missed by Doctors
- New Model Predicts 10 Year Breast Cancer Risk
- AI Tool Accurately Predicts Cancer Three Years Prior to Diagnosis
- Ground-Breaking Tool Predicts 10-Year Risk of Esophageal Cancer
- AI Tool Analyzes Capsule Endoscopy Videos for Accurately Predicting Patient Outcomes for Crohn’s Disease