Prostate Cancer Detection in MRI Using Random Forest

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current prostate cancer diagnosis methods, particularly biopsies, often miss clinically significant lesions due to their random nature, leading to false negatives or overdiagnosis, and manual contouring of prostate MR images is labor-intensive and prone to variability, limiting accurate automated segmentation.

Innovation Solution

The development of computer-aided diagnosis (CAD) systems using Random Forest classifiers for detecting prostate cancer in multi-parametric MRI scans and deep learning methods for automatic prostate segmentation, incorporating features like Haralick texture features, instance-level weighting, and holistically nested networks to improve detection accuracy and segmentation efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If manual contouring is used for prostate segmentation, then segmentation can be performed with simple methods, but it is labor intensive and prone to inter- and intra-observer variance

Engineering Contradiction:
Improvesegmentation method simplicityVSAvoidsegmentation consistency
Core Design Contradiction:
Ease of manufactureVSReliability

Solution Approach 1:

The patent replaces manual mechanical contouring with an automated image processing system that uses gradient computation and edge detection algorithms. The system automatically identifies prostate boundaries by detecting intensity gradients in MRI images, eliminating the need for manual slice-by-slice contouring while improving consistency and reducing observer variance.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Ease of operation

If random biopsy is used for prostate cancer diagnosis, then biopsy procedure is simple, but it can miss clinically significant lesions and result in false negatives

Engineering Contradiction:
Improvebiopsy procedure simplicityVSAvoidlesion detection accuracy
Core Design Contradiction:
Ease of operationVSMeasurement precision

Solution Approach 1:

The patent introduces an intermediate imaging and analysis system between the biopsy procedure and diagnosis. The system uses multi-parametric MRI imaging combined with automated segmentation and Random Forest classification to identify and prioritize suspicious lesions before biopsy, serving as a mediator that guides the biopsy process to target clinically significant lesions while maintaining procedural simplicity.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Productivity

If automated segmentation is used for prostate MR images, then efficiency and reproducibility are improved, but it is challenging due to heterogeneous intensity distribution and complex anatomical structures

Engineering Contradiction:
Improvesegmentation efficiencyVSAvoidsegmentation algorithm complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent divides the prostate segmentation task into multiple processing stages: initial boundary detection using gradient methods, refinement through multi-parametric MRI fusion, and final classification using Random Forest algorithms. This segmented approach breaks down the complex heterogeneous intensity problem into manageable steps, improving both efficiency and handling of anatomical variability.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent employs multi-parametric MRI imaging that captures multiple tissue properties (T2-weighted, diffusion-weighted, dynamic contrast-enhanced) simultaneously. This multi-functional imaging approach provides comprehensive tissue characterization that works across diverse anatomical structures and intensity distributions, enabling robust automated segmentation despite heterogeneity.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentEP3923237A1Detection of prostate cancer in multi-parametric MRI using random forest
Publication Date: 2021.12.15 THE GOVERNMENT OF THE UNITED STATES OF AMERICA AS REPRESENTED BY THE SECRETARY DEPARTMENT OF HEALTH & HUMAN SERVICES
  • EP3923237A1 patent drawingFigure 1
  • EP3923237A1 patent drawingFigure 2
  • EP3923237A1 patent drawingFigure 3

AI summary

Disclosed prostate computer aided diagnosis (CAD) systems employ a Random Forest classifier to detect prostate cancer. System classify individual pixels inside the prostate as potential sites of cancer using a combination of spatial, intensity and texture features extracted from three sequences. The Random Forest training considers instance-level weighting for equal treatment of small and large cancerous lesions and small and large prostate backgrounds. Two other approaches are based on an AutoContext pipeline intended to make better use of sequence-specific patterns. Also disclosed are methods and systems for accurate automatic segmentation of the prostate in MRI. Methods can include both patch-based and holistic (image-to-image) deep learning methods for segmentation of the prostate. A patch-based convolutional network aims to refine the prostate contour given an initialization. A method for end-to-end prostate segmentation integrates holistically nested edge detection with fully convolutional networks. HNNs automatically learn a hierarchical representation that improve prostate boundary detection.