Optical Spectroscopy Classification with Error Rejection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current methods for classifying optical spectroscopy data for cancer and other pathologies have limitations in sensitivity and specificity, often resulting in invasive procedures and high false-negative rates, necessitating improved diagnostic accuracy.

Innovation Solution

A method involving training classifiers to determine a rejection region for biomedical spectra data, combining classifications from multiple classifiers using support vector machines with embedded error rejection, and incorporating feature selection techniques like Sequential Floating Forward Selection and Principal Component Analysis to enhance diagnostic decision-making.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If optical spectroscopy methods are used for tissue classification, then noninvasive real-time diagnosis is achieved, but sensitivity and specificity are insufficient leading to high false-negative rates

Engineering Contradiction:
Improvenoninvasive real-time diagnosisVSAvoidsensitivity and specificity
Core Design Contradiction:
Ease of operationVSReliability

Solution Approach 1:

The patent combines multiple spectroscopy techniques (Raman, autofluorescence, fluorescence, reflectance, and elastic-scattering spectroscopy) into an integrated diagnostic system. By merging these different optical methods, the system achieves both noninvasive real-time operation and improved diagnostic reliability through complementary information from multiple spectral modalities

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The patent employs composite classification algorithms that integrate multiple machine learning approaches (support vector machines, neural networks, decision trees) to process spectroscopy data. This composite analytical framework enhances sensitivity and specificity while maintaining the noninvasive real-time diagnostic capability

Inventive Principle:
Principle #40Composite materials

2Reliability

If invasive biopsy procedures are performed to improve diagnostic accuracy, then sensitivity and specificity increase, but patient distress and healthcare costs increase

Engineering Contradiction:
Improvediagnostic accuracyVSAvoidpatient distress and surgical risk
Core Design Contradiction:
ReliabilityVSObject-affected harmful factors

Solution Approach 1:

The patent replaces the mechanical invasive biopsy procedure with an optical spectroscopy-based diagnostic system. The optical probe uses noninvasive light interaction with tissue to obtain spectral information, eliminating the need for surgical tissue removal while achieving comparable or superior diagnostic accuracy through advanced pattern recognition

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

Solution Approach 2:

The patent introduces optical spectroscopy as an intermediary diagnostic method between visual inspection and invasive biopsy. The spectroscopic analysis serves as a noninvasive mediator that provides sufficient diagnostic information to avoid unnecessary biopsies while maintaining high diagnostic reliability

Inventive Principle:
Principle #24Intermediary (Mediator)

3Measurement precision

If multiple random biopsies are taken to detect pre-cancerous changes, then detection sensitivity improves, but procedure time and complexity increase

Engineering Contradiction:
Improvedetection sensitivityVSAvoidprocedure complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent extracts the diagnostic function from multiple random biopsy procedures and concentrates it in a single optical spectroscopy measurement. The optical probe captures spectral information from the entire tissue volume in one noninvasive measurement, eliminating the need for multiple sequential biopsies and their associated complexity

Inventive Principle:
Principle #2Taking out (Extraction)

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

This approach improves the accuracy of tissue classification by identifying samples at high risk of misclassification, reducing the need for invasive procedures and enhancing diagnostic reliability through improved sensitivity and specificity.

Implementation Method 1

Some of these proposed spectroscopic techniques include Raman spectroscopy

Methodology Applied
Scientific EffectRaman spectroscopy:

Implementation Method 2

autofluorescence spectroscopy

Methodology Applied
Scientific EffectAutofluorescence spectroscopy: Fluorescence

Implementation Method 3

fluorescence spectroscopy

Methodology Applied
Scientific EffectFluorescence spectroscopy: Fluorescence

Implementation Method 4

reflectance spectroscopy

Methodology Applied
Scientific EffectReflectance spectroscopy: Reflection

Implementation Method 5

elastic-scattering spectroscopy

Methodology Applied
Scientific EffectElastic-scattering spectroscopy: Scattering

Data Source

PatentEP2862505B1Classification Techniques for Medical Diagnostics Using Optical Spectroscopy
Publication Date: 2016.11.23 TRUSTEES OF BOSTON UNIV
  • EP2862505B1 patent drawingFigure 1A~2B
  • EP2862505B1 patent drawingFigure 3
  • EP2862505B1 patent drawingFigure 4

AI summary

Mathematical/statistical pattern-recognition systems and methods to distinguish between different pathologies and benign conditions (e.g., normal or cancerous tissue) given spectra measured using optical spectroscopy such as elastic-scattering spectroscopy.