Personalized Tissue Classification Thresholds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for discriminating tumor tissue from normal tissue during biopsies or surgical resections lack real-time feedback and are hindered by inter-patient variance, resulting in moderate sensitivity and specificity, making them unsuitable for individual patient approaches.
Innovation Solution
A system and method that acquires and processes optical spectra from within the body to define classification thresholds relative to individual patient characteristics, allowing for efficient discrimination of normal, benign, and malignant tissues using spectroscopic measurements, with thresholds set based on reference measurements from normal tissue types.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If a database is built with spectra from many patients for classification, then the database can be used for regression to predict tissue class, but inter-patient variance hampers tissue discrimination and reduces sensitivity and specificity
Solution Approach 1:
The patent segments the classification approach by creating separate classification models for each patient rather than using a single aggregate database. Each patient's spectra are used to build their own personalized reference database, eliminating inter-patient variance interference and improving tissue discrimination accuracy for individual cases.
2Ease of operation
If spectroscopic point measurements are used for tissue classification, then the method can be applied during biopsy or surgical resection, but the sensitivity and specificity are moderate (50-85%) and results strongly vary in the literature
Solution Approach 1:
The patent applies preliminary action by collecting and analyzing multiple spectra from each patient before establishing their personalized classification threshold. This pre-processing step creates a robust reference database for each patient, enabling more accurate real-time tissue classification during biopsy or surgical procedures with higher sensitivity and specificity.
3Device complexity
If classification is performed using all-patient database, then the workflow is simplified, but inter-patient variations cause low sensitivity and moderate classification accuracy
Solution Approach 1:
The patent changes the key parameter from a fixed aggregate database to dynamic patient-specific databases. By adjusting the database composition parameter to be personalized for each patient rather than universal, the system achieves higher classification sensitivity while maintaining manageable complexity through automated individual model generation.
4Measurement precision
If individual patient spectroscopic data is collected and processed in real-time, then tissue discrimination can be tuned to the individual patient, but it is time consuming to build the database during intervention
Solution Approach 1:
The patent applies preliminary action by collecting multiple spectra from each patient before the actual biopsy or resection procedure to build their personalized classification database in advance. This pre-processing ensures that when real-time classification is needed during the intervention, the database is already established, minimizing time loss during the critical procedure while maintaining high individualized accuracy.
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 enhances the sensitivity and specificity of tissue discrimination, achieving 94% and 90% classification accuracy for normal and malignant breast tissue, and 100% accuracy for individual patients, improving the effectiveness of tissue classification during minimally invasive procedures.
Implementation Method 1
Various optical methods can be employed, e.g., diffuse reflectance spectroscopy (DRS) and autofluorescence measurement as the techniques that are most commonly investigated
Implementation Method 2
Various optical methods can be employed, e.g., diffuse reflectance spectroscopy (DRS) and autofluorescence measurement as the techniques that are most commonly investigated
Data Source
AI summary
The present invention deals with discrimination of malignant tissue from normal and benign tissue in a single patient on the basis of optical spectroscopic measurements. Starting from spectroscopic measurements in normal tissue, reference values are obtained for the normal class. With spectroscopic measurements in other tissues data points can be assigned to new class(es) when the spectral characteristics fall outside a threshold defining the reference class. Thresholds between different classes can also be defined. Finding (the transition to) malignant tissue is based on comparing the spectroscopic values to the classification threshold discriminating normal and benign versus malignant tissue. Thus, the basis of normal spectroscopic measurements is tuned to the individual patient characteristic. Discriminating the normal plus benign and malignant from that reference is more efficient compared to the reference of the all patient database.