Dual-Modal Lesion Classification Using PET/CT Normalization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image-based lesion classification methods, particularly for prostate cancer, lack accuracy and rely heavily on invasive biopsies, necessitating improved non-invasive image-based classification techniques.
Innovation Solution
A system utilizing dual-modal CT and PET images, registered and normalized to account for physiological variations, extracts a set of image features for accurate lesion classification, employing machine learning architectures to determine lesion severity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If biopsy-based diagnosis is used, then diagnostic accuracy is improved, but patient invasiveness increases
Solution Approach 1:
The patent replaces the mechanical biopsy procedure with a purely image-based classification system using machine learning. The system processes CT and PET images through trained models to classify lesions, eliminating the need for physical tissue sampling while maintaining diagnostic accuracy.
Solution Approach 2:
The patent introduces machine learning models as intermediaries between imaging data and diagnostic conclusions. These models learn from training datasets containing labeled lesions and serve as mediators that translate image features into accurate classifications without requiring invasive procedures.
2Object-affected harmful factors
If image-based classification is used, then patient invasiveness is reduced, but diagnostic accuracy deteriorates
Solution Approach 1:
The patent merges multiple imaging modalities (CT and PET) and combines them with machine learning analysis. By integrating anatomical information from CT and functional/metabolic information from PET, the system achieves diagnostic accuracy comparable to biopsy while remaining non-invasive.
Solution Approach 2:
The patent transforms image data into meaningful classification parameters through machine learning feature extraction. The models identify and utilize relevant imaging parameters and patterns that correlate with lesion characteristics, enabling accurate non-invasive classification.
3Measurement precision
If multiple imaging modalities are combined, then diagnostic accuracy is improved, but device complexity increases
Solution Approach 1:
The patent employs a unified machine learning framework that can process multiple imaging modalities (CT, PET) through a single integrated system. The machine learning models serve as universal processors that handle different image types and extract relevant features, reducing the operational complexity despite combining multiple modalities.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A system (100) for classifying lesions comprising an image providing unit (101) for providing a dual-modal image of an imaging region is provided, the dual-modal image comprising a CT image (10, 10') and a registered PET image (20'), the imaging region including a tissue region of interest comprising a lesion and a reference tissue region. The system further comprises an identifying unit (102) for identifying, in the CT and the PET image, a respective lesion image segment and, in the PET image, a reference image segment. Moreover, the system comprises a normalizing unit (103) for normalizing the lesion image segment in the PET image with respect to the reference image segment, an image feature extracting unit (104) for extracting image feature values from both lesion image segments, and a classifying unit (105) for classifying the lesion based on the extracted values.