Longitudinal Biomarker Assessment with Acquisition Variability Modeling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for automatically tracking the longitudinal evolution of biomarkers are hindered by variability in acquisition and processing techniques, leading to unreliable estimates due to differences in image acquisition protocols and hardware, which complicates the detection of subtle changes in biological objects like brain lesions.
Innovation Solution
A method and system that quantify longitudinal biomarker evolution by calculating statistical parameters to account for acquisition technique variability, generate synthetic data distributions, and use bootstrapping to fit and compare fitting parameters against a reference, providing a robust estimate of biomarker changes over time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automatic estimation methods are used to track longitudinal biomarker evolution, then productivity is improved, but measurement precision deteriorates due to variability in acquisition and processing techniques
Solution Approach 1:
The patent transforms the measurement problem by changing parameters - specifically, it models the biomarker values not as fixed points but as distributions characterized by mean and standard deviation. This parameter transformation allows the system to accommodate acquisition variability while maintaining longitudinal tracking capability, resolving the contradiction between automation and precision.
Solution Approach 2:
The patent introduces an intermediary statistical model that mediates between the raw measurements and the longitudinal evolution assessment. This intermediary layer absorbs the variability from different acquisition techniques and processing methods, allowing automated comparison while preserving measurement precision through proper statistical handling.
2Measurement precision
If statistical parameters are calculated to account for acquisition variability, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent creates a simplified representation (copy) of the complex measurement problem through synthetic data generation. By generating synthetic biomarker values that reflect the statistical distribution of real measurements, the system can assess longitudinal evolution without repeatedly processing complex real data, thus improving precision while managing complexity.
Solution Approach 2:
The patent performs preliminary statistical characterization of acquisition variability before conducting the longitudinal assessment. By pre-calculating standard deviations and creating synthetic distributions in advance, the system prepares the data in a form that simplifies subsequent longitudinal comparison, reducing the complexity burden during the actual measurement process.
3Measurement precision
If synthetic data distributions are generated from measured values, then measurement precision is improved through statistical comparison, but loss of time increases due to multiple bootstrapping rounds
Solution Approach 1:
The patent applies partial bootstrapping - performing a limited number of bootstrap iterations rather than exhaustive sampling. This partial action provides sufficient statistical significance assessment for clinical decision-making while dramatically reducing processing time. The system achieves the necessary precision without the time cost of complete bootstrap analysis.
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
The invention relates to a system (200) and method (100) for evaluating a longitudinal evolution of a quantitative biomarker measured for a biological object (210), the method (100) comprising: - receiving (101), for said biological object (210), longitudinal measurements of said quantitative biomarker, wherein N measurements M_1,...,M_N are received, wherein each measurement M_i, with i = 1,...,N, is performed at a timepoint T_i according to an acquisition technique A_i, wherein if i ≠ j, then T_i ≠ T_j ∀i,j ∈ {1,...,N}, wherein for each measurement M_i, a value V_i is obtained for said biomarker; - calculating (102), for each measurement M_i, a statistical parameter, which characterizes a variability of the used acquisition technique A_i, said statistical parameter being a measure of a statistical distribution of possible values for the biomarker when carrying out the measurement M_i according to said acquisition technique; - for each obtained biomarker value V_i, using the calculated statistical parameter for generating (103) synthetic data whose distribution follows the statistical distribution of possible values for the biomarker, wherein each synthetic data represents a simulated value for said biomarker and is assigned the timepoint T_i at which the measurement M_i was performed; - performing (104) several bootstrapping rounds, wherein, for each bootstrapping, one or several values among the measured value V_i and the biomarker simulated values generated for said measured value V_i are randomly sampled for each timepoint T_i; - for each bootstrapping, fitting (105) the sampled values by means of a fitting function; - for each fitting, extracting (106) a fitting parameter of the fitting function; - evaluating (107) the longitudinal evolution of the quantitative biomarker by statistically comparing the extracted fitting parameters against a reference value.