Decomposition-based framework for tumor classification and prediction of treatment response from longitudinal MRI

Abstract

Objective: In the field of radiation oncology, the benefit of MRI goes beyond that of providing high soft-tissue contrast images for staging and treatment planning. With the recent clinical introduction of hybrid MRI linear accelerators (MR-Linacs) it has become feasible to map physiological parameters describing diffusion, perfusion, and relaxation during the entire course of radiotherapy, for example. However, advanced data analysis tools are required for extracting qualified prognostic and predictive imaging biomarkers from longitudinal MRI data. In this study, we propose a new prediction framework tailored to exploit temporal dynamics of tissue features from repeated measurements. We demonstrate the framework using a newly developed decomposition method for tumor characterization. Approach: Two previously published MRI datasets with multiple measurements during and after radiotherapy, were used for development and testing: T$_2$-weighted multi-echo images obtained for two mouse models of pancreatic cancer, and diffusion-weighted images for patients with brain metastases. Initially, the data was decomposed using the novel monotonous slope non-negative matrix factorization (msNMF) tailored for MR data. The following processing consisted of a tumor heterogeneity assessment using descriptive statistical measures, robust linear modelling to capture temporal changes of these, and finally logistic regression analysis for stratification of tumors and volumetric outcome. Main Results: The framework was able to classify the two pancreatic tumor types with an \textit{area under curve} AUC of 0.999, $P<0.001$ and predict the tumor volume change with a correlation coefficient of 0.513, $P=0.034$. A classification of the human brain metastases into responders and non-responders resulted in an AUC of 0.74, $P=0.065$. Significance: A general data processing framework for analyses of longitudinal MRI data has been developed and applications were demonstrated by classification of tumor type and prediction of radiotherapy response. Further, as part of the assessment, the merits of msNMF for tumor tissue decomposition were demonstrated.

Prediction of overall survival in patients with locally advanced pancreatic cancer using longitudinal diffusion-weighted MRI

“…The current study made use of this opportunity to investigate image-based parameters derived from longitudinal DWI, exploring different methods for extraction of parameters. The msNMF method has previously been tested in patients with brain metastasis, showing borderline significant differences between responders and non-responders ( 21 ), but has not been tested as a predictor of OS in pancreatic cancer. It was included in the current study to investigate whether it could provide information useful for the prediction of OS in pancreatic cancer, as a proof of concept.…”

“…Inspired by Rahbek et al. ( 21 ), a time-trend of each percentile was extracted using a linear fit ( Figure 2G ). The slope of the linear fit and the value at fraction one were used as parameters in the statistical analysis.…”

“…DWI parameters were derived using a data-driven method recently proposed by Rahbek et al. ( 20 , 21 ). The potential advantage of data-driven approaches is that there are no initial model assumptions which could lead to biased parameters.…”

A weakly supervised NMF method to decipher molecular subtype-related dynamic patterns in breast DCE-MR images

“…The NMF method acquired a more accurate estimation of kinetic parameters than the CAM method on breast DCE-MRI data. Rahbek et al (2023) presented a tumor tissue signal decomposition using monotonous slope NMF to identify possible MRI biomarkers. However, these methods decompose tumors in an unsupervised manner without considering any clinical information.…”