Tumor likelihood estimation on MRI prostate data by utilizing k-Space information

2024-06-04

arXiv.org

AI Summary

Accurate and efficient MRI-based risk assessment of prostate cancer remains challenging, particularly due to limitations in conventional image-domain reconstruction and apparent diffusion coefficient (ADC) map computation. Method: We propose an end-to-end k-space–aware classification framework that directly processes raw complex-valued k-space data—including phase information—bypassing image reconstruction and ADC calculation. The method integrates digital undersampling, PCA-based coil compression, and GRAPPA-free reconstruction to enable near-real-time inference. Contribution/Results: This work is the first to systematically demonstrate the discriminative value of full k-space information—especially phase—for prostate cancer classification. On a clinical cohort, the model achieves an AUROC of 86.1% ± 1.8% with fully sampled k-space and maintains robust performance (71.4% ± 2.9% AUROC) under 16× acceleration. By eliminating reconstruction and post-processing delays, the framework significantly reduces total scan-to-diagnosis time, establishing a novel paradigm for rapid, precise clinical diagnosis.

Technology Category

• Machine Learning: Kernel Methods
• Computer Vision: Multi-modal Vision
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Security and Privacy: Data transparency and provenance
• User Modeling, Personalization and Recommendation: User privacy protection in personalized systems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

We present a novel preprocessing and prediction pipeline for the classification of magnetic resonance imaging (MRI) that takes advantage of the information rich complex valued k-Space. Using a publicly available MRI raw dataset with 312 subject and a total of 9508 slices, we show the advantage of utilizing the k-Space for better prostate cancer likelihood estimation in comparison to just using the magnitudinal information in the image domain, with an AUROC of $86.1%pm1.8%$. Additionally, by using high undersampling rates and a simple principal component analysis (PCA) for coil compression, we reduce the time needed for reconstruction by avoiding the time intensive GRAPPA reconstruction algorithm. By using digital undersampling for our experiments, we show that scanning and reconstruction time could be reduced. Even with an undersampling factor of 16, our approach achieves meaningful results, with an AUROC of $71.4%pm2.9%$, using the PCA coil combination and taking into account the k-Space information. With this study, we were able to show the feasibility of preserving phase and k-Space information, with consistent results. Besides preserving valuable information for further diagnostics, this approach can work without the time intensive ADC and reconstruction calculations, greatly reducing the post processing, as well as potential scanning time, increasing patient comfort and allowing a close to real-time prediction.

Problem

• Improving prostate cancer likelihood estimation using k-Space MRI data
• Reducing reconstruction time via PCA coil compression and undersampling
• Enabling near real-time prediction without time-intensive ADC calculations

Innovation

• Utilizes k-Space data for tumor likelihood estimation
• Employs PCA for fast coil compression
• Reduces scan time via digital undersampling