Correlated diffusion imaging

Alexander Wong, Jeffrey Glaister, Andrew Cameron, Masoom A. Haider

Abstract

Correlated Diffusion Imaging (CDI) is a novel MRI post-processing technique designed to improve the detection and localization of prostate cancer. By analyzing the joint correlation of water diffusion signals across multiple sensitivities, CDI overcomes the limitations of standard Apparent Diffusion Coefficient (ADC) maps. In preliminary trials involving 20 patients, CDI achieved a diagnostic accuracy of 93.63% and an Area Under the ROC Curve ($A_z$) of 0.9789, significantly outperforming traditional methods. This software-based innovation offers a promising path toward reducing unnecessary biopsies and improving the treatment of aggressive tumors.

Introduction

Prostate cancer remains a leading cause of mortality among men, yet early and accurate detection is highly correlated with positive clinical outcomes. A primary obstacle in current diagnostics is the "localization" problem: identifying the exact spatial coordinates of a tumor within the prostate.

Existing imaging modalities often fall short. Transrectal ultrasound (TRUS) frequently fails to distinguish tumors from surrounding tissue, while standard T2-weighted MRI often shows weak delineation between healthy and malignant zones. Even standard Diffusion-Weighted Imaging (DWI)—which measures the Brownian motion of water—suffers from signal overlap. In many cases, the diffusion signature of a small, aggressive tumor is indistinguishable from healthy tissue using standard metrics. This ambiguity leads to a double-edged sword of clinical error: the over-diagnosis of indolent conditions leading to unnecessary biopsies, and the under-treatment of aggressive foci that are missed during screening.

Methodology

To address these limitations, researchers have developed Correlated Diffusion Imaging (CDI). While standard DWI typically calculates an Apparent Diffusion Coefficient (ADC) based on signal attenuation at a few "b-values" (sensitivities), CDI adopts a holistic, statistical framework.

Multi-gradient Acquisition

The process begins with the acquisition of multiple signals across a broad range of gradient pulse strengths and timings. Unlike standard sequences that rely on a limited set of points, this approach captures a richer dataset of water motion characteristics.

The echo-planar sequence with two gradient pulses (G1 and G2) used to dephase and rephase spins for signal acquisition.

Signal Correlation and Mixing

The core innovation of CDI lies in its "Mixer" function. Rather than treating each acquisition as an independent data point, CDI calculates the local joint correlation of signal attenuation across all sensitivities within a small spatial sub-volume.

Schematic of the CDI signal mixing process, where multiple diffusion signals (q1 through qN) are correlated to produce a high-contrast output (C).

Because cancerous tissue is characterized by high cell density, it restricts water diffusion in a highly specific manner. CDI amplifies the unique correlation "signature" of this restricted environment, making malignant zones stand out against the background of healthy tissue. Crucially, this is a post-processing innovation; it requires no new hardware and can be implemented on existing 3.0T MRI machines via software updates.

Key Results

The study evaluated CDI using a cohort of 20 patients with known prostate cancer at Sunnybrook Health Sciences Centre. The quantitative improvements over standard methods were substantial:

  • Diagnostic Accuracy: In Leave-One-Out Cross-Validation (LOOCV) trials, CDI reached 93.63% accuracy, compared to only 76.91% for standard ADC.
  • Area Under the ROC Curve ($A_z$): CDI achieved a score of 0.9789, significantly outperforming the ADC map's 0.9183.
  • Specificity: CDI demonstrated a specificity of 94.44% (vs. 76.79% for ADC), suggesting a significant potential to reduce false positives in clinical settings.

In several clinical cases, T2-weighted images showed no discernible signal change where a tumor was present. However, CDI reconstruction produced high signal intensity that clearly delineated the malignant foci, demonstrating its superior sensitivity.

Discussion and Future Work

While the preliminary results are highly promising, several hurdles remain before widespread clinical adoption. The current study was limited to a small sample size of 20 patients; larger, multi-center trials are necessary to validate these findings across diverse populations. Furthermore, future research must correlate CDI signals directly with "whole-mount histopathology"—the physical examination of tissue after surgical removal—to confirm the technique's accuracy against the true gold standard.

There is also the challenge of Benign Prostatic Hyperplasia (BPH), which can sometimes mimic cancer on MRI. Future studies will need to determine if CDI's correlation signatures can reliably distinguish between malignancy and BPH. Finally, because the physics of restricted water diffusion applies to many types of tumors, the authors suggest that CDI could eventually be adapted for the detection of pancreatic, breast, and liver cancers.

Summary updated

Abstract

Background Prostate cancer is one of the leading causes of cancer death in the male population. Fortunately, the prognosis is excellent if detected at an early stage. Hence, the detection and localization of prostate cancer is crucial for diagnosis, as well as treatment via targeted focal therapy. New imaging techniques can potentially be invaluable tools for improving prostate cancer detection and localization. Methods In this study, we introduce a new form of diffusion magnetic resonance imaging called correlated diffusion imaging, where the tissue being imaged is characterized by the joint correlation of diffusion signal attenuation across multiple gradient pulse strengths and timings. By taking into account signal attenuation at different water diffusion motion sensitivities, correlated diffusion imaging can provide improved delineation between cancerous tissue and healthy tissue when compared to existing diffusion imaging modalities. Results Quantitative evaluation using receiver operating characteristic (ROC) curve analysis, tissue class separability analysis, and visual assessment by an expert radiologist were performed to study correlated diffusion imaging for the task of prostate cancer diagnosis. These results are compared with that obtained using T2-weighted imaging and standard diffusion imaging (via the apparent diffusion coefficient (ADC)). Experimental results suggest that correlated diffusion imaging provide improved delineation between healthy and cancerous tissue and may have potential as a diagnostic tool for cancer detection and localization in the prostate gland. Conclusions A new form of diffusion magnetic resonance imaging called correlated diffusion imaging (CDI) was developed for the purpose of aiding radiologists in cancer detection and localization in the prostate gland. Preliminary results show CDI shows considerable promise as a diagnostic aid for radiologists in the detection and localization of prostate cancer.
BibTeX
@article{wong2013correlated,
  author = {Alexander Wong and Jeffrey Glaister and Andrew Cameron and Masoom A. Haider},
  title = {{Correlated diffusion imaging}},
  journal = {BMC Medical Imaging},
  year = {2013},
  doi = {10.1186/1471-2342-13-26},
}
Figure