MICSI PROSTATE IMAGING BIOMARKERS
PROSTATE DWI ON A
0.55T MRI SYSTEM
MICSI denoising enables prostate imaging at prohibitively low field strengths
The limited accessibility to MR systems poses major limitation on the widespread adoption of MPMRI as a screening or diagnostic tool. Roughly 10% of MRI scanners are 3.0T and would meet the demanding signal-to-noise ratio (SNR) recommendation from PIRADS.
With MICSI denoising, we demonstrate that the SNR of the diffusion MRI, the noisiest MPMRI image, can be acquired with high SNR not only for 3.0T, but even on 0.55T.
We show that MICSI denoising can bridge the 6x SNR gap to allow for high-end diffusion imaging on the vast majority of clinical MRI systems.
Microstructure imaging for prostate diagnositics
The SNR boost from MICSI denoising does not only enable standard of care imaging on low-end systems, but enable the high SNR acquisition of diffusion weighted images at extremely long echo times. This enables the modeling diffusion properties within cellular and luminal compartments. This enables quantifying compartment-specific changes in prostate cancer (PCa) progression in:
(1) Diffusivity. ADC is one of the most critical parameters of MPMRI exam. We hypothesize that cellular and luminal ADC would outperform the clinical ADC(TE~50-90 ms), by sensing restrictions of each compartment separately.
(2) Anisotropy. Fractional Anisotropy is a critical biomarker for neuroimaging but is overlooked as a biomarker in prostate due to low SNR. The anisotropy of the stroma and lumen is greatest for low-grade PCa.
(3) Fraction. The luminal volume fraction decreases with PCa progression.
(4) Relaxometry. Compartmental relaxation times may be sensitive towards changes in the chemical composition of luminal secretions.
The SNR boost from MICSI denoising does not only enable standard of care imaging on low-end systems, but enable the high SNR acquisition of diffusion weighted images at extremely long echo times. This enables the modeling diffusion properties within cellular and luminal compartments. This enables quantifying compartment-specific changes in prostate cancer (PCa) progression in:
(1) Diffusivity. ADC is one of the most critical parameters of MPMRI exam. We hypothesize that cellular and luminal ADC would outperform the clinical ADC(TE~50-90 ms), by sensing restrictions of each compartment separately.
(2) Anisotropy. Fractional Anisotropy is a critical biomarker for neuroimaging but is overlooked as a biomarker in prostate due to low SNR. The anisotropy of the stroma and lumen is greatest for low-grade PCa.
(3) Fraction. The luminal volume fraction decreases with PCa progression.
(4) Relaxometry. Compartmental relaxation times may be sensitive towards changes in the chemical composition of luminal secretions.
Microstructure imaging for prostate diagnositics
References
1. Lemberskiy et al. Time-Dependent Diffusion in Prostate Cancer. (2017) Invest Radiol. 52, 405-411
2. Novikov et al. System, method and computer accessible medium for noise estimation, noise removal and gibbs ringing removal. (2016) US10698065B2.
3. Lemberskiy et al. System, method and computer-accessible medium for characterizing prostate microstructure using water diffusion and nuclear magnetic resonance relaxation. (2019) US11366190B2.
4. Lemberskiy et al. Characterization of prostate microstructure using water diffusion and NMR relaxation. (2018) Front Phys. 6
Through biophysical modeling of diffusion and relaxometry in prostate parameter maps will more closely reflect tissue microstructure. For instance, luminal ADC may serve as a proxy for the luminal diameter, which shrinks dramatically (~20x) with Gleason Score (GS), as the contribution cellular ADC and the partial volume fraction have been accounted for. The fractional anisotropy, FA, of stroma and lumen is the largest for low-grade PCa, which is evident on Donald Gleason’s diagram.
The quantitative nature of our modeling approach will be far more amenable to training AI classifiers than MPMRI, as the underlying biophysical parameters do not depend on scan parameters, field strength or vendor.
Towards a non-invasive analog of Gleason Score
References
1. Lemberskiy et al. Time-Dependent Diffusion in Prostate Cancer. (2017) Invest Radiol. 52, 405-411
2. Novikov et al. System, method and computer accessible medium for noise estimation, noise removal and gibbs ringing removal. (2016) US10698065B2.
3. Lemberskiy et al. System, method and computer-accessible medium for characterizing prostate microstructure using water diffusion and nuclear magnetic resonance relaxation. (2019) US11366190B2.
4. Lemberskiy et al. Characterization of prostate microstructure using water diffusion and NMR relaxation. (2018) Front Phys. 6
References
1. Lemberskiy et al. Time-Dependent Diffusion in Prostate Cancer. (2017) Invest Radiol. 52, 405-411
2. Novikov et al. System, method and computer accessible medium for noise estimation, noise removal and gibbs ringing removal. (2016) US10698065B2.
3. Lemberskiy et al. System, method and computer-accessible medium for characterizing prostate microstructure using water diffusion and nuclear magnetic resonance relaxation. (2019) US11366190B2.
4. Lemberskiy et al. Characterization of prostate microstructure using water diffusion and NMR relaxation. (2018) Front Phys. 6