The impeded diffusion fraction quantitative imaging assay demonstrated in multi‐exponential diffusion phantom and prostate cancer

Abstract

Purpose To demonstrate a method for quantification of impeded diffusion fraction (IDF) using conventional clinical DWI protocols. Methods The IDF formalism is introduced to quantify contribution from water coordinated by macromolecules to DWI voxel signal based on fundamentally different diffusion constants in vascular capillary, bulk free, and coordinated water compartments. IDF accuracy was studied as a function of b‐value set. The IDF scaling with restricted compartment size and polyvinylpirrolidone (PVP) macromolecule concentration was compared to conventional apparent diffusion coefficient (ADC) and isotropic kurtosis model parameters for a diffusion phantom. An in vivo application was demonstrated for six prostate cancer (PCa) cases with low and high grade lesions annotated from whole mount histopathology. Results IDF linearly scaled with known restricted (vesicular) compartment size and PVP concentration in phantoms and increased with histopathologic score in PCa (from median 9% for atrophy up to 60% for Gleason 7). IDF via non‐linear fit was independent of b‐value subset selected between b = 0.1 and 2 ms/µm2, including standard‐of‐care (SOC) PCa protocol. With maximum sensitivity for high grade PCa, the IDF threshold below 51% reduced false positive rate (FPR = 0/6) for low‐grade PCa compared to apparent diffusion coefficient (ADC > 0.81 µm2/ms) of PIRADS PCa scoring (FPR = 3/6). Conclusion The proposed method may provide quantitative imaging assays of cancer grading using common SOC DWI protocols.

Technical note: hydrogel‐based mimics of prostate cancer with matched relaxation, diffusion and kurtosis for validating multi‐parametric MRI

“…Multi-exponential diffusion decay was created by addition of lamellar vesicles (Table 1) composed of cetearyl alcohol (CA), behentrimethyl ammonium chloride (BTAC), and stearylamidopropyl dimethylamine (SA) at molar ratios of 7:1:1 ("CSB" vesicles). 30 An acrylamide/bis-acrylamide mixture is added (typically 5% PAG) with 0.3% irgacure (cross-link initiator) and photo-crosslinked for 20 min to provide mechanical stability. Two samples were prepared with 10% PAG.…”

“…and diffusion parameter values of separate components characterized in previous studies 25,26,30 are summarized in Supporting Information Tables S-2 (PVP) and S-3 (CSB). Multi-layered phantoms were assembled by sequential addition, cooling and UV curing of the 10 mL layers in 50 mL polypropylene centrifuge tubes from select materials mimicking relaxation and diffusion parameters of normal and cancerous prostate tissue.…”

“…Multi-layered phantoms were assembled by sequential addition, cooling and UV curing of the 10 mL layers in 50 mL polypropylene centrifuge tubes from select materials mimicking relaxation and diffusion parameters of normal and cancerous prostate tissue. 8,16,17,19,20,22,23,31,32 The tubes were placed in 1 L jar with water bath and studied over a month period to test for compartmental stability of relaxation and diffusion parameters. Repeated preparations were attempted for select formulations to assess reproducibility.…”

“…The realistic target tissue transverse relaxation (T 2 ) and diffusion parameter ranges were derived as population-average values across literature reports 8,16,17,19,20,22,23,31,32 with histopathology confirmation, and used multi-echo (T 2 ) and multi-b (diffusion) standard NMR acquisition methods 13,25 to fit relaxation and diffusion model parameters (e.g., Supporting Information Figure S-1.). The derived tentative target ranges and population-wide standard deviation (SD) are summarized in Supporting Information Table S-1 for normal peripheral zone (nPZ), normal transition zone (nTZ), indolent Gleason Score 6 (GS6), high-risk Gleason Score 7 (GS7), and atrophy (Atr) and depicted as boxes in Figure 1a-c.…”

“…9,[19][20][21][22] The ideal biomimetic mpMR reference would faithfully reproduce all MR parametric and spatial features of tissue. In reality, MR phantoms typically mimic selective individual properties such as T 1 or T 2 , 13 or diffusion, 13,23 or spatial structures. 24 The viable mpMRI material design needs to address multiple challenges.…”