Diffusion MRI of cancer: From low to high b-values

Abstract

Following its success in early detection of cerebral ischemia, diffusion-weighted imaging (DWI) has been increasingly used in cancer diagnosis and treatment evaluation. These applications are propelled by the rapid development of novel diffusion models to extract biologically valuable information from diffusion-weighted MR signals, and significant advances in MR hardware that has enabled image acquisition with high b-values. This article reviews recent technical developments and clinical applications in cancer imaging using DWI, with a special emphasis on high b-value diffusion models. The article is organized in four sections. First, we provide an overview of diffusion models that are relevant to cancer imaging. The model parameters are discussed in relation to three tissue properties-cellularity, vascularity, and microstructures. An emphasis is placed on characterization of microstructural heterogeneity, given its novelty and close relevance to cancer. Second, we illustrate diffusion MR clinical applications in each of the following three categories: 1) cancer detection and diagnosis; 2) cancer grading, staging, and classification; and 3) cancer treatment response prediction and evaluation. Third, we discuss several practical issues, including selection of image acquisition parameters, reproducibility and reliability, motion management, image distortion, etc., that are commonly encountered when applying DWI to cancer in clinical settings. Lastly, we highlight a few ongoing challenges and provide some possible future directions, particularly in the area of establishing standards via well-organized multicenter clinical trials to accelerate clinical translation of advanced DWI techniques to improving cancer care on a large scale. Level of Evidence: 5 Technical Efficacy: Stage 2 J. Magn. Reson. Imaging 2019;49:23-40.

© 2018 International Society for Magnetic Resonance in Medicine.

Figures

Figure 1.

Diffusion-induced signal attenuation as a function of b-value. Three regions are highlighted in the plot with low, intermediate, and high b-values, respectively, each corresponding to a specific tissue property that can be probed by DWI. The b-values on the horizontal axis are illustrating examples and can be varied depending on the applications.

Figure 2.

An image voxel in DWI, containing intra- and extra-cellular spaces and capillary vasculature as annotated.

Figure 3.

Examples of using DWI for tumor detection. Images in each row were acquired from the same patient. (a)-(b): On a patient with gastric cancer in antrum (arrow), the contrast between the lesion (arrow) and the background tissue was much higher on the DWI with b=1000 s/mm2 (b) than on the T2W image (a). (c)-(d): T1 enhancement was observed on a patient with esophagogastric junction cancer (arrow; (c)). However, the opposite larger curvature wall also displayed strong enhancement (double arrow), making it difficult to determine the tumor border. On the DWI with b=1000 s/mm2 (d), the normal wall signal was effectively suppressed, making the cancer easily detectable (arrow). (e)-(h): On a patient with gastric cancer in antrum, the conspicuity of the tumor improved substantially when the b-value was increased from 300 s/mm2 in (e) to 1200 s/mm2 in (f). Similar improvement was observed on another patient when the b-value was increased from 500 s/mm2 in (g) to 1200 s/mm2 in (h).

Figure 4.

Examples of using DWI to detect small tumors that are not easily visible in convention MRI. (a)-(b): A gastric malignancy in lesser curvature (double arrow) was seen in the T2W image (arrow in (a)). Using DWI with b=1000s/mm2, an additional small lesion (arrow in (b)) was detected and proven as a GIST by operational pathology. (c)-(d): DWI was effective in highlighting small metastatic lesions in the small lymph nodes (c) and in the liver (d). (e)-(f): In this patient with gastric cancer, no definite sign of metastasis on the T2W image was observed (e). However, diffuse dissemination in perihepatic peritoneum (arrow) and omentum (double arrows) were clearly seen on DWI with b=1000s/mm2 (f).

Figure 5.

Maps of D, α and β from a CTRW model demonstrating good contrasts between a medulloblastoma and its surrounding normal brain tissue from an 18-month old girl. These parameter maps should be used conjointly with conventional images based on T1, T2 and FLAIR contrasts to improve tumor detection and characterization.

Figure 6.

Contrast-enhanced T1W images (top row) and diffusion-weighted images (bottom row) of an HCC patient who received RFA treatment. No obvious enhancement was seen near the post-RFA area (arrow in (a)). However, small recurrent lesions were detected in DWI (arrow in (b)) with much improved contrast. These lesions were confirmed in both contrast-enhanced T1W (c) and diffusion-weighted images (d) at one-month follow-up.

Figure 7.

FROC maps of a grade II glioma patient (upper row) and a grade IV glioblastoma patient (lower row). D,β, and μ are the FROC parameters defined in Eqs. [6] and [7] with their physical meanings explained in the text. In particular, β has been related to intra-voxel tissue heterogeneity. Differences between the two tumors can be seen in each of the three maps.

Figure 8.

Demonstration of DWI for quantitatively monitoring early change of GIST following imatinib targeted therapy in a patient with diffusely metastasis lesions in the abdomen. T2W and diffusion-weighted images are shown in the top and bottom row, respectively. Each column represents a time point: (a) and (e): pre-treatment; (b) and (f): three days; (c) and (g): one week; and (d) and (h): four weeks after initiation of the treatment. T2W images showed fused irregular tumors without appreciable change in tumor size until a later time point (d). In addition, the tumor size was difficult to measure because of the irregular shape. In contrast, the mean ADC exhibited a substantial increase only after 3 days of treatment (1.09 × 10−3 mm2/s in (e) versus 1.34 × 10−3 mm2/s in (f)). The mean ADC continued to increase for weeks during the treatment (1.83 × 10−3 mm2/s in (g) after one week and 1.96 × 10−3 mm2/s in (h) after four weeks).

Figure 9.

Demonstration of FROC parameters for monitoring early changes of a GIST patient following sunitinib targeted therapy. Images of pre-treatment baseline and two weeks after initiation of treatment are displayed in the top and bottom row, respectively. The first two columns correspond to ADC ((a) and (f)) and T2W images ((b) and (g)). The remaining three columns show the D, β and μ color maps of a lesion superimposed on the T2W image, respectively. The color bars in these three columns display the quantitative scales for the respective parameter (D in units of x10−3 mm2/s, β between 0 and 1, andμ in units of μm). The tumor size measured from the T2W images ((b) vs. (g)) did not change. ADC ((a) vs. (f)) and D ((c) vs. (h)) showed minimal change. However, β decreased substantially ((d) vs. (i)), and μ increased moderately ((e) vs. (j)) two weeks after the initiation of sunitinib targeted therapy. The results may signify potentially good response of tumors to sunitinib, which may confirm the treatment decision more quickly.

Figure 10.

Demonstration of confounding pathologic changes during targeted therapy of liver metastasis. T2W and diffusion-weighted images are shown in the top and bottom row, respectively. Each column represents a time point: (a) and (e): pre-treatment; (b) and (f): one week; (c) and (g): one month; and (d) and (h): three months after initiation of treatment. Prior to treatment, a large tumor on the right liver lobe was clearly visible with mixed signals ((a) and (e)). One week after the treatment, no appreciable tumor size change was observed in the T2W image (b). The DWI showed decreased signal (f) with the corresponding increase in ADC (not shown). At the one-month time point, the tumor size decreased (c). However, hypersignal appeared on DWI in the center of tumor (arrow in (g)), which may signify progression. At the three-month time point, however, the tumor size continued to shrink (d), suggesting that the hypersignals in (g) and (h) were most likely inner degeneration or abscess that mimics recurrence, instead of recurrence.

Figure 11.

A case of diffuse metastasis tumors to demonstrate that a small ROI can be more sensitive than WTV to detecting progressive disease. T2W and diffusion-weighted images are shown in the top and bottom row, respectively. From left to right, each column represents a time point: pre-treatment baseline, three days, one week, two weeks, four weeks, and three months after treatment, sequentially. Throughout the time course, both the mean ADC (ADCmean) and minimum ADC (ADCmin) were measured from the tumors. The values of ADCmean were 0.91 × 10−3 (g), 1.19 × 10−3 (h), 1.39 × 10−3 (i), 1.72 × 10−3 (j), 1.38 × 10−3 (k), and 0.96 × 10−3 mm2/s (l), whereas the values of ADCmin were 0.77 × 10−3 (g), 0.94 × 10−3 (h), 1.02 × 10−3 (i), 0.91 × 10−3 (j), 0.85 × 10−3 (k), and 0.95 × 10−3 mm2/s (l). A quantitative analysis showed that ADCmin was more sensitive than ADCmean during the tumor response and recurrent processes, concurrent with the increase and decrease in ADC, respectively. This illustrates the benefit of using a small ROI to focus on high DWI signal regions in this specific example.