A combination of MRI diffusion-derived vessel density (DDVD) and slow diffusion coefficient (SDC) can reliably diagnose liver hemangioma: a testing of three centers’ data

Introduction

Liver hemangioma (HG) has an incidence rate ranging between 0.4% and 20.0% and is commonly discovered during any abdominal imaging work-up (1-3). From a histopathological perspective, these neoplasms are characterized by cavernous venous spaces delineated by a lining of vascular endothelial cells and interspersed with connective tissue septa. The primary subtypes include cavernous HG, capillary HG, and sclerosing HG. The principal criterion for this classification is the extent of fibrous tissue present within the body of HG. Cavernous HG represents the most frequent subtype with the presence of larger vascular spaces coupled with a low quantity of connective tissue, and is closely aligned with the typical imaging profile of liver HG. Capillary HGs, also known as flash-filling or rapidly-filling HGs, account for approximately 16% of all liver HGs. This subtype is notably more prevalent in HG measuring less than 1 cm in diameter. Hyalinized or sclerosed HG is unusual and is believed to represent the end stage of a HG. Due to the replacement of the vascular spaces by fibrotic tissue, it is virtually impossible to propose a definitive diagnosis for sclerosed HG based on imaging, thus pathologic proof is necessary.

Morphologically, liver HGs are a well-defined lesion with round or lobulated margins. On T1-weighted magnetic resonance (MR) images, liver HGs display low signal intensity, and on T2-weighted MR images liver HGs show high signal due to the long T2 of its blood-filled vascular channels. In some cases, a liver HG can be diagnosed based on typical imaging features without the need for a contrast-enhanced scan. However, a substantial portion of liver HG shows apparent diffusion coefficient (ADC) and T2 overlapping with those of HCC and liver metastasis (4-13). When magnetic resonance imaging (MRI) is the first-line examination for the liver, for the majority of HG cases, the diagnosis is established with the application of a contrast-enhanced imaging (14).

The diffusion-weighted imaging (DWI) derived surrogate biomarker diffusion-derived vessel density (DDVD) works on the principle that on spin-echo type echo-planar-imaging DWI, blood vessels (including micro-vessels) show high signal when there is no motion probing gradient (b=0 s/mm²), while they show low signal even when very low b-values (such as b=1 or 2 s/mm²) are applied. Thus, the signal difference between images when the motion probing gradient is off and on reflects the extent of tissue vessel density. DDVD is derived from the equation (15-17):

DDVD(b0b2)=Sb0ROIarea0−Sb2ROIarea2[unit:arbitraryunit(au)/pixel][1]

where ROIarea0 and ROIarea2 refer to the number of pixels in the selected region-of-interest (ROI) on b=0 and b=2 DWI, respectively. Sb0 refers to the measured total signal intensity within the ROI when b=0, and Sb2 refers to the measured total signal intensity within the ROI when b=2, thus Sb/ROIarea equates to the mean signal intensity within the ROI. Sb2 and ROIarea2 can also be approximated by other low b-values (such as b=10 s/mm²) DWI. DDVD is useful as an imaging biomarker in diverse clinical scenarios. For the liver, DDVD analysis demonstrates liver parenchyma has an age-dependent decrease of micro-perfusion in healthy men and women (18,19). This agrees with the known physiological age-dependent reduction in liver blood flow which has been well documented using a variety of technical methods including histology, dye dilution, and indicator clearance. DDVD is a useful parameter for distinguishing between livers with and without fibrosis, and livers with severer fibrosis tend to have even lower DDVD measurements than those with milder liver fibrosis (15,17). Li et al. applied DDVD to assess the perfusion of hepatocellular carcinoma (HCC), showing higher perfusion of HCC relative to adjacent liver parenchyma (20). When an echo time (TE) of 59 ms [repetition time (TR) =1,600 ms, free breathing acquisition] and b=0, 2 mm²/s were used for the DDVD calculation of 26 cases of HCC, the mean DDVD_HCC/DDVD_liver ratio was 1.42. This value agrees well with the perfusion computed tomography (CT) blood volume literature results median ratio of 1.38 (21). With DDVD analysis, Zheng et al. (22) demonstrated that per unit micro-circulation of the spleen is decreased in viral hepatitis-b liver fibrosis patients. This is consistent with, for example, the report of Gitlin et al. (23) with analysis of the washout curves of Xenon 133 injected in the splenic artery in patients with liver cirrhosis and portal hypertension. Among the patients, splenic blood flow, expressed as mL per 100 g of splenic tissue, was decreased. Zheng et al. (24) recently reported that liver focal nodular hyperplasia (FNH) has a lower DDVD value than liver malignant lesions [HCC and metastases (Mets)]. DDVD analysis alone allowed the separation of FNH and malignant lesions with receiver operating characteristic area under the curve (ROAUC) of around 0.9. While liver simple cysts show DDVD value close to 0 (25,26), liver HG shows very high DDVD value. Hu et al. (4) tested the differentiating of liver HG from liver mass-forming lesions (MFLs) with two small-sized datasets totaling 22 HG and 28 MFLs (HCC n=24 and FNH n=4). Solely based on DDVD pixelwise map (DDVDm), a correct classification was made by a trained reader in 90.9% of the HGs (among them 77.7% with diagnostic confidence) and 96.4% of the MFLs (among them 85.7% with diagnostic confidence), suggesting the value of DDVDm in diagnosing HG.

Most recently, a novel metric ’slow diffusion coefficient (SDC)’ was proposed to measure tissue slow diffusion (5):

SDC=[S(b1)−S(b2)]/(b2−b1)(unit:au/s)[2]

where b₁ and b₂ refer to a high b-value (e.g., 400 mm²/s) and a higher b-value respectively (e.g., 600 mm²/s), and S(b₁) and S(b₂) denote the image signal-intensity acquired at the high b-value and the higher b-value respectively. With the conventional approach, it has been reported that the spleen has a much lower ADC than the liver, HCCs have a lower ADC than adjacent liver parenchyma, and liver simple cysts have a higher ADC than liver HGs. On the other hand, with SDC analysis, Xu et al. (5) reported that the spleen has a faster diffusion than the liver, HCCs have a faster diffusion than liver parenchyma, and liver HGs have a faster diffusion than liver simple cysts. The liver and spleen have a similar amount of blood perfusion, the spleen is waterier than the liver, and the spleen tissue has a higher contrast-enhanced CT extracellular volume fraction than the liver (27). HCCs are mostly associated with increased blood supply and increased proportion of arterial blood supply along with edema. It is more reasonable with SDC results that the spleen and HCC have a faster diffusion than liver parenchyma. Due to the ‘flushing’ of blood flow inside a HG, it is also more reasonable with SDC results that the diffusion of liquid in liver HGs is faster than the more ‘static’ liquid of the liver cysts. Moreover, abscess liquid has been reported to be of low ADC with diffusion restriction which appears to be unreasonable (28). SDC measure suggests that liver abscess liquid may have faster diffusion than the adjacent liver parenchyma (29).

Since the DDVDm-based diagnosis of liver HG was not perfect (4), and liver HG shows very high SDC signal (5), we hypothesize that a combination of DDVDm and SDC pixelwise map (SDCm) can further increase the diagnostic confidence for liver HG. In this study, we tested whether a combination of DDVDm and SDCm can reliably differentiate liver HG from MFL. We present this article in accordance with the STARD reporting checklist (available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-424/rc).

Methods

Study materials

There were three datasets of historical liver DWI samples initially not collected for the purpose of HG evaluation (Figure 1). The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was approved by the institutional ethics committees of Sun Yat-sen Memorial Hospital (No. S1709442250839) and Zhujiang Hospital (No. 2017-YXZDK-002), with informed consent obtained from individual participants. For dataset-1 (MRI scanner: Ingenia, 3.0-T, Philips Healthcare, Best, Netherlands), an intravoxel incoherent motion (IVIM) imaging DWI sequence was acquired with TR of 1,600 ms (free breathing), TE of 59 ms, and an acquisition spatial resolution of 3.02×3.11×7 mm³. DDVDm was reconstructed with b=0 [number of excitations (NEX) =5] and b=2 (NEX =5) s/mm² images. SDCm was with b=400 (NEX =2) and b=600 (NEX =2) s/mm² images. For dataset-2 (MRI scanner: Vida Magneton, 3.0-T, Siemens Healthineers, Erlangen, Germany), an IVIM DWI sequence was acquired with TR of 2,500 ms (respiratory gating), TE of 84 ms, and an acquisition spatial resolution of 2.734×2.734×5 mm³. DDVDm was reconstructed with b=0 and b=10 s/mm² images (both NEX =1), and SDCm was with b=500 and b=800 s/mm² images (both NEX =3). For dataset-3 (MRI scanner, Avanto, 1.5-T, Siemens Healthineers), an IVIM DWI sequence was acquired with TR of 6,900 ms (respiratory gating), TE of 81 ms, and an acquisition spatial resolution of 2.478×1.979×6 mm³. DDVDm was reconstructed with b=0 and b=50 s/mm² images (a lower non-zero b-value image was not available), and SDCm was with b=700 and b=900 s/mm² images, NEX was 2 for all images. Dataset-1 consisted of 17 HGs, 35 HCCs, 6 intra-hepatic cholangiocarcinomas (ICCs), 2 Mets, and 7 FNHs. Dataset-2 consisted of 7 HGs, 56 HCCs, 4 ICCs, 14 Mets, and 8 FNHs. Dataset-3 consisted of 8 HGs, 12 HCCs, 2 ICCs, 13 Mets, and 1 FNH (Figure 1). HGs were counted lesion-by-lesion, while Mets mostly had multiple lesions per case and were counted case-by-case. All HCCs, ICCs, and FNHs were single lesion per case on liver imaging in this study. FNH, HCC, and ICC all had tissue confirmation. HGs were diagnosed with typical contrast-enhanced imaging appearances and/or with pathological confirmation. The diagnosis of Mets was based on histopathology or a combination of complete patient history and typical imaging features.

Figure 1 Patient enrolment flow diagram. FNH, focal nodular hyperplasia; HCC, hepatocellular carcinoma; HG, hemangioma; ICC, intra-hepatic cholangiocarcinoma; Mets, metastases.

Image analysis

Two readers jointly read the images and made the diagnosis. Reader-1 (C.Y.L.) was a senior trainee in radiology, and reader-2 (Y.X.J.W.) was a specialist radiologist. DWI images were used to locate the site and size of the lesions but not used for diagnosis, instead diagnosis was solely based on DDVDm and SDCm. DWI images were also used to confirm that no major ‘position-shift’ occurred for the two DWI images to reconstruct DDVDm, and the two DWI images to reconstruct SDCm. For dataset-1 and dataset-2, a diagnostic decision for a lesion was made with five possible choices: (I) HG with confidence; (II) HG without confidence (lesion features suggesting HG but the diagnosis could not be made firmly); (III) solid MFL with confidence; (IV) solid MFL without confidence (lesion features suggesting MFL but the diagnosis could not be made firmly); and (V) undecided. Therefore, readers were not asked to differentiate between FNH, HCC, Mets, and ICC as they were grouped together as MFL with mainly solid content. The image quality was notably noisier for dataset-3 than for dataset-1 and dataset-2. For dataset-3, a diagnostic decision was made for a lesion with three choices: (I) suggesting HG; (II) suggesting MFL; and (III) undecided.

The first step of lesion signal analysis was based on SDCm, and then the suggestive classifications were further diagnosed with DDVDm. On SDCm, a liver HG was expected to show ‘liquid signal’, which is similar to or higher than the gallbladder signal and gastric liquid signal (5). Liver HG was expected to show higher signal than the kidneys on SDCm. When necessary to help visualization, the color of kidney signals was adjusted to be reddish-orange color but not saturated, and then HG signal would appear ‘red’ or saturated with the color scheme applied in the current study. The DDVDm difference between liver HG and MFL was described earlier (4). Liver HGs generally show substantially higher DDVD signal relative to background liver parenchyma (4). Note that, only a combination of DDVDm and SDCm can suggest the diagnosis of HG, as liquid signal on SDCm and very low signal on DDVDm would suggest liver cyst (25). For dataset-3, since DDVD was calculated with b=0 and b=50 s/mm² images and liver cysts might appear high signal on DDVDm with these b-values (25), thus all liver cysts in dataset-3 were initially excluded from analysis in this study.

Results

Visual differences between liver HG and MFL are shown in Figures 2-9. HG typically showed high signal on DDVDm and very high liquid signal on SDCm. MFL typically showed iso- or slightly high signal on DDVDm and lower than liquid signal on SDCm. A number of large malignant tumors showed very heterogeneous signals, with necrotic liquid showing focal very high signals on SDCm (Figures 5-7).

Figure 2 A patient from dataset-1 with two HGs. (A) CE-T1W. The two HGs (arrows) show typical peripheral nodular enhancement. (B) DWI b=0 image. The larger lesion (white arrow) shows very high signal, while the smaller lesion shows modestly high signal (pink arrow). (C) DDVDm. (D) SDCm. The larger lesion (white arrows) shows typical HG signal (very high signal on both DDVDm and SDCm). For the smaller lesion, its signal also agrees with HG. As described in (4), the thick rim for the smaller lesion (pink arrows) suggests HG. The smaller lesion also shows very high signal on SDCm. On SDCm, the two HGs (arrows) show liquid signal (comparable to the gastric liquid signal, thick arrow). Thus, a combination of the DDVD and SDC confirms the HG diagnosis for these two lesions. Color scale: DDVD: −130 to 500 au/pixel; SDC: −1,000 to 14,000 (au/s). Au, arbitrary unit; CE-T1W, contrast-enhanced T1-weighted image; DDVD, diffusion-derived vessel density; DDVDm, DDVD pixelwise map; DWI, diffusion-weighted imaging; HG, hemangioma; SDC, slow diffusion coefficient; SDCm, SDC pixelwise map.

Figure 3 DDVDm (A1,B1,C1) and SDCm (A2,B2,C2) of three HG lesions (thin arrows) of typical appearance. Cases (A,B) are from dataset-1. Case (C) is from dataset-2. For case (A), the HG lesion has higher signals than those of the kidney parenchyma on DDVDm and on SDCm, and higher signals than that of the gastric liquid (thick arrow). For case (B), the HG lesion has higher signal relative to the liver parenchyma on DDVDm and very high signal on SDCm [being higher than that of the gastric liquid (thick arrow)]. For case (C), very high signals are shown on both DDVDm and on SDCm. *, denotes an area with DWI artifacts (visible on DWI images). Note that, due to respiratory motion, DDVDm and SDCm might not be exactly at the same section for each case. Color scale for (A,B): DDVD: −130 to 500 au/pixel; SDC: −1,000 to 14,000 au/s. Color scale for (C): DDVD: −20 to 100 au/pixel; SDC: −100 to 1,200 au/s. Au, arbitrary unit; DDVD, diffusion-derived vessel density; DDVDm, DDVD pixelwise map; DWI, diffusion-weighted imaging; HG, hemangioma; SDC, slow diffusion coefficient; SDCm, SDC pixelwise map.

Figure 4 Three patients with HG lesion showing very high signal on SDCm. (A) A HG case from dataset-2, the lesion (arrow) has SDC signal higher than that of the left kidney. (B) A HG case from dataset-1, the lesion (arrow) has SDC signal higher than that of the left kidney. (C) A HG case from dataset-1, the lesion (arrow) has SDC similar to or higher than that of the gastric fluid. Color scale: SDC: (A) −100 to 1,200 au/s; (B,C) −1,000 to 14,000 au/s. Au, arbitrary unit; HG, hemangioma; SDC, slow diffusion coefficient; SDCm, SDC pixelwise map.

Figure 5 DDVDm (A1,B1,C1) and SDCm (A2,B2,C2) of three HCC cases (thin arrows) from dataset-1. For cases (A,C), the HCC lesions have higher signals than adjacent liver parenchyma, however, these lesions’ signal is lower than the expected signal of HG on DDVDm, and the signal is much lower than the gastric liquid (thick arrows) on SDCm. For case (B), the HCC lesion has slightly lower signal than adjacent liver parenchyma on DDVDm and higher signal than adjacent liver parenchyma on SDCm. On SDCm, the lesion signal is lower than those of the gastric liquid (thick arrow) and spleen. Note that, due to respiratory motion, DDVDm and SDCm might not be exactly at the same section for each case. Color scale: DDVD: −130 to 500 au/pixel; SDC: −1,000 to 14,000 au/s. Au, arbitrary unit; DDVD, diffusion-derived vessel density; DDVDm, DDVD pixelwise map; HCC, hepatocellular carcinoma; HG, hemangioma; SDC, slow diffusion coefficient; SDCm, SDC pixelwise map.

Figure 6 DDVDm (A1,B1) and SDCm (A2,B2) of two HCC cases (arrows) from dataset-2. The lesion in (A) shows moderately high signal on both DDVDm and SDCm. Due to tumor necrosis, the lesion in (B) shows mixed high signal and iso-signal on both DDVDm and SDCm. Color scale: DDVD: −20 to 100 au/pixel; SDC: −100 to 1,200 au/s. Au, arbitrary unit; DDVD, diffusion-derived vessel density; DDVDm, DDVD pixelwise map; HCC, hepatocellular carcinoma; SDC, slow diffusion coefficient; SDCm, SDC pixelwise map.

Figure 7 DDVDm (A2,B2) and SDCm (A3,B3) appearance of two cases of HCC from dataset-2. (A) A large HCC (arrows) with central necrosis (*). (A1) DWI b=0 s/mm² image shows the lesion with heterogeneous high signal and with very high signal central necrosis (*). (A2) DDVDm shows the lesion with heterogeneous iso-signal. (A3) SDCm shows moderately high signal for HCC tissue and very high signal for the central necrosis. (B1) DWI b=0 s/mm² image shows the lesion (arrows) with iso-signal. *, denotes ascites with very high signal. (B2) DDVDm shows the lesion with slightly low signal (arrows). *, denotes ascites with very high signal. As described earlier (25), misalignment between DWI b=0 and DWI b=10 s/mm² images may cause an artificial very high signal for liquid on DDVDm. (B3) SDCm shows faintly high signal for HCC tissue (arrows). Color scale: DDVD: −20 to 100 au/pixel; SDC: −100 to 1,200 au/s. Au, arbitrary unit; DDVD, diffusion-derived vessel density; DDVDm, DDVD pixelwise map; DWI, diffusion-weighted imaging; HCC, hepatocellular carcinoma; SDC, slow diffusion coefficient; SDCm, SDC pixelwise map.

Figure 8 A case of FNH (arrows) from dataset-1. (A) DWI b=0 s/mm² images. (B) DWI b=2 s/mm² images. On both (A) and (B), the lesion shows modestly high signal. On DDVDm (C) and SDCm (D), the lesion shows faintly high signal which is typical for FNH (25). Color scale: DDVD: −130 to 500 au/pixel; SDC: −1,000 to 14,000 au/s. Au, arbitrary unit; DDVD, diffusion-derived vessel density; DDVDm, DDVD pixelwise map; DWI, diffusion-weighted imaging; FNH, focal nodular hyperplasia; SDC, slow diffusion coefficient; SDCm, SDC pixelwise map.

Figure 9 DDVDm and SDCm show a HG lesion (arrows) patient (A) and a Mets lesion (arrows) patient (B) from dataset-3. The HG shows very high signal on DDVDm (A2) and on SDCm (A3). On the other hand, the Mets show iso-signal on DDVDm (B2) and signal lower than that of liquid on SDCm (B3). (A1,B1) DWI b=0 s/mm² images. Color scale: DDVD: −40 to 100 au/pixel; SDC: −500 to 1,200 au/s. Au, arbitrary unit; DDVD, diffusion-derived vessel density; DDVDm, DDVD pixelwise map; DWI, diffusion-weighted imaging; HG, hemangioma; Mets, metastases; SDC, slow diffusion coefficient; SDCm, SDC pixelwise map.

The classification performances for dataset-1 and dataset-2 are shown in Table 1, with 95.8% of the HG and 97.7% of the MFL being correctly classified with confidence. Only one HG and two MFLs were correctly classified but without confidence, and one MFL was classified as undecided. Dataset-3 overall had a low signal-to-noise ratio. The classification performance for dataset-3 is shown in Table 2, with 75.0% (6/8) of the HG and 96.4% of the MFL being correctly suggested. One HG was undecided, while one HG was incorrectly suggested to be MFL.

Figure 10 shows a comparison of an earlier analysis of 22 HG lesions and 28 MFL lesions (HCC n=24, FNH n=4) based solely on DDVDm (A1,B1) with the analysis in the current study based on a combination of DDVDm and SDCm (A2,B2). In the earlier analysis, correct diagnosis was made in 90.9% of the HGs (72.7% with confidence) and 96.4% of the MFLs (85.7% with confidence) (4). In the current analysis of these same lesions, all were correctly classified with confidence both for HG and for MFL.

Figure 10 A comparison of an earlier analysis of 22 HG lesions and 28 MFL lesions based on DDVDm only (A1,B1) with the analysis in the current study based on a combination of DDVDm and SDCm (A2,B2). Diameter is defined as the longest length of the lesion on the slice with the largest area on DWI. In the current analysis of these lesions, all were correctly classified with confidence (A2,B2). (A1,B1) Reproduced from (4) with permission. Green dot: correctly classified with confidence in the initial analysis; blue dot: correctly classified but without confidence in the initial analysis; red dot: incorrectly classified (but without confidence) in the initial analysis. Confi, confidence; DDVD, diffusion-derived vessel density; DDVDm, DDVD pixelwise map; DWI, diffusion-weighted imaging; HG, hemangioma; MFL, mass-forming lesion; SDC, slow diffusion coefficient; SDCm, SDC pixelwise map.

Discussion

In some cases, a liver HG can be diagnosed based on typical MRI features without the need for contrast-enhanced scan. However, when MRI is the first-line examination for the liver, contrast-enhanced scan is commonly acquired to increase diagnostic confidence for HG. Contrast-enhanced MRI incurs additional cost and additional MRI scan duration, besides the potential side-effects associated with gadolinium agents. In our preliminary testing with sole DDVDm (without referring to DWI and anatomical imaging), two out of 22 liver HGs were miss-labeled as MFL, and one out of 28 MFLs were miss-labeled as HG. Four out of 22 liver HG were correctly labeled as HG but without high confidence, and three out of 28 MFL were correctly labeled as MFL but without high confidence (4). Moreover, in that study, MFL did not include Mets and ICC. In the current study, SDCm was integrated into the diagnostic procedure, MFL number was increased and MFL included Mets and ICC as well. The current study shows the majority of liver HGs and MFLs can be reliably separated with confidence. For dataset-1 and dataset-2, 95.8% of the HG and 97.7% of the MFL were correctly classified with confidence. Dataset-3 overall had a low signal-to-noise ratio, while 96.4% of the MFL and 75.0% (6/8) of the HG were correctly suggested (one HG was undecided, while one HG was incorrectly suggested to be MFL).

DWI plays an important role in MRI evaluation of a variety of pathologies. However, clinical application of the DWI-derived quantitative metrics of ADC and IVIM parameters [perfusion fraction (PF), D_slow, D_fast] has not been very successful. ADC is indeed widely used, while it is generally regarded by physicians and radiologists that its role is only ‘supportive’ rather than ‘confirmatory’. IVIM technique has largely remained in the research phase (30-32). One of the difficulties of ADC and IVIM quantification is that, in addition to tissue perfusion and diffusion, these two metrics are substantially affected by tissue T2 relaxation time (33-37). For ADC, it is conventionally considered that the spleen has a much lower diffusion than the liver, HCC has a lower diffusion than adjacent liver parenchyma, liver HG has a lower diffusion than liver cyst, and liver pyogenic abscess liquid has a lower diffusion than adjacent liver parenchyma. These do not appear to be reasonable (5,29). With conventional modelling, IVIM-PF of the spleen has been consistently measured being only half of that of the liver (35), IVIM-PF of HCC has been consistently measured lower than that of the adjacent liver parenchyma (20,38), and IVIM-PF has been shown to be higher paradoxically in livers with steatosis than in livers without steatosis (39), and higher in liver and spleen of older subjects than liver and spleen of younger subject (18,40). In published literature, D_slow values do not appear to be reasonable as a slow diffusion metric. For example, in a review article by Englund et al. (41), it was noted that skeletal muscle has a D_slow of 1.46±0.30 mm²/s which is much higher than the liver D_slow of 1.1 mm²/s (42,43). We would think that the D_slow of skeletal muscles will not be higher than that of liver with the liver more richly perfused by hepatic artery and portal vein and with lots of sinusoids and space of Disse. Majority of literature reported a lower D_slow in HCC tissue than in liver parenchyma (43,44). However, HCC is associated with faster blood transit time and higher free water content than liver parenchyma. The application of IVIM for liver HG evaluation has not been successful (45). IVIM technique is also commonly associated with long data acquisition time and data fitting instability (31,46). The analysis of DDVD and SDC requires only four b-values (with one being b=0 s/mm²) in total, allowing a significantly shorter scanning time (i.e., approximately with the scan duration of ADC data acquired twice) than contrast-enhanced CT/MRI or IVIM imaging, and does not involve contrast injection. A combination of DDVD and SDC may overcome many of the limitations associated with ADC and IVIM metrics.

This study tested historically acquired liver IVIM image data, data acquisition was not optimized for DDVD and SDC calculation, and without extra efforts being made to minimize the position shift between images. While most HGs and MFLs can be reliably separated by a combination of DDVDm and SDCm, lesions of very small size can cause difficulty for classification (Figure 11). As we discussed earlier (4,25), ‘position shift’ between b=0 and b=2 (or 10 or 50 in this study) s/mm² images or between b=400 (or 500 or 700 in this study) and b=600 (or 800 or 900 in this study) s/mm² images can be a major source of quantification error for DDVD and SDC calculation of the liver, as the liver is heavily subject to respiratory motion (Figures 12-14). Smaller lesions more likely cause diagnostic challenges due to this ‘position shift’. Future studies should include smaller lesions with well position-marched DWI images. One possible way to overcome the difficulty is to scan the DDVD and SDC protocols twice (or even three times for small lesions), and manually select the pair of images with the most similar positions to reconstruct DDVDm and SDCm. In our empirical experience, a HG can usually be established as liquid high signal on SDCm; on the other hand, it is not always easy to confirm a HG diagnosis on DDVDm (4). Thus, the first step of image reading can be to confirm a HG is with liquid high signal on SDCm. Then, the differential diagnosis would be between HG or simple cyst, with simple cysts showing low signal on DDVDm (Figure 15). It is less likely that a HG will show cyst signal on DDVDm unless the lesion is very small or subject to motion artifacts. DWI images may also help with this. Simple cyst shows bright liquid signal on DWI image. A HG may show bright liquid signal on DWI image (such as the larger lesion in Figure 2B), but some portions of HGs show signal lower than liquid signal on DWI image (such as the smaller lesion in Figure 2B) (4). It should be noted that some Mets with large necrotic area may mimic cyst lesion on DDVDm and SDCm, and morphological images are required to make the correct diagnosis (Figure 16).

Figure 11 A case of small HG (arrows) from dataset-2. (A1) DWI b=0 s/mm² images. (A2) DWI b=10 s/mm² images. There is a ‘position-shift’ between (A1) and (A2), and reliable DDVDm could not be calculated for this lesion. (B1,B2) SDCm calculated with b=500 and b=800 s/mm² images and both use the same color scale. (C1,C2) SDCm calculated with b=400 and b=800 s/mm² images and both use the same color scale. (D1,D2) SDCm calculated with b=150 and b=800 s/mm² images and both use same color scale. (B2,C2,D2) Sections at the kidney level to show the relative kidney signal. This case demonstrates DDVD/SDC diagnostic difficulty arises for small lesions, though (B1,C1,D1) may suggest this is a ‘liquid lesion’, as for such a small lesion, its signal is too high for MFL. Both in the axial axis and in the longitudinal axis, this lesion likely contains pixels with partial volume effect involving lower signals of the liver tissue. Color scale: SDC: −100 to 1,200 au/s. Au, arbitrary unit; DDVD, diffusion-derived vessel density; DDVDm, DDVD pixelwise map; DWI, diffusion-weighted imaging; HG, hemangioma; MFL, mass-forming lesion; SDC, slow diffusion coefficient; SDCm, SDC pixelwise map.

Figure 12 The gallbladder and a simple cyst appear artificially high signal on DDVDm (* for the gallbladder and arrows for the cyst). (A,B) There is a position-shift between b=0 and b=10 s/mm² images, so that the gallbladder and the simple cyst appear in different shapes on b=0 and b=10 s/mm² images. (C,D) The same DDVDm with different color scales [−65 to 368 au/pixel for (C) and −20 to 100 au/pixel for (D)], and the cyst shows high signal and the gallbladder shows mixed high signal and very low signal. (E) SDCm (scale: DDVD: −100 to 100 au/pixel) showing the gallbladder and the cyst with signal higher than that of the spleen. This additional case was not included in the materials of this study. Au, arbitrary unit; DDVD, diffusion-derived vessel density; DDVDm, DDVD pixelwise map; DWI, diffusion-weighted imaging; SDC, slow diffusion coefficient; SDCm, SDC pixelwise map.

Figure 13 Small liver simple cysts (arrows) in a patient appear artificially high signal ring on DDVDm. There was a slight position shift between (A1) and (A2) (the liver vasculature appears different in shape on these two images). On DDVDm (A3), two small liver simple cysts appear artificially high signal ring, and an additional cyst show very low signal. There was a slight position-shift between (B1) and (B2), and the cyst shows artificially high signal ring on DDVDm (B3). There was a minimal position-shift between (C1) and (C2), and the cyst shows artificially high signal ring on DDVDm (C3). The lower central signal suggests cyst lesion rather than HG. Color scale: DDVD: −20 to 100 au/pixel. This additional case was not included in the materials of this study. Au, arbitrary unit; DDVD, diffusion-derived vessel density; DDVDm, DDVD pixelwise map; DWI, diffusion-weighted imaging; HG, hemangioma.

Figure 14 Position-shift between two DWI images to construct SDCm can cause alternative high signal and low signal on adjacent maps. (A,B) A large HCC lesion. (C) DDVDm (color scale: −130 to 500 au/pixel) shows the lesion is slightly high signal, consistent with a solid MFL. (D1,E1,F1) Slice sections with high signal for the lesion, and (D2,E2,F2) slice sections with lower signal for the lesion. On (D1), the lesion signal mimics liquid signals as those in the stomach. However, slice (D2) shows the lesion is only slightly higher signal than the adjacent liver. On (E1,F1), the lesion remains high signal but that is lower than the liquid signal in the stomach. Focal necrotic areas are also noted. Sections (E2,F2) show the lesions signal is lower than that in (E1,F1). (D3,E3,F3) Sections at the kidney level for reference. SDC color scale: −1,000 to 14,000 au/s for (D1-D3); −800 to 13,000 au/s for (E1-E3); −1,000 to 17,000 au/s for (F1-F3). Au, arbitrary unit; DDVD, diffusion-derived vessel density; DDVDm, DDVD pixelwise map; DWI, diffusion-weighted imaging; HCC, hepatocellular carcinoma; MFL, mass-forming lesion; SDC, slow diffusion coefficient; SDCm, SDC pixelwise map.

Figure 15 Typical DDVDm and SDCm appearances of liver simple cysts (arrows) in three patients. (A-C) From three patients, respectively, and there are multiple cysts in (C) (five labeled with arrows, a few very small cysts not labeled). Cysts typically show a very low signal on DDVDm, and a very high liquid signal on SDCm. Cyst 2 on DDVDm and cyst 1 on SDCm suffered partial motion artifact effect. Color scale: DDVD: −130 to 500 au/pixel; SDC: −1,000 to 14,000 au/s. These additional three cases were not included in the materials of this study. Au, arbitrary unit; DDVD, diffusion-derived vessel density; DDVDm, DDVD pixelwise map; DWI, diffusion-weighted imaging; SDC, slow diffusion coefficient; SDCm, SDC pixelwise map.

Figure 16 A case of Mets in the liver (original: adenocarcinoma of the appendix) with large cystic change. (A) b=0 s/mm² DWI; (B) T1WI+C; (C) DDVDm, (D) SDCm. DDVDm shows cystic signal. SDCm shows predominantly liquid signal but with a thicker wall (arrow). The tumor wall is better shown on (A) (arrow). Color scale: DDVD: −20 to 100 au/pixel; SDC: −5 to 35 au/s. This additional case was not included in the materials of this study. Au, arbitrary unit; DDVD, diffusion-derived vessel density; DDVDm, DDVD pixelwise map; DWI, diffusion-weighted imaging; Mets, metastases; SDC, slow diffusion coefficient; SDCm, SDC pixelwise map; T1WI+C, contrast-enhanced T1-weighted image.

There are a number of other limitations to this study. The NEX was only 1 for dataset-2 b=0 and b=10 s/mm² images. Dataset-3 suffered from low signal-to-noise ratio, leading to the diagnostic confidence level for DDVDm and SDCm being not high. This study only assessed the diagnostic performance of DDVDm and SDCm, without considering the integration of other T2, ADC, and morphological features. Other less common liver lesions, such as hepatic adenoma, were not included for the analysis. However, we do not anticipate difficulties in separating liver HG from these lesions. The less common sclerosing HG might not have been well represented in the current study. For dataset-3, DDVDm was reconstructed with b=0 and b=50 s/mm² images as a lower non-zero b-value image was not available. DDVD_b0b50 would be contributed by T2 relaxation (16,26), and not be suitable for separating cyst and HG. The role of dataset-3 is considered to be supportive only. Moreover, it is possible that some rare vascular tumors or lesions, such as hemangioendothelioma, angiosarcoma, artero-venous fistula (47-49), may appear similar or partially similar in signal to HG on DDVDm and SDCm. Future studies with optimized scan parameters and with larger sample size and with a variety of liver rare lesions are desirable.

Conclusions

In conclusion, this study shows, when a combination of DDVDm and SDCm is used to evaluate the liver, HG and MFL can mostly be reliably separated. With the integration of DDVDm and SDCm into liver MRI, the number of gadolinium-contrast-enhanced scans can be saved in a high proportion of patients, particularly for patients with larger lesions.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the STARD reporting checklist. Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-424/rc

Data Sharing Statement: Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-424/dss

Peer Review File: Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-424/prf

Funding: This study received funding from a Hong Kong GRF Project (No. 14112521), National Natural Science Foundation of China (No. 82171893), Shenzhen Basic Research Plan Fund Project (Natural Science Foundation of Shenzhen) (No. JCYJ20240813104112017), Shenzhen People’s Hospital Physician Scientist Training “Five Three” Program (No. SYWGSLCYJ202404).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-424/coif). M.S.Y.Z. contributed to the development of Yingran Medicals Co., Ltd. Y.X.J.W. reports that there is a Chinese patent pending related to this article. He is also the founder of Yingran Medicals Co., Ltd., a company that develops medical imaging-based diagnostic software. The other authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was approved by the institutional ethics committees of Sun Yat-sen Memorial Hospital (No. S1709442250839) and Zhujiang Hospital (No. 2017-YXZDK-002), with informed consent obtained from individual participants.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Kacała A, Dorochowicz M, Matus I, et al. Hepatic Hemangioma: Review of Imaging and Therapeutic Strategies. Medicina (Kaunas) 2024;60:449. [Crossref] [PubMed]
2. Glinkova V, Shevah O, Boaz M, et al. Hepatic haemangiomas: possible association with female sex hormones. Gut 2004;53:1352-5. [Crossref] [PubMed]
3. Giannitrapani L, Soresi M, La Spada E, et al. Sex hormones and risk of liver tumor. Ann N Y Acad Sci 2006;1089:228-36. [Crossref] [PubMed]
4. Hu GW, Li CY, Zhang G, et al. Diagnosis of liver hemangioma using magnetic resonance diffusion-derived vessel density (DDVD) pixelwise map: a preliminary descriptive study. Quant Imaging Med Surg 2024;14:8064-82. [Crossref] [PubMed]
5. Xu FY, Xiao BH, Wáng YXJ. The rationale for proposing a magnetic resonance slow diffusion metric and its proof-of-concept testing showing spleen parenchyma and hepatocellular carcinoma have faster diffusion than liver parenchyma. Quant Imaging Med Surg 2025;15:3792-806. [Crossref] [PubMed]
6. Yamada I, Aung W, Himeno Y, et al. Diffusion coefficients in abdominal organs and hepatic lesions: evaluation with intravoxel incoherent motion echo-planar MR imaging. Radiology 1999;210:617-23. [Crossref] [PubMed]
7. Parikh T, Drew SJ, Lee VS, et al. Focal liver lesion detection and characterization with diffusion-weighted MR imaging: comparison with standard breath-hold T2-weighted imaging. Radiology 2008;246:812-22. [Crossref] [PubMed]
8. Kim HC, Seo N, Chung YE, et al. Characterization of focal liver lesions using the stretched exponential model: comparison with monoexponential and biexponential diffusion-weighted magnetic resonance imaging. Eur Radiol 2019;29:5111-20. [Crossref] [PubMed]
9. Ohtomo K, Itai Y, Furui S, et al. Hepatic tumors: differentiation by transverse relaxation time (T2) of magnetic resonance imaging. Radiology 1985;155:421-3. [Crossref] [PubMed]
10. Lombardo DM, Baker ME, Spritzer CE, et al. Hepatic hemangiomas vs metastases: MR differentiation at 1.5 T. AJR Am J Roentgenol 1990;155:55-9. [Crossref] [PubMed]
11. Cieszanowski A, Szeszkowski W, Golebiowski M, et al. Discrimination of benign from malignant hepatic lesions based on their T2-relaxation times calculated from moderately T2-weighted turbo SE sequence. Eur Radiol 2002;12:2273-9. [Crossref] [PubMed]
12. Duran R, Ronot M, Kerbaol A, et al. Hepatic hemangiomas: factors associated with T2 shine-through effect on diffusion-weighted MR sequences. Eur J Radiol 2014;83:468-78. [Crossref] [PubMed]
13. Holzapfel K, Bruegel M, Eiber M, et al. Characterization of small (≤10 mm) focal liver lesions: value of respiratory-triggered echo-planar diffusion-weighted MR imaging. Eur J Radiol 2010;76:89-95. [Crossref] [PubMed]
14. EASL Clinical Practice Guidelines on the management of benign liver tumours. J Hepatol 2016;65:386-98. [Crossref] [PubMed]
15. Wáng YXJ. Living tissue intravoxel incoherent motion (IVIM) diffusion MR analysis without b=0 image: an example for liver fibrosis evaluation. Quant Imaging Med Surg 2019;9:127-33. [Crossref] [PubMed]
16. Ma FZ, Xiao BH, Wáng YXJ. MRI signal simulation of liver DDVD (diffusion derived 'vessel density') with multiple compartments diffusion model. Quant Imaging Med Surg 2025;15:1710-18. [Crossref] [PubMed]
17. Xiao BH, Huang H, Wang LF, et al. Diffusion MRI Derived per Area Vessel Density as a Surrogate Biomarker for Detecting Viral Hepatitis B-Induced Liver Fibrosis: A Proof-of-Concept Study. SLAS Technol 2020;25:474-83. [Crossref] [PubMed]
18. Huang H, Zheng CJ, Wang LF, et al. Age and gender dependence of liver diffusion parameters and the possibility that intravoxel incoherent motion modeling of the perfusion component is constrained by the diffusion component. NMR Biomed 2021;34:e4449. [Crossref] [PubMed]
19. Deng YY, Sun MH, Huang H, et al. Age-and gender-related variations of liver diffusion metrics apparent diffusion coefficient (ADC) and diffusion derived vessel density (DDVD), and explanations with the known physiological T2 relaxation time variations among different volunteers' groups. BMC Med Imaging 2025;25:185. [Crossref] [PubMed]
20. Li XM, Yao DQ, Quan XY, et al. Perfusion of hepatocellular carcinomas measured by diffusion-derived vessel density biomarker: Higher hepatocellular carcinoma perfusion than earlier intravoxel incoherent motion reports. NMR Biomed 2024;37:e5125. [Crossref] [PubMed]
21. Wáng YXJ, Li CY, Yao DQ, et al. The impacts of time of echo (TE) and time of repetition (TR) on diffusion-derived 'vessel density' (DDVD) measurement: examples of liver, spleen, and hepatocellular carcinoma. Quant Imaging Med Surg 2025;15:3771-8. [Crossref] [PubMed]
22. Zheng CJ, Huang H, Xiao BH, et al. Spleen in viral Hepatitis-B liver fibrosis patients may have a reduced level of per unit micro-circulation: non-invasive diffusion MRI evidence with a surrogate marker. SLAS Technol 2022;27:187-94. [Crossref] [PubMed]
23. Gitlin N, Grahame GR, Kreel L, et al. Splenic blood flow and resistance in patients with cirrhosis before and after portacaval anastomoses. Gastroenterology 1970;59:208-13.
24. Zheng CJ, Yao DQ, Li CY, et al. Lower diffusion-derived vessel density (DDVD) measure of liver focal nodular hyperplasia than those of hepatocellular carcinoma and liver metastasis allows potential differential diagnosis: quantitative and semi-quantitative analyses of two-center data. J Gastrointest Oncol 2025;16:1144-56. [Crossref] [PubMed]
25. Yao DQ, Zheng CJ, Deng YY, et al. Potential diverse applications of diffusion-derived vessel density (DDVD) pixel-by-pixel mapping. Quant Imaging Med Surg 2024;14:2136-45. [Crossref] [PubMed]
26. Ju ZG, Leng XM, Xiao BH, et al. Influences of the second motion probing gradient b-value and T2 relaxation time on magnetic resonance diffusion-derived 'vessel density' (DDVD) calculation: the examples of liver, spleen, and liver simple cyst. Quant Imaging Med Surg 2025;15:74-87. [Crossref] [PubMed]
27. Yeung J, Sivarajan S, Treibel TA, et al. Measurement of liver and spleen interstitial volume in patients with systemic amyloid light-chain amyloidosis using equilibrium contrast CT. Abdom Radiol (NY) 2017;42:2646-51. [Crossref] [PubMed]
28. Wáng YXJ. The very low magnetic resonance imaging apparent diffusion coefficient (ADC) measure of abscess is likely due to pus's specific T2 relaxation time. Quant Imaging Med Surg 2023;13:8881-5. [Crossref] [PubMed]
29. Ling GH, Wáng YXJ. MRI slow diffusion coefficient (SDC) shows liver pyogenic abscess has faster diffusion than adjacent liver parenchyma. Quant Imaging Med Surg 2025;15:6547-51. [Crossref] [PubMed]
30. Wang Q, Yu G, Qiu J, et al. Application of Intravoxel Incoherent Motion in Clinical Liver Imaging: A Literature Review. J Magn Reson Imaging 2024;60:417-40. [Crossref] [PubMed]
31. Wáng YXJ, Wang X, Wu P, et al. Topics on quantitative liver magnetic resonance imaging. Quant Imaging Med Surg 2019;9:1840-90. [Crossref] [PubMed]
32. Lu PX, Huang H, Yuan J, et al. Decreases in molecular diffusion, perfusion fraction and perfusion-related diffusion in fibrotic livers: a prospective clinical intravoxel incoherent motion MR imaging study. PLoS One 2014;9:e113846. [Crossref] [PubMed]
33. Wáng YXJ, Aparisi Gómez MP, Ruiz Santiago F, et al. The relevance of T2 relaxation time in interpreting MRI apparent diffusion coefficient (ADC) map for musculoskeletal structures. Quant Imaging Med Surg 2023;13:7657-66. [Crossref] [PubMed]
34. Wáng YXJ. An explanation for the triphasic dependency of apparent diffusion coefficient (ADC) on T2 relaxation time: the multiple T2 compartments model. Quant Imaging Med Surg 2025;15:3779-91. [Crossref] [PubMed]
35. Yu WL, Xiao BH, Ma FZ, et al. Underestimation of the spleen perfusion fraction by intravoxel incoherent motion MRI. NMR Biomed 2023;36:e4987. [Crossref] [PubMed]
36. Jerome NP, d'Arcy JA, Feiweier T, et al. Extended T2-IVIM model for correction of TE dependence of pseudo-diffusion volume fraction in clinical diffusion-weighted magnetic resonance imaging. Phys Med Biol 2016;61:N667-80.
37. Führes T, Riexinger AJ, Loh M, et al. Echo time dependence of biexponential and triexponential intravoxel incoherent motion parameters in the liver. Magn Reson Med 2022;87:859-71. [Crossref] [PubMed]
38. Ma FZ, Wáng YXJ. T2 relaxation time elongation of hepatocellular carcinoma relative to native liver tissue leads to an underestimation of perfusion fraction measured by standard intravoxel incoherent motion magnetic resonance imaging. Quant Imaging Med Surg 2024;14:1316-22. [Crossref] [PubMed]
39. Wáng YXJ. Observed paradoxical perfusion fraction elevation in steatotic liver: An example of intravoxel incoherent motion modeling of the perfusion component constrained by the diffusion component. NMR Biomed 2021;34:e4488. [Crossref] [PubMed]
40. Yu WL, Ma FZ, Huang H, et al. Age and gender differences of normative values of spleen diffusion MRI parameters. Rofo 2025;197:535-45. [Crossref] [PubMed]
41. Englund EK, Reiter DA, Shahidi B, et al. Intravoxel Incoherent Motion Magnetic Resonance Imaging in Skeletal Muscle: Review and Future Directions. J Magn Reson Imaging 2022;55:988-1012. [Crossref] [PubMed]
42. Riexinger AJ, Martin J, Rauh S, et al. On the Field Strength Dependence of Bi- and Triexponential Intravoxel Incoherent Motion (IVIM) Parameters in the Liver. J Magn Reson Imaging 2019;50:1883-92. [Crossref] [PubMed]
43. Li YT, Cercueil JP, Yuan J, et al. Liver intravoxel incoherent motion (IVIM) magnetic resonance imaging: a comprehensive review of published data on normal values and applications for fibrosis and tumor evaluation. Quant Imaging Med Surg 2017;7:59-78. [Crossref] [PubMed]
44. Yue X, Lu Y, Jiang Q, et al. Application of Intravoxel Incoherent Motion in the Evaluation of Hepatocellular Carcinoma after Transarterial Chemoembolization. Curr Oncol 2022;29:9855-66. [Crossref] [PubMed]
45. Wáng YXJ, Sabarudin A. Underestimation of liver hemangioma perfusion fraction by standard intravoxel incoherent motion diffusion magnetic resonance imaging. Quant Imaging Med Surg 2024;14:2128-35. [Crossref] [PubMed]
46. Chevallier O, Zhou N, Cercueil JP, et al. Comparison of tri-exponential decay versus bi-exponential decay and full fitting versus segmented fitting for modeling liver intravoxel incoherent motion diffusion MRI. NMR Biomed 2019;32:e4155. [Crossref] [PubMed]
47. Bonatti M, Valletta R, Corato V, et al. I thought it was a hemangioma! A pictorial essay about common and uncommon liver hemangiomas' mimickers. Insights Imaging 2024;15:228. [Crossref] [PubMed]
48. Paulatto L, Dioguardi Burgio M, Sartoris R, et al. Colorectal liver metastases: radiopathological correlation. Insights Imaging 2020;11:99. [Crossref] [PubMed]
49. Choi BI, Lee KH, Han JK, et al. Hepatic arterioportal shunts: dynamic CT and MR features. Korean J Radiol 2002;3:1-15. [Crossref] [PubMed]