The role of number-of-excitations (NEX) for the usability of free-breathing acquired diffusion-derived ‘vessel density’ (DDVD) and slow diffusion coefficient (SDC) maps in separating liver cysts, hemangiomas, and solid masses with a focus on small lesions

Introduction

Recently two new types of diffusion weighted imaging (DWI) contrast, diffusion-derived ‘vessel density’ (DDVD) weighted imaging and slow diffusion coefficient (SDC) weighted imaging, have been tested in the liver. Liver vessels and micro-vessels show high-signal when there is no diffusion gradient (b=0 s/mm²) and low-signal when even very low b-values (such as b=2 s/mm²) are applied. Thus, the signal difference between images when the diffusion gradient is ‘off’ and ‘on’ reflects the extent of functional tissue vessel density, and this difference is termed as DDVD (1-3). DDVD is a useful parameter for distinguishing livers with and without fibrosis, and livers with severer fibrosis tend to have even lower DDVD measurements than those with milder liver fibrosis (1,3,4). With DDVD analysis, Zheng et al. (5) demonstrated that per unit micro-circulation of spleen is decreased in viral hepatitis B liver fibrosis patients. This is consistent with, for example, the report of Gitlin et al. (6) 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. On the contrary, total splenic blood flow, calculated by multiplying specific splenic flow by spleen volume, was increased (6). In a recent analysis, when a time of echo (TE) of 59 ms [time of repetition (TR) =1,600 ms, free breathing acquisition] and b=0, 2 mm²/s were used for the DDVD calculation of 26 cases of hepatocellular carcinomas (HCCs), the mean DDVD_HCC/DDVD_liver ratio was 1.42. This value agrees well with the literature results showing perfusion-computed tomography (CT) blood volume ratio of HCC to liver (blood volume_HCC/blood volume_liver) median value being 1.38 (7). Liver focal nodular hyperplasia (FNH) has been shown to have a lower DDVD measure than HCC and metastasis (Mets), and on average Mets had a higher DDVD measure than HCC (8). Slow diffusion coefficient (SDC) was proposed to measure tissue slow diffusion (9). In its basic form, SDC is derived from a high b-value DWI image (typically with b-value of 400–500 s/mm²) and a higher b-value DWI image (typically with b-value of 600–800 s/mm²). With the conventional apparent diffusion coefficient (ADC) approach, the spleen has been reported to have a much lower ADC than liver, HCCs have a lower ADC than liver parenchyma. On the other hand, with SDC analysis, the spleen has faster diffusion than liver and HCCs have faster diffusion than liver parenchyma (9). The liver and spleen have a similar amount of blood perfusion, the spleen is more watery than the liver. HCCs are mostly associated with increased blood supply and increased proportion of arterial blood supply and with edema. It is more reasonable with SDC results that spleen and HCC have faster diffusion than liver parenchyma. A combination of SDC and DDVD (abbreviated as SDC/DDVD in this article) has been tested to evaluate liver focal lesions. Based on T2 weighed imaging (T2WI) signal, SDC/DDVD, and ADC, a semi-quantitative score scheme termed ‘LiverMss-FNH’ was used to evaluate liver solid mass. In two studies totaling 25 FNHs and 132 liver malignant tumors, it was shown that LiverMss-FNH ≥3.0 suggests the possibility of a liver mass being FNH, and LiverMss ≥4 can strongly favor the diagnosis for FNH (10,11). A liver mass with an iso-signal or slightly high DDVD signal while with an SDC higher or equal to that of the kidneys had an odds ratio of 34.7 in favor of intrahepatic cholangiocarcinoma (ICC) over HCC (12). SDC/DDVD has also been tested to separate hemangioma (HG) from solid or solid component dominant mass-forming lesion (MFL) [for details, see (13,14)]. Hu et al. (14) reported that SDC/DDVD offers an accuracy of >95% in separating liver HGs and liver MFLs (composed of HCC and FNH).

Earlier studies did not test the reliability of SDC/DDVD in separating liver cyst and HG. In theory, cyst shows very low DDVD signal when the second b-value is ≤10 s/mm², and HG always shows very high DDVD (2,14). However, in practice, misalignment between DWI_b0 (DWI when b-value is 0 s/mm²) and second b-value DWI image due to various motions can lead to watery long T2 component showing artificially high signal on DDVD map [such as the case for cerebrospinal fluid (CSF)] (15). Earlie studies included relatively large lesions. For example, Hu et al. (13) evaluated 22 HGs with a median size of 3.26 cm (range: 1.14–8.85 cm) and 28 MFLs with a median size of 4.46 cm (range: 1.72–12.77 cm). Xu et al. (11) evaluated 12 FNHs with a median size of 3.5 cm (range: 1.4–5.8 cm) and 50 liver malignant tumors with a median size of 5.6 cm (range: 1.6–16.9 cm). In addition, earlier studies on liver HG and MFL separation only used SDC/DDVD maps, without the integration of T2WI information (13,14). In this study, aiming to include a substantial portion of lesions sized ≤2 cm and using free-breathing acquired SDC/DDVD maps as well as T2WI signal and morphology of liver focal lesions, we investigate how reliably liver MFL, cyst and HG can be separated without the need for a contrast-enhanced (CE) imaging and how the number-of-excitations (NEX) during the data acquisition can impact the usability of SDC and DDVD maps. Compared with respiration-gated liver magnetic resonance imaging (MRI) data acquisition, free-breathing can save the scan duration by half, and DWI metric qualification is not inferior to respiration-gated acquisition (16,17). According to the Barcelona Clinic Liver Cancer staging system and its treatment strategy, a single tumor smaller than 2 cm is one of the criteria for very early-stage HCC, and these HCCs are often eligible for curative treatments. Due to the patient sample characteristics of this study, we defined lesions <2.2 cm as small lesions.

Methods

Patient data

The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. All imaging data were prospectively acquired with institutional ethical approvals [reference No. KY2025058 for The Fifth Affiliated Hospital of Anhui Medical University (Center 1), reference No. 2026-LS-ky009 for The Second Hospital of Nanjing (Center 2)], and with informed consent obtained from individual participants. The patient data with liver space occupying lesions were consecutively sampled without inclusion or exclusion criteria. The liver DWI was based on a single-shot spin-echo type echo-planar imaging (EPI) sequence with free breathing. The default spectral pre-saturation technique was used for fat suppression. The scan parameter and patient data are shown in Table 1. Counting by lesion number (rather than patient number), in total there were 148 MFLs (91 HCC, 11 ICC, 33 Mets, 13 FNH), 22 cysts, 45 HGs, and 5 liver abscesses (in three patients). All HCC and ICC cases had histopathological diagnosis. The diagnosis of Mets was based on histopathology or a combination of complete patient history and typical imaging features. Three FNH lesions had histopathological diagnosis, and the remaining 10 lesions were diagnosed with typical findings of hepatobiliary contrast agent enhanced MRI and follow-up. The diagnosis of cysts and 45 HGs were confirmed by contrast enhanced MRI or CT. Liver abscesses were diagnosed confirmed by a combination of patient history, typical MRI appearances, and resolution of the lesions after anti-biotics treatment and drainage in one patient.

SDC/DDVD calculation

DDVD weighted maps were derived from (1-3):

DDVD=S(b0)−S(b10)[unit:arbitraryunit(au)/pixel][1]

where b₀ and b₁₀ refer b=0 and 10 s/mm², respectively; S_b0 and S_b10 denote the DWI image signal-intensity acquired at b=0 and 10 s/mm², respectively.

SDC weighted maps were derived from Eq. [2]:

SDC=S(b1)−S(b2)b1−b2[unit:au/s][2]

where b₁ and b₂ refer to a high b-value and a higher b-value, respectively; S_b1 and S_b2 denote the DWI image signal-intensity acquired at the high b-value and the higher b-value, respectively. SDC was calculated with b₁=400 and b₂=600 s/mm² images for Center 1 data, with b₁=600 and b₂=800 s/mm² images for Center 2 data.

Image interpretation

Image interpretation was performed in census by a senior trainee (C.Y.L.) and two radiology specialists (M.H.S., Y.X.J.W.). Lesions were analyzed sequentially on T2WI, SDC, and then DDVD. T2WI was used for lesion localization and size measurement. Lesion size was measured on the slice where the lesion showed the largest size, and the largest diameter was taken. The paired DWI datasets used to reconstruct DDVD or SDC maps were visually checked for substantial spatial misregistration, if any (10-15). Ideally, a lesion on the two DWI images [such as DWI_b0 and DWI_b10 (DWI when b-value is 10 s/mm²)] to construct a diffusion metric map should have the same location and the same shape on these two DWI images. However, due to the respiratory motion and EPI associated image distortion, this requirement is often not satisfied and problematic for small lesions. Based on the existence of notable misalignment between the two images to construct SDC map or DDVD map or other artifacts, SDC maps and DDVD maps were subjectively classified as ‘usable’ or ‘unusable’.

On T2WI, cysts typically show CSF-signal. Small cysts can also show sub-CSF-signal due to partial volume effect but they still tend to show signal higher than small MFLs. HG can show CSF-signal or sub-CSF-signal, but typically with signal higher than the spleen. On SDC map, a lesion was considered to be liquid signal—a key feature of cyst and HG—if its signal intensity was similar to that of CSF and those of gallbladder/gastric fluid and consistently higher than that of the kidneys. To facilitate visual assessment, the SDC color scale was adjusted so that the kidneys appeared as a non-saturated reddish-orange, typically rendering liquid signal as saturated red on this scale (14). A diagnosis of HG was made if a liquid signal lesion on SDC also demonstrated high signal on DDVD map. A diagnosis of cyst was made if a liquid signal lesion on SDC showed very low signal on the DDVD map. When T2WI and SDC map quality were acceptable, but DDVD map was non-diagnostic, the lesion was only classified as liquid lesion (LL). An MFL typically shows mildly or modestly high signal on both SDC and DDVD maps. Some ICC and Mets could also show very high signal on SDC yet they usually do not reach liquid signal level and they are commonly associated with internal heterogeneity [for details, see (11,12,14)]. For morphology, cysts and MFL typically have an expansive shape while HG may have an expansive shape or a non-expansive shape (Figure 1).

Figure 1 Step-by-step diagnosis process applied in this study for liver focal lesions. Due to the higher spatial resolution and stability of the T2WI, reading T2WI is the first step, and its result takes priority over the results of SDC and DDVD. If the lesion has signal equal to CSF-signal or gastric liquid signal on T2WI, then it is an LL. If an LL is expansively round or oval shape (A), it can be cyst or HG. If an LL is non-expansive multilocular or irregular shape (B), it is likely to be a HG. On T2WI, while cysts typically show CSF-signal, a large portion of HGs show sub-CSF signal. HGs with sub-CSF signal commonly have signal higher than spleen signal (C1) but may also not be necessarily so. SDC can usually establish a lesion being liquid signal (C2). Typically, a cyst shows very low DDVD signal and an HG shows very high DDVD signal. However, for the data acquired in this study, DDVD signal may be less reliable for lesions <2 cm. Larger HGs (>2 cm) usually show high DDVD signal. CSF, cerebrospinal fluid; DDVD, diffusion-derived ‘vessel density’; HG, hemangioma; LL, liquid lesion; SDC, slow diffusion coefficient; T2WI, T2-weighted imaging.

For diagnostic assessment, each lesion was assigned to one of four confidence levels: (I) diagnosis confident; (II) highly suggestive (immediate further examination may not be required); (III) suggestive or diagnosis not confident; (IV) undecided. Due to the ever-existing possibility of respiratory motion induced misalignment between DWI images and image distortion associated with EPI, the reading based on T2WI was given the priority. In this study, readers were not required to differentiate among specific solid tumor types (i.e., HCC, ICC, Mets, or FNH). Five hepatic abscesses were not formally evaluated for classification in this study.

Statistical analysis

In this study, no lesion was excluded from the analysis. Data were presented only descriptively without formal statistical analysis.

Results

The MFLs had a median size of 2.9 cm (range: 0.6–25 cm). Two percent (3/148) of MFLs’ SDC maps were not usable, while 16.2% (24/148) of MFLs’ DDVD maps were not usable (Table 2). MFLs with unusable SDC map or DDVD map mostly had a size <2.0 cm (Figure 2). With a further analysis of DDVD maps for MFLs with a size ≤2.2 cm (Figure 3), 47.4% (9/19) of the Center 1 data were usable while 68.9% (31/45) of the Center 2 data were usable. The median lesion size was smaller for Center 2 data. The smallest usable DDVD map in the Center 1 data was a case of 1.6 cm size, while smallest usable DDVD map in the Center 2 data was three cases of 0.9 cm in size. Figure 3 suggests, by increasing a DDVD NEX of two in Center 2 to a DDVD NEX of six in Center 2, the DDVD map usability was improved.

Figure 2 The relationship between the usability of SDC and DDVD maps and the size of liver MFLs. (A) Combined Centers 1 and 2 data. (B) Center 1 data. (C) Center 2 data. One dot denotes one lesion. Blue lines denote median size of the lesions. Center 1: The Fifth Affiliated Hospital of Anhui Medical University. Center 2: The Second Hospital of Nanjing. DDVD, diffusion-derived ‘vessel density’; MFL, mass-forming lesion; NEX, number-of-excitation; SDC, slow diffusion coefficient.

Figure 3 The relationship between the usability of DDVD maps and the size of small liver MFLs (size: ≤2.2 cm), and the NEX for data acquisition. (A) Center 1 data. (B) Center 2 data. One dot denotes one lesion. Blue lines denote median size of the lesions. Results in this graph show, by increasing NEX from 2 in Center 1 to NEX from 6 in Center 2, a higher proportion of smaller lesions became usable, even though the lesion size was on average smaller for Center 2 data. Center 1: The Fifth Affiliated Hospital of Anhui Medical University. Center 2: The Second Hospital of Nanjing. DDVD, diffusion-derived ‘vessel density’; MFL, mass-forming lesion; NEX, number-of-excitation.

LLs (i.e., cysts and HGs) had an overall median size of 1.8 cm (range: 0.5–8.1 cm). For the SDC maps of LLs, 7.5% (5/67) were not usable; while for the DDVD maps of the LL, 40.3% (27/67) were not usable (Figure 4). All LLs with unusable SDC maps had a size <1.5 cm, while LLs with unusable DDVD maps mostly had a size <2.5 cm.

Figure 4 The relationship between the usability of SDC and DDVD maps and the size of liver LLs. (A) Combined Centers 1 and 2 data. (B) Centers 1 and 2 data presented separately. One dot denotes one lesion. Lines denote median size of the lesions. There were too few LLs from Center 2 to allow a comparison of DDVD map usability between Centers 1 and 2. The lesion with unusable DDVD map and a size of 3.1 cm (arrow) located close to the diaphragm and the heart. Center 1: The Fifth Affiliated Hospital of Anhui Medical University. Center 2: The Second Hospital of Nanjing. DDVD, diffusion-derived ‘vessel density’; LL, liquid lesion; SDC, slow diffusion coefficient.

Generally, lesions with unusable SDC map or unusable DDVD map had a size smaller than those of lesions with usable SDC map or usable DDVD map.

For 92.3% (144/148) of the MFLs, both T2WI and SDC confidently classified them as MFL. Three MFLs were classified as MFL on T2WI with confidence (size: 1.8, 1.2, and 0.8 cm), but both SDC and DDVD maps were unusable for these three lesions (Figure 5A). One MFL was confidently classified as solid lesion on T2WI, but both SDC map and DDVD map were not suggestive of MFL (a lesion with internal necrosis and slight misalignment between DWI images).

Figure 5 Diagnostic classification performance of T2WI, SDC map and DDVD map (A-C). Due to the ever-existing possibility of respiratory motion induced misalignment between DWI images and image distortion associated with EPI, the reading based on T2WI was given the priority. Hereby, ‘SDC_DDVD had an inconsistent reading’ means: (I) both SDC and DDVD maps were unusable, or (II) both lesion SDC and DDVD signal did not agree with liquid signal, or (III) lesion DDVD signal did not agree with the liquid signal and SDC maps were unusable; or (IV) lesion SDC signal did not agree with the liquid signal and DDVD maps were unusable. (C) For HGs in groups F and G, further examination would be recommended. Arrow in (A) denotes a lesion with internal necrosis and slight misalignment between DWI images. One dot denotes one lesion. Lines denote median size of the lesions. DDVD, diffusion-derived ‘vessel density’; DWI, diffusion weighted imaging; EPI, echo-planar imaging; HG, hemangioma; MFL, mass-forming lesion; SDC, slow diffusion coefficient; T2WI, T2-weighted imaging.

For the 22 liver cysts, 72.7% (16/22, median size: 1.65 cm, range: 0.8–3.1 cm) were CSF-signal on T2WI, and both T2WI and SDC/DDVD confidently classified them as LL (Figure 5B). A total of 27.3% (6/22, median size: 0.9 cm, range: 0.6–1.1 cm) were sub-CSF-signal, and SDC/DDVD improved the diagnostic confidence of these lesions being LL. Thus, SDC/DDVD may support the diagnosis of cysts being LL for those with a size around and less than 1 cm.

The 45 liver HGs were divided into seven groups according to the classification performance of T2WI, SDC, and DDVD (Figure 5). Group A had 17 lesions with CSF-signal on T2WI (median size: 3.30 cm, range: 1.1–7.7 cm), both T2WI and SDC/DDVD confidently classified them as LL. Group B 11 lesions (median size: 1.70 cm, range: 0.5–8.1 cm) had sub-CSF signal on T2WI while highly suggestive of them being LL, SDC/DDVD improved the confidence of LL diagnosis. Group C had one lesion (size 2.6 cm) which was CSF-signal on T2WI but SDC/DDVD had inconsistent reading (for the meaning of ‘inconsistent reading’, see the legend of Figure 5). Group D had 4 lesions (median size: 1.20 cm, range: 0.5–1.8 cm) with sub-CSF-signal on T2WI but SDC/DDVD had inconsistent reading; however due to their morphology and higher signal than typical MFL lesions, the diagnosis was ‘highly suggestive of LL’. Group E 6 lesions (median size: 2.45 cm, range: 1.5–3.2 cm) were confidently diagnosed as LL by SDC/DDVD, while T2WI was only suggestive of them being LL. Group F had 2 lesions (sizes: 2.1 and 1.5 cm), both T2WI and SDC/DDVD suggested them being LL. Group G had 4 lesions (median size: 1.60 cm, range: 1.2–3.5 cm), T2WI suggested them to be LL while SDC/DDVD had inconsistent reading. Overall, for the HGs which were not typical CSF-signal (n=27/45, 60%), SDC and/or DDVD confirmed the diagnosis of 22.2% (6/27) of them being LL independent of T2 weighted imaging, and further supported the T2 weighted imaging-based diagnosis of 40.7% (11/27) of them being LL.

For the differentiation between cysts and HGs (Figure 6), 50% of the cysts were confidently diagnosed as cyst, 74.4% of the HGs were confidently diagnosed as HG. A total of 36.4% of the cysts and 7.7% of the HG were undecided (to be a cyst or an HG). Two cysts (9.1%) and three HGs (7.7%) were misclassified (being a cyst or an HG).

Figure 6 The diagnosis of liver cyst (n=22) or liver HG (n=39) among the lesions initially classified as LL in Figure 5. HGs overall have a larger size than cysts. A total of 50% of the cysts and 74.4% of the HGs were confidently diagnosed. Arrow: a lobulated T2WI very bright CSF-signal HG adjacent to another liver cyst, this HG was classified as a cyst initially; in hindsight, a HG would be suggested due to its lobulation. One dot denotes one lesion. Lines denote median size of the lesions. CSF, cerebrospinal fluid; HG, hemangioma; LL, liquid lesion; T2WI, T2-weighted imaging.

For the five hepatic abscesses, though all had features of liquefaction, these liquefactions might mimic tumor necrosis. The extensive edema around the focal lesions would suggest them being infectious change; however this had to be considered together with the patients’ history and clinical signs.

Visualizations of patient examples are shown in Figures 7-14.

Figure 7 Appearance of two cases of HCC (A,B) and two cases of Mets (C,D), all indicated by arrows. The primaries from pancreas (C) and stomach (D), respectively. For (D), the lesion in (D4) shows signal higher than that of the spleen, however still has signal lower than the anticipated liquid signal. The high ring signal shown in (D3) tentatively agrees with the richly perfused ring shown on (D2). Yellow asterisk (*) in (C4) denotes the upper pole of right kidney. CE, contrast-enhanced; DDVD, diffusion-derived ‘vessel density’; HCC, hepatocellular carcinoma; Mets, metastasis; MRI, magnetic resonance imaging; SDC, slow diffusion coefficient; T2WI, T2-weighted imaging.

Figure 8 Appearance of a liver cyst (arrows) in two cases (A,B). In both cases, the lesion shows CSF-signal on T2WI, very low signal on DDVD, and markedly high signal on SDC. CE MRI (A2) or CE CT (B2) shows cyst appearance. CE, contrast-enhanced; CSF, cerebrospinal fluid; CT, computed tomography; DDVD, diffusion-derived ‘vessel density’; MRI, magnetic resonance imaging; SDC, slow diffusion coefficient; T2WI, T2-weighted imaging.

Figure 9 Appearance of liver HG (arrows) in three cases (A-C). There are two HGs in (A), denoted with red arrow and orange colored arrow respectively. All four lesions show CSF-signal on T2WI, very high signal on DDVD, and markedly high signal on SDC. (A2,B2) CE MRI. (C2) CE CT, showing HG appearance. CE, contrast-enhanced; CSF, cerebrospinal fluid; CT, computed tomography; DDVD, diffusion-derived ‘vessel density’; HG, hemangioma; MRI, magnetic resonance imaging; SDC, slow diffusion coefficient; T2WI, T2-weighted imaging.

Figure 10 Appearance of two large liver HGs in a patient. The arrow pointed lesion in (A) and (E) shows sub-CSF signal on T2WI, but its signal is still higher than the spleen signal. This lesion shows liquid signal on SDC, however it shows lower signal on DDVD in (G). This may be an artifact as the spleen also shows lower signal. In (C), the ‘semi-ring signal’ (a high signal half ring and a low signal half ring) around the lesion’s border and the modestly high signal inside the ring suggest the diagnosis of HG, as discussed in (13). The arrow pointed lesion in (I) shows CSF-signal on T2WI, markedly high signal on DDVD and on SDC. (B,F,J) are CE MRI, showing HG appearance. CE, contrast-enhanced; CSF, cerebrospinal fluid; DDVD, diffusion-derived ‘vessel density’; HG, hemangioma; MRI, magnetic resonance imaging; SDC, slow diffusion coefficient; T2WI, T2-weighted imaging.

Figure 11 Appearance of a liver HG (arrows) in three cases (A-C). All three lesions show very high signal on DDVD, and markedly high signal on SDC map in (A4) and (B4). In (C4), the HG lesion shows only high signal likely due to that its location is close to the diaphragm thus susceptible to respiratory motion. (A2,B2,C2) CE MRI, showing HG appearance. CE, contrast-enhanced; DDVD, diffusion-derived ‘vessel density’; HG, hemangioma; MRI, magnetic resonance imaging; SDC, slow diffusion coefficient; T2WI, T2-weighted imaging.

Figure 12 Four liver HGs in three patients located near the diaphragm (A-C). The lesions (yellow and blue arrows) show lower than anticipated signal on DDVD and SDC maps as these locations are more susceptible to respiratory motion. There are two HG lesions in (B). All lesions show signals lower than CSF on T2WI. However, the non-expansive and irregular shape of the lesions as shown in (A1) and (C1) still suggests HG. The lesion in (B) with blue arrow is too long-oval shaped to be an MFL. (A2,B2,C2) CE MRI, showing HG appearance. (A1) The case in group G of Figure 5C with the largest lesion size. CE, contrast-enhanced; CSF, cerebrospinal fluid; DDVD, diffusion-derived ‘vessel density’; HG, hemangioma; MFL, mass-forming lesion; MRI, magnetic resonance imaging; SDC, slow diffusion coefficient; T2WI, T2-weighted imaging

Figure 13 Three cases of liver HG patients (A-C) and a liver cyst patient (D). The patient in (C) has two HGs, and the patient in (D) has multiple cysts. On DDVD maps, it has been noted that a liquid focal lesion with ‘semi-ring signal’ (a high signal half ring and a low signal half ring) associated with modest or high internal signal suggests HG, while ‘semi-ring signal’ associated with low internal signal (the low signal half ring may thus be invisible) suggests cyst (13,14). CE, contrast-enhanced; DDVD, diffusion-derived ‘vessel density’; HG, hemangioma; SDC, slow diffusion coefficient; T2WI, T2-weighted imaging.

Figure 14 Four pyogenic liver abscess patients (A-D). The lesions (arrows) show mixed signals on DDVD and SDC. Liquid components show markedly high signal on SDC maps, and mixed high and low signals on DDVD maps due to respiratory motion artifacts. (A3,B3,C3,D1) Asterisks show edema beyond the lesion extent seen on T2WI. DDVD, diffusion-derived ‘vessel density’; SDC, slow diffusion coefficient; T2WI, T2-weighted imaging.

Discussion

A combination of SDC and DDVD can provide diffusion and perfusion information initially promised by intravoxel incoherent motion (IVIM) imaging, while with much faster data acquisition and allowing more convenient pixelwise mapping (3,15,18-21). It is noted that the addition of the diffusion gradients can induce a shortening of the ‘observed T2 value’ as compared with the T2 value when there is no diffusion gradient, and a greater b-value of the diffusion gradients is likely associated with a greater shortening of the ‘observed T2 value’ (22-24). The wide range of b-values applied for IVIM imaging and ADC imaging, for example, from b=0 to 800 s/mm², leading to IVIM metrics and ADC being heavily affected by T2, with the ‘observed T2 shortening effect’ being more apparent than when the native T2 is short (for example, muscle T2=32 ms, liver T2=40 ms, 3.0 T) (24). The intervals for the two b-values to calculate DDVD (for example, b=0 and 10 s/mm²) and SDC (for example, b=400 and 600 s/mm²) are shorter than typical IVIM and ADC imaging, thus the T2 effect for DDVD and SDC is potentially mitigated in one aspect (9,24,25). In addition to the liver, a combination of SDC and DDVD has been tested to evaluate parotid gland tumor classification and diagnose isocitrate dehydrogenase genotypes in diffuse gliomas, with promising results (26,27). The primary goal of this study is to assess the usability of SDC map and DDVD map for small liver lesion classification in the presence of respiratory motion. Ideally, liver MFL, cyst, and HG can all be separated if SDC mapping and DDVD mapping are both ‘perfect’.

In this study, most of the SDC maps (98% of MFLs and 92.5% of LLs) were usable (Table 2, Figure 2). Center 1 MFLs had a median size of 3.8 cm and Center 2 MFLs had a median size of 2.5 cm. The NEX for SDC mapping was four for Center 1 data (b₁=400 s/mm² and b₂=600 s/mm²) and six for Center 2 data (b₁=600 s/mm² and b₂=800 s/mm²). All Center 1 SDC maps were usable, and 96.7% (87/90) of the Center 2 SDC maps were usable. All the unusable SDC maps had a size ≤1.8 cm. The results in this study tentatively suggest that there is no difference in usability between Center 1 and Center 2 SDC maps. In this study, SDC was calculated with b₁=400 and b₂=600 s/mm² images for Center 1 data, with b₁=600 and b₂=800 s/mm² images for Center 2 data. According to liver IVIM observations, once b-values is ≥400 s/mm², the relationship between DWI signal decay and b-value increase follows a linear pattern except when DWI signal is contaminated by noises (28).

In this study, with overall 16.2% for MFL and 40.3% for LL being unusable (Figure 2A, Figure 4A), DDVD map usability rate remained a concern for small liver lesions. LLs were over smaller than MFL and particularly so for cysts. In addition, LLs were mostly from Center 1, with an NEX of two for DDVD mapping. For MFLs ≤2.2 cm, 52.6% of the Center 1 data DDVD were unusable while 31.1% of the Center 2 data DDVD were unusable. A trend was noted that by increasing NEX from two in Center 1 to six in Center 2, DDVD map usability rate increased (Figure 3). Besides the difference in NEX, the spatial resolutions and TE values are approximately the same between Center 1 DWI images and Center 2 DWI images. TR was 5,500 ms for Center 1 DWI and 2,700 for Center 2 DWI, though such a difference in TR is unlikely to have a major impact on DWI signal-to-noise ratio and signal-to-contrast ratio (29), the long TR in Center 1 images would in principle favors Center 1 images from the TR’s aspect. Thus, our results in this study suggested that it is more cost-effective to use a higher NEX than to use a longer TR as long as TR is already reasonably long.

For unusable DDVD maps, 60% (6/10) of MFL and 33.3% (9/27) of LLs were ≥1.6 cm, 40% (4/10) of MFL and 66.7% (18/27) of LLs were <1.6 cm (Figure 3A, Figure 4A). These results tentatively suggest that due to the higher signal of LLs, DDVD map can evaluate smaller LLs than MFLs.

With the data shown in Figure 2C, for MFLs from Center 2 which all had an NEX of six both for SDC mapping and DDVD mapping, DDVD maps had a usability of 84.4% and SDC maps had a usability of 96.7%. Thus, for the same lesions and the same NEX, the DDVD map usability rate was lower than the SDC map usability rate. The cause for the lower DDVD map usability rate (as compared to that of SDC map) is likely that, there is a higher degree of difference in image distortions between DWI_b0 and DWI_b10 than that between the two DWI images for SDC mapping. It is well recognized that EPI DWI images are associated with image distortions. The difference in image distortions may be more apparent between DWI_b0 (where the diffusion gradients are ‘off’) and DWI_b10 (where the diffusion gradients are ‘on’). This observation may also have relevance for the interpretation of liver ADC maps. Typically, ADC is calculated with b=0 s/mm² DWI image and a high b-value DWI image (such as b=800 s/mm²), or calculated with low b-value DWI image (such as b=50 s/mm²) and a high b=0 DWI image (such as b=800 s/mm²) (30). Following the observation in this study, for small liver lesions, ADC_b0b800 may suffer from a higher degree of image distortion (thus less reliable) than ADC_b50b800. This point deserves to be confirmed in future studies.

To address the issue of the low DDVD usability rate for small liver focal lesions, three approaches may be attempted. As noted above, even for small lesions, DDVD map usability rate can be improved by using a higher NEX. Thus, after T2WI is acquired and an MRI radiographer sees a liver focal lesion and wants to scan DWI for DDVD mapping, the radiographer can choose to increase the NEX even further (for example to an NEX of 8) for small lesions. As anticipated, we have earlier observed that lesions in the left liver, lesions close to the diagram, and lesions close to the anterior abdominal wall are more likely to suffer from respiration motion. NEX may be increased if a lesion is found at these locations during the anatomical imaging. Another approach will be to use single breathhold data acquisition. In one of our testings using a 3.0 T Philips scanner (Ingenia Elition X, Philips Healthcare, Best, Netherlands), we applied the following parameters: single-shot spin-echo type EPI sequence, TR =1,000 ms, TE =80 ms, spatial resolution 2 mm × 2 mm × 6 mm, or 2 mm × 3 mm × 6 mm, or 3 mm × 3 mm × 6 mm, number of slices =20, b=0 and 10 s/mm², breathhold duration =12 seconds. The acquired DDVD maps, as shown in Figures 15,16, appeared to be of good quality for the resolutions of 2 mm × 3 mm × 6 mm and 2 mm × 2 mm × 6 mm. However, we also noted that these parameters derived too low image signals for SDC mapping even for b=400 and 600 s/mm². The image distortion associated with EPI acquisition may also be partially overcome by turbo spin echo DWI. Compared with traditional EPI DWI used in the current study, turbo spin echo DWI can reduce image geometric distortions (31-33). This approach can be tested in the future.

Figure 15 Images of 4 sections of a normal liver acquired with a single breathhold of 12 seconds for b=0 and 10 s/mm² imaging. The spatial resolution is 2 mm × 3 mm × 6 mm for (A) and (B), 3 mm × 3 mm × 6 mm for (C) and (D). (A1,B1,C1,D1) DWI b=0 s/mm² image. (A2,B2,C2,D2) DWI b=10 s/mm² image. (A3,B3,C3,D3) DDVD map. DDVD, diffusion-derived ‘vessel density’; DWI, diffusion weighted imaging.

Figure 16 Images of 3 sections of a normal liver acquired with a single breathhold of 12 seconds for b=0 and 10 s/mm² imaging. The spatial resolution is 2 mm × 2 mm× 6 mm. (A1,B1,C1) DWI b=0 s/mm² image. (A2,B2,C2) DWI b=10 s/mm² image. (A3,B3,C3) DDVD map. DDVD, diffusion-derived ‘vessel density’; DWI, diffusion weighted imaging.

This study assessed the roles of SDC map and DDVD map for classifying liver MFL, cyst and HG. Not surprisingly, all MFLs were classified as MFL with T2WI, and SDC/DDVD classified 97.3% (144/148) of them being MFL (Figure 5A). The diagnostic hallmark for liver cysts is that they show on every structural sequence the same signal intensity as CSF. However, due to the partial volume effect, small lesions do not always show typical CSF-signal. In this study, SDC/DDVD improved the confidence of 27.3% (6/22) of cysts being LL (Figure 5C). On T2WI, both CSF-signal and sub-CSF signal are common for HG regardless of its size (34-36). In this study, 18/45 (40%) of the HGs were CSF-signal on T2WI, while 60% of the HGs were not typical CSF-signal on T2WI. A total of 6 HGs (13.3%) were confidently classified as LL by SDC/DDVD, where T2WI was only suggestive of them being LL. A total of 11 HGs (24.5%, median size: 1.70 cm, range: 0.5–8.1 cm) had sub-CSF-signals on T2WI, they were confidently classified as LL with the further contribution by SDC/DDVD. Therefore, SDC/DDVD can support the diagnosis of 24.5% of HGs being LL and confirm the diagnosis of 13.3% of the HGs being LL. In this study, based on T2WI, 26.7% (12/65, groups E, F, and G lesions in Figure 5) would require further CE imaging to confirm the diagnosis being LL. However, in current routine practice, a much higher proportion of HGs would undergo CE imaging. Based on a combination of T2WI and SDC/DDVD, only 6 HGs (13.3%, groups F and G lesions in Figure 5C) would require further CE imaging to confirm the diagnosis being LL.

For LLs, the differentiation between cyst and HG was only partially successful (Figure 6). This could partially be due to the fact that most of LLs were from Center 1. The differentiation between cyst and HG relies on the quality of DDVD map, while Center 1 data had an NEX of only two. Only half of the cysts could be confidently diagnosed, on the other hand 74.4% of the HGs were confidently diagnosed as HG. The lesion size was overall larger for HGs. Moreover, some of the HGs had a ‘non-expansive’ shape which would suggest the diagnosis of a LL being HG. The differentiation between cyst and HG is usually not critical for clinical management.

There are many limitations to this study. The scope of lesion variety remained limited in study. The MFLs in this study were of typical T2WI signals and thus we had no difficulty in classifying them. In practice, false positive diagnosis of LL with MRI could be caused by metastatic lesions from endocrine tumors that may have an extremely high signal intensity on T2WI due to their hypervascularity (37-39). Necrotic metastases may also show very high signal intensity on T2WI, but these are mostly not as sharply marginated with a thin wall as cysts or HGs (14,40,41). Probably there was no sclerosing HG in this study as all HGs showed typical CE imaging appearance. Some other rare pathologies, such as angiosarcoma and peliosis hepatis, may mimic HG (42-44). Thus, this study remains a proof-of-principle study, rare appearance of various liver focal lesions should be studied in the future. From technical point of view, we did not apply any image registration algorithms to mitigate misregistration between b-values. However, with the non-rigid pattern of motion of the liver due to respiration motion, image registration algorithm tends not to perform well, and such methods are not routinely applied in clinical practice. The diagnostic decisions in this study were made subjectively, and decisions were thus influenced by the skill and expertise of the readers. Deciding SDC map or DDVD map being as ‘usable’ or ‘unusable’ was also subjective. Inter-reader agreement analysis (such as Kappa value analysis) was not obtained. In fact, the proportions of usable or unusable maps and the proportions of lesions which could or could not be reliably classified will vary according to the particularity of the dataset and the expertise of the readers. The percentages described in this article could only serve a rough principle and show the trend direction. Thus, one of the limitations is that no formal statistical analysis was conducted in this study. It should be noted that, it is not our argument that SDC/DDVD can replace contrast enhanced liver imaging, instead we argue that the application of SDC/DDVD may save some patients from contrast enhanced liver imaging particularly for those with larger liver lesions.

Conclusions

In conclusion, this study shows that most of the SDC maps were usable, while a substantial portion of DDVD maps were unusable for small liver lesions. This is likely due to that there is a higher degree of difference in image distortions between DWI_b0 and DWI_b10 than between the DWI images for SDC mapping. Lesion size has an important impact on the usability of SDC and DDVD maps. For small lesions, DDVD map usability rate can be improved by using a higher NEX. In this study, on T2WI, 40% of HGs were typical CSF-signal, while 60% of HGs were not typical CSF-signal. For the HGs which were not typical CSF-signal (n=27), SDC and/or DDVD confirmed the diagnosis of 22.2% (6/27) of them being LL independent of T2 weighted imaging, and further supported the T2 weighted imaging-based diagnosis of 40.7% (11/27) of them being LL.

Acknowledgments

Parts of the research was conducted at CUHK MRI Facility, which received instruments support from Kai Chong Tong, HKSAR Research Matching Grant Scheme and Department of Imaging and Interventional Radiology, The Chinese University of Hong Kong.

Footnote

Data Sharing Statement: Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2026-1-0084/dss

Peer Review File: Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2026-1-0084/prf

Funding: This study was supported by the Natural Science Research Projects in Higher Education Institutions in Anhui Province (No. 2023AH050584) and Natural Science Foundation of Anhui Medical University (No. 2022xkj214).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jgo.amegroups.com/article/view/10.21037/jgo-2026-1-0084/coif). Y.X.J.W. is the founder of Yingran Medicals Ltd., which develops medical image-based diagnostics software. The metric DDVD is associated with a granted China patent (ZL201910125747.2, inventorship). Y.X.J.W. reports that the metric SDC is associated with a pending patent application (inventorship). 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. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the institutional ethics committee of The Fifth Affiliated Hospital of Anhui Medical University (reference No. KY2025058) and the institutional ethics committee of The Second Hospital of Nanjing (reference No. 2026-LS-ky009). Informed consent was obtained from all 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. 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]
2. 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;15:1710-8. [Crossref] [PubMed]
3. 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]
4. Hu GW, Zheng CJ, Zhong WX, et al. Usefulness of diffusion derived vessel density computed from a simplified IVIM imaging protocol: An experimental study with rat biliary duct blockage induced liver fibrosis. Magn Reson Imaging 2021;84:115-23. [Crossref] [PubMed]
5. 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]
6. 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. [Crossref] [PubMed]
7. 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]
8. 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]
9. 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]
10. Li CY, Zheng CJ, Duan XH, et al. Separation of liver focal nodular hyperplasia (FNH) and liver malignant tumors by a combination of T2-weighted imaging signal and three diffusion magnetic resonance metrics of diffusion-derived vessel density, slow diffusion coefficient, and apparent diffusion coefficient. J Gastrointest Oncol 2026;17:20. [Crossref] [PubMed]
11. Xu XY, Zeng WT, Sun MH, et al. Reliable diagnosis of liver focal nodular hyperplasia by a combination of T2 weighted signal and three diffusion magnetic resonance metrics of diffusion-derived ‘vessel density’, slow diffusion coefficient, and apparent diffusion coefficient: a validation study of two centers’ data. Quant Imaging Med Surg 2026;16:80. [Crossref] [PubMed]
12. Wáng YXJ, Zheng CJ, Xu XY, et al. Diffusion-derived “Vessel Density” (DDVD) and Slow Diffusion Coefficient (SDC) Signal Features of Mass-forming Intrahepatic Cholangiocarcinoma Compared to Those of Hepatocellular Carcinoma and Liver Metastases: Preliminary Results. Rofo 2026; Epub ahead of print. [Crossref] [PubMed]
13. 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]
14. Hu GW, Li CY, Ling GH, et al. 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. J Gastrointest Oncol 2025;16:2336-56. [Crossref] [PubMed]
15. 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]
16. Baltzer PA, Schelhorn J, Benndorf M, et al. Diagnosis of focal liver lesions suspected of metastases by diffusion-weighted imaging (DWI): systematic comparison favors free-breathing technique. Clin Imaging 2013;37:97-103. [Crossref] [PubMed]
17. Li XM, Ma FZ, Quan XY, et al. Repeatability and reproducibility comparisons of liver IVIM imaging with free-breathing or respiratory-triggered sequences. NMR Biomed 2024;37:e5080. [Crossref] [PubMed]
18. 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]
19. 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]
20. Ma FZ, Wáng YXJ. A New Approach for Estimating Liver IVIM Fast Diffusion Coefficient in Two-Compartment Model and Fast Diffusion Fraction in Three-Compartment Model: Theory and In Vivo Validation of Two Centers’ Data. NMR Biomed 2026;39:e70221. [Crossref] [PubMed]
21. Li CY, Chen L, Ma FZ, et al. High performance of the diffusion magnetic resonance imaging biomarker diffusion-derived ‘vessel density’ (DDVD) for separating placentas associated with pre-eclampsia from placentas in normal pregnancy. Quant Imaging Med Surg 2025;15:1-14. [Crossref] [PubMed]
22. Egnell L, Jerome NP, Andreassen MMS, et al. Effects of echo time on IVIM quantifications of locally advanced breast cancer in clinical diffusion-weighted MRI at 3 T. NMR Biomed 2022;35:e4654. [Crossref] [PubMed]
23. 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]
24. Wáng YXJ. An explanation for why tissues with T2 of around 70 milliseconds always measure low MRI apparent diffusion coefficient (ADC). Quant Imaging Med Surg 2025;15:12948-56. [Crossref] [PubMed]
25. 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]
26. Zheng WY, Shi YT, He YY, et al. A novel MRI diffusion metric ‘slow diffusion coefficient’ (SDC) for diagnosing isocitrate dehydrogenase (IDH) genotype in diffuse gliomas: initial promising results. BMC Med Imaging 2025;25:427. [Crossref] [PubMed]
27. Yao DQ, King AD, Zhang R, et al. Added value of slow diffusion coefficient (SDC) for the separation of benign and malignant parotid gland tumors by diffusion weighted imaging: preliminary results. BMC Med Imaging 2025;25:497. [Crossref] [PubMed]
28. Chevallier O, Zhou N, He J, et al. Removal of evidential motion-contaminated and poorly fitted image data improves IVIM diffusion MRI parameter scan-rescan reproducibility. Acta Radiol 2018;59:1157-67. [Crossref] [PubMed]
29. Xu FY, Wáng YXJ. The Impact of TE and TR on Apparent Diffusion Coefficient and Diffusion-derived Vessel Density Quantification: Studies of Healthy Liver and Spleen. Rofo 2026; Epub ahead of print. [Crossref] [PubMed]
30. Guiu B, Cercueil JP. Liver diffusion-weighted MR imaging: the tower of Babel? Eur Radiol 2011;21:463-7. [Crossref] [PubMed]
31. Yoshizako T, Yoshida R, Asou H, et al. Comparison between turbo spin-echo and echo planar diffusion-weighted imaging of the female pelvis with 3T MRI. Acta Radiol Open 2021;10:2058460121994737. [Crossref] [PubMed]
32. Mori N, Mugikura S, Miyashita M, et al. Turbo Spin-echo Diffusion-weighted Imaging Compared with Single-shot Echo-planar Diffusion-weighted Imaging: Image Quality and Diagnostic Performance When Differentiating between Ductal Carcinoma in situ and Invasive Ductal Carcinoma. Magn Reson Med Sci 2021;20:60-8. [Crossref] [PubMed]
33. Liu Y, Hu L, Qian L, et al. Intravoxel incoherent motion diffusion-weighted imaging in nasopharyngeal carcinoma: comparison of the turbo spin-echo and echo-planar imaging techniques. Quant Imaging Med Surg 2024;14:9101-11. [Crossref] [PubMed]
34. Sasaki K, Ito K, Koike S, et al. Differentiation between hepatic cyst and hemangioma: additive value of breath-hold, multisection fluid-attenuated inversion-recovery magnetic resonance imaging using half-Fourier acquisition single-shot turbo-spin-echo sequence. J Magn Reson Imaging 2005;21:29-36. [Crossref] [PubMed]
35. Chanthasittiporn T, Sirikhunseang M, Geraplangsub S, et al. Hepatic Cyst versus Hemangioma: Useful T2-FLAIR Image for Differentiation. Bangkok Med J 2017;13:41-4. [Crossref]
36. Burim CV, Pecci Neto L, Torlai FG, et al. Differentiation between simple cyst and hepatic hemangioma utilizing T2-weighted magnetic resonance imaging with gradient-echo (b-FFE) technique. Radiol Bras 2008;41:355-60. [Crossref]
37. Soyer P, Gueye C, Somveille E, et al. MR diagnosis of hepatic metastases from neuroendocrine tumors versus hemangiomas: relative merits of dynamic gadolinium chelate-enhanced gradient-recalled echo and unenhanced spin-echo images. AJR Am J Roentgenol 1995;165:1407-13. [Crossref] [PubMed]
38. Ahn SJ, Choi SJ, Kim HS, et al. Well-Differentiated Pancreatic Neuroendocrine Tumor with Solitary Hepatic Metastasis Presenting as a Benign Cystic Mass: A Case Report. J Korean Soc Radiol 2014;70:351-4. [Crossref]
39. Semelka RC, Worawattanakul S, Noone TC, et al. Chemotherapy-treated liver metastases mimicking hemangiomas on MR images. Abdom Imaging 1999;24:378-82. [Crossref] [PubMed]
40. Mortelé KJ, Ros PR. Cystic focal liver lesions in the adult: differential CT and MR imaging features. Radiographics 2001;21:895-910. [Crossref] [PubMed]
41. Chen MY, Bechtold RE, Savage PD. Cystic changes in hepatic metastases from gastrointestinal stromal tumors (GISTs) treated with Gleevec (imatinib mesylate). AJR Am J Roentgenol 2002;179:1059-62. [Crossref] [PubMed]
42. Mathew RP, Sam M, Raubenheimer M, et al. Hepatic hemangiomas: the various imaging avatars and its mimickers. Radiol Med 2020;125:801-15. [Crossref] [PubMed]
43. Iannaccone R, Federle MP, Brancatelli G, et al. Peliosis hepatis: spectrum of imaging findings. AJR Am J Roentgenol 2006;187:W43-52. [Crossref] [PubMed]
44. Itai Y, Teraoka T. Angiosarcoma of the liver mimicking cavernous hemangioma on dynamic CT. J Comput Assist Tomogr 1989;13:910-2. [Crossref] [PubMed]