1. Introduction
The most common malignant liver lesions are hepatic metastases, resulting in the most common indication for liver imaging [
1]. As such, diagnosis with exact detection and localization of liver metastases are essential, as these can affect the clinical course of the malignant disease. Because of its high specificity and lack of radiation exposure, magnetic resonance imaging (MRI) has rapidly become the modality of choice for the detection and characterization of liver lesions [
2,
3]. The most sensitive MRI sequences for metastasis detection are diffusion-weighted imaging (DWI) and hepatobiliary phase (HP) images after intravenous application from a liver-specific contrast agent. Indicative features such as peripheral ring enhancement, diffusion restriction, and hypointensity on HP are typically seen in hepatic metastases [
4,
5,
6].
Over the past decade, numerous studies have confirmed the diagnostic value of DWI in oncologic and non-oncologic applications [
3,
4,
5,
6,
7,
8,
9]. The strength of DWI lies in its ability to qualitatively and quantitatively assess the diffusion properties of tissues and its ability to reflect tumor microstructure. In short, malignant tumors have higher cellularity, with resulting diffusion restriction detected by DWI [
3,
10]. Thus, DWI plays an essential role in tumor detection and characterization based on the high contrast between the lesion and surrounding tissue [
11]. DWI increases the sensitivity and specificity for lesion detection in the liver [
12,
13]. However, DWI presents some limitations, such as low image resolution and poor signal-to-noise ratio (SNR), partly due to the short echo time (TE). Moreover, DWI can also be prone to image artifacts such as blurring or ghosting artifacts [
5].
In hepatic diagnostics, DWI at higher b-values leads to low background signals from the normal liver parenchyma, which increases the contrast between the background liver and lesions. In clinical routine, b-values up to 800–1000 s/mm
2 are acquired as high b-values [
14]. There is still a lack of data for the diagnostic accuracy of high b-values over 1000 s/mm
2 to date.
Lately, there has been a growing interest in the clinical assessment of computed DWI (c-DWI), especially in the field of oncology, with very promising results [
8,
14,
15,
16]. C-DWI is a mathematical postprocessing technique that produces virtually high b-values images by using real DWI data with at least two different lower b-values [
17]. With this method, higher diffusion effects and SNR, by fitting input data with shorter TE, can be generated. Thus, the above-mentioned disadvantages of standard DWI could be avoided. Furthermore, c-DWI does not need additional acquisition time [
17,
18]. In other tumors, such as prostate cancer and breast cancer, higher b-value images showed higher conspicuity compared to standard b-value images [
15,
17].
For liver imaging, Shimizu et al. showed that there was no significant difference between acquired DWI and c-DWI images with b-values of 1000 s/mm
2 in the detection of hepatic metastases at 3T [
19].
Kawahara et al. validated the additional benefit of c-DWI with b-values of 1000 s/mm
2 derived by a DWI obtained with lower b-values of 500 s/mm
2 for the detection of hepatic metastases at 1.5T scanner [
18].
One of the promising diagnostic abilities of c-DWI on higher b-values over 1000 s/mm
2 is reduced T2 shine-through effect compared to standard DWI. This effect can lead to the misdiagnosis of benign, cystic lesions as malignant lesions due to the hyperintensity on the b 800 image. Another important aspect is that on higher b-value images, malignant lesions can have better conspicuity compared to standard DWI [
5,
15].
However, despite the promising diagnostic abilities of c-DWI, there is still no reliable data regarding the diagnostic value of c-DWI with higher b-values over 1000 s/mm2 for the diagnosis of hepatic metastases. This is of interest, as in other body localizations and tumor entities, a promising benefit of higher b-value DWI was proposed.
The aim of this study was therefore to investigate the usefulness of high b-value c-DWI in comparison to acquired DWI in patients with hepatic metastases and to compare different high b-values in regard to their visibility and extension of the liver lesions.
4. Discussion
The present study investigated the possible benefit of high b-values c-DWI in the diagnosis of hepatic metastasis. As shown, high b-values up to 1500 s/mm2 can potentially be used in clinical routine but do not provide a superior detectability compared to standard 800 and/or 1000 s/mm2 DWI images. As a second key finding, c-DWI images above 2000 s/mm2 cannot be recommended due to poor image quality.
Correct diagnosis of liver metastasis is very important for treatment planning in oncologic patients because even the diagnostic suspicion of a small liver metastasis can change the patient’s course from curative intended treatment to a palliative setting [
1].
There is no doubt regarding the clinical benefit of DWI in liver imaging. It was shown that ADC values might be useful in the discrimination of benign to malignant liver lesions [
20]. For liver metastasis in particular, DWI was identified to be very sensitive, even for small lesions below 1 cm in size [
21]. It was clearly stated that metastasis detection with the addition of DWI and HP is superior to the conventional MRI technique, with a reported sensitivity of 0.88 in an analysis on colorectal liver metastasis of 1121 patients [
22].
The definition of the best pair of b-values is of great clinical interest and can change the diagnostic abilities of DWI. In a study by Kaya et al., 124 hepatic lesions were evaluated, with 7 different b-values ranging from 0 to 1000 s/mm
2 [
4]. The authors concluded that b-values of 0 and 800 s/mm
2 are preferable. In another meta-analysis of 1775 hepatic lesions, an overall pooled sensitivity of 0.86 and a specificity of 0.82 were reported for discrimination between malignant and benign liver lesions [
23]. The authors further compared standard DWI obtained with b-values up to 1000 s/mm
2 to low b-values in subanalyses, with significantly better accuracy for DWI based upon b-values of 800 and 1000 s/mm
2. However, the authors concluded that there is a definite need for studies investigating the possible benefit of higher b-values [
23]. A preliminary study on malignant primary liver tumors and metastases compared the diagnostic utility of different ADC values derived from b-values of 400, 800, 1600 and 2000 s/mm
2 [
24]. The authors could not identify significant differences of the different ADC values, but they did not assess the image quality of the high b-value images, as in the present study. In another interesting study, the addition of DWI to HP images significantly improved the diagnostic performance for residents [
25]. In short, there is no doubt regarding the benefit of DWI in liver MRI. Yet, there are still uncertainties regarding the b-value choice.
There is a growing interest in c-DWI around oncologic imaging. The principal hypothesis is that the generated high b-values images allow a better lesion contrast with reduced T2 shine-through effect compared to standard DWI [
17]. Due to the higher cellularity of malignant tumors, the diffusion restriction can be better visualized by high b-value images, as high b-value images are more sensitive to kurtosis effects [
15,
17]. As another important point, there is no need for further acquisition time.
For ischemic stroke imaging, there are reliable data that high b-value DWI can better display diffusion restriction, which was shown for acquired and computed images [
6,
7]. Notably, the b-value of 2000 s/mm
2 had the best image quality [
6,
7].
For prostate cancer, a high b-value DWI of 1500 s/mm
2 up to 2000 s/mm
2 is recommended for the standard MRI protocol due to its superior diagnostic abilities compared to standard DWI. There are enough data that the computed high b-value image is as good as the acquired b-value image, as shown in a recent study employing b-values of 2000 s/mm
2 [
26]. There were even results that the c-DWI 2000 s/mm
2 image might be superior in regard to image quality compared to the acquired one [
27].
Similar results were reported for breast cancer patients [
28]. The high c-DWI images up to 2000 in one study [
29] and 2500 s/mm
2 in another study [
28] were of similar diagnostic quality compared to acquired images. For pancreatic cancer, it was recently published that c-DWI images with b-values of 1500 and 2000 s/mm
2 are superior in visualization compared to standard DWI [
30].
However, only one report was published regarding c-DWI in liver imaging. Kawahara et al. used a 1.5 T scanner and evaluated the diagnostic benefit of c-DWI b-values images of 1000 s/mm
2 based on 56 patients with hepatic metastases [
18]. The study could show that combined c-DWI and acquired DWI of 1000 s/mm
2 is superior to acquired DWI alone. Yet, the authors did not evaluate c-DWI images with higher b-values.
Therefore, there are still no data regarding the possible benefit of high b-value DWI over 1000 s/mm2 for liver MRI. The present study provides new insight for c-DWI that contrary to other tumor entities, there is no diagnostic benefit of higher b-value images for liver metastasis based upon subjective measurements.
This might be caused by respiratory motion artifacts with possible misalignment of the acquired b-values, which could have an influence on c-DWI image quality [
17]. Moreover, the influence of cardiac pulsation has a relevant effect, especially on high b-value images, which are sensitive to microscopic motion [
17].
The present study measured tumor sizes on the different DWI images. Notably, metastases were smaller on c-DWI images of 2000 s/mm2 and above. This finding should be kept in mind when reporting the correct size of the metastases, and one should not use c-DWI images over 1500 s/mm2 for the measurement. There are several limitations of the present study to address. First, it is a retrospective study with possible inherent bias based upon a relatively small sample size. The reading was performed with the knowledge that liver metastasis is present, which could have an influence on the results. Second, due to the study design, we could not compare the c-DWI images with acquired high b-value images. It is therefore not known whether acquired high b-value images are superior compared to standard DWI and c-DWI images. Third, we could not perform subanalyses for primary tumors other than colorectal cancer due to the small sample size. Fourth, the present analysis was based on a subjective assessment of the DWI images. We did not assess quantitative parameters, such as SNR. Fifth, the volume assessment was performed as a columnar measurement, which is not as correct as a full volumetry. However, the performed measurement can easily be translated into clinical routine.