Researchers interested in susceptibility-weighted imaging (SWI), quantitative susceptibility mapping (QSM) and the benefits of these imaging techniques to classify melanoma metastasis.T1-weighted contrast-enhanced and T2-weighted images are core sequences in the detection of metastatic melanoma in the brain (1). It was reported that some malignant melanoma metastases appear hyperintense in native T1w images (2). This pattern may be either resulting from the presence of melanin (3) or hemorrhage (2). Furthermore, intra-metastatic bleeding might not be visible on T1w-native images since the T1 signal depends on the age of hemorrhage. We investigate the feasibility of QSM for the detection and quantification of bleeding of brain metastases of malignant melanoma.The study was approved by the local ethics committee, every patient gave written informed consent. Thirty-two patients with melanoma stage IV diagnosed with brain metastases (32‑77 years; mean age 56.7 years) were measured 83 times in total on a 1.5 T whole-body MR system (Magnetom Symphony, Siemens Healthcare) with a 12-channel head-matrix coil during routine clinical workup. The clinical protocol included a T1-weighted SE (pre- and post-contrast), a T2-weighted TSE, a diffusion-weighted EPI, a T2-weighted FLAIR and a fully flow-compensated 3D gradient-echo sequence. Imaging parameters for the 3D-GRE were: flip angle=15°, TR/TE=49/40ms, matrix=320x250x72, voxel size=0.75x0.88x1.9 mm³, bandwidth=80 Hz/pixel, partial parallel imaging (GRAPPA) with acceleration factor R=2 and 24 reference lines. T2w-FLAIR: flip angle=150°, TI/TR/TE=2340/8000/98 ms, matrix=320x256x30, voxel size=0.72x0.9x5 mm3, slice gap=0.5 mm, bandwidth=130 Hz/pixel, averages=2. After non-contrast enhanced sequences were acquired, a T1-weighted SE sequence was acquired (flip angle=90°, TR/TE=500/17 ms, matrix=320x260x39, voxel size=0.72x0.79x4 mm3, slice gap=0.4 mm, bandwidth=130 Hz/pixel) after application of Gadobenate Dimeglumine (MultiHance, Bracco Imaging, Italy) for contrast enhancement. Phase images were combined using the vendor-provided adaptive combine method. Brain masks were generated from magnitude images using FSL-BET (4). Phase images were unwrapped using a Laplacian-based phase unwrapping (5,6,7). The background field was removed with V-SHARP (6,7) (with kernel size up to 25 mm). Susceptibility maps were calculated using iLSQR (4,8) and iLSQR-parameters recommended for effective removal of streaking artifacts and accurate quantification of susceptibility were used (8). Susceptibility maps of patients with bleedings were calculated using superposition (9) to minimize artifacts. Susceptibility maps were referenced to cerebrospinal fluid from the atrium of the lateral ventricles. Regions of interest for different appearing metastases were drawn on QSM using MITK Software (10,11).25 of the 32 patients showed hypointensities in SWI and hyperintensities in QSM in at least one measurement (Fig.1, Fig.2a,b). Twenty-two patients showed T1w-native hyperintense brain metastases (Fig.1, Fig.2c,f). Using SWI/QSM, in 17 patients, T1w-native hyperintensities could be related to hemorrhagic events (Fig.1f-o, Fig.2a-c), whereas in 5 patients, they could be related to melanoma metastases with no detectable bleedings (Fig.1a-e). 10 patients with T1w-native hyperintensities related to bleedings showed variable contrast in T1w native and QSM related to bleeding age (Fig.3). Mean susceptibility of 20 hemorrhagic metastases which were not T1w-native hyperintense was 0.355±0.097 ppm and of 11 hemorrhagic metastases which were T1w-native hyperintense 0.239±0.123 ppm. 11 patients revealed new bleedings as determined by SWI/QSM while undergoing therapy. In 6 patients, SWI-hypointensites/QSM-hyperintensities showed only partial signal alteration of the lesion indicating incomplete bleeding (Fig.2a-c). One patient showed a metastasis in putamen (Fig.2d-f), that was T1w-native hyperintense (Fig.2f). We found that the mean susceptibility of the metastasis (-0.005 ppm) is less paramagnetic than the mean susceptibility of the same region of putamen in the other hemisphere (0.111 ppm).Metastases of melanoma show a very high propensity to bleed in the present cohort in accordance with the literature (12,13). Hyperintensity on QSM images easily identifies bleedings which may both be T1 hyper- or hypointense (Fig.1) thus representing different age of hemorrhage. The ROI-based assessment of QSM allows for quantitative assessment of intra-metastatic bleeding where SWI only provides qualitative results. Therefore, QSM could be discussed as treatment monitoring tool to indicate response to therapy of brain metastases, since many patients developed new intra-metastatic bleedings while undergoing therapy. Surgery is only indicated in patients with single lesions, thus histopathological ground truth could not be obtained in the majority of patients, mostly presenting with multiple metastases. However, SWI was reported with histopathology to reliably visualize paramagnetic blood in the brain (14) and these reports are used as reference for the present study. Our results also suggest that the susceptibility of non-bleeding melanoma metastases is diamagnetic especially compared to structures known for paramagnetism like putamen (15).QSM maps should be discussed as part of the regular clinical protocol for metastases of melanoma since they allow rapid detection and quantification of intra-metastatic bleeding.