Improving small brain metastasis detection with composite and augmented T₁-weighted imaging in contrast-enhanced multi-parametric MRI

Introduction

Brain metastases are the most common tumors of the central nervous system and typically arise from the hematogenous spread of circulating tumor cells, most commonly from lung cancer (40–50%), to the brain microvasculature (1). Early and accurate detection is critical for guiding treatment strategies, which often include a combination of surgery, radiotherapy, and systemic therapies (2). Moreover, earlier identification of brain metastases significantly improves patient prognosis, as untreated lesions are associated with rapid neurological deterioration, cognitive impairment, and reduced quality of life (3). Therefore, improving the accuracy and sensitivity of diagnostic imaging is imperative for early therapeutic decisions and ultimately improving clinical outcomes.

Due to its excellent soft-tissue contrast and high-resolution depiction of tissue anatomy (4), contrast-enhanced magnetic resonance imaging (MRI) is the gold standard for brain metastasis detection (5-7). Gadolinium-based contrast-enhanced T₁-weighted (T₁W) imaging, which exploits blood-brain barrier disruption in metastatic lesions, is widely used to detect brain metastases (7-9), providing clear lesion delineation with relatively high accuracy (10). Nevertheless, current T₁W imaging has limited sensitivity for detecting small metastases, particularly those with a diameter of less than 5 mm (11,12). To address this limitation, various advanced imaging sequences have been developed to optimize brain metastasis detection (6).

Recently, a gradient echo (GRE)-based MULTI-parametric MRI with fLEXible design (MULTIPLEX) was introduced, featuring a dual repetition time (TR), dual flip angle (FA), and multi-echo design (13). The composite T₁-weighted (cT₁W) image is derived from the combination of all echo images from the medium FA of the MULTIPLEX sequence, using the sum-of-square method to mitigate the T₂ star-weighted effect from later echoes. The resulting cT₁W imaging displays T₁W contrast with an elevated signal-to-noise ratio (SNR) (14). Meanwhile, the augmented T₁-weighted (aT₁W) image, which is defined as the ratio of the T₁W and proton density-weighted (PDW) signals, was recently developed to achieve computationally improved T₁W contrasts based on routine GRE signals (15). Thus, the MULTIPLEX-aT₁W images are calculated by first dividing the medium FA signals by the small FA signals for each TR separately, and then averaging the results from both TRs to yield the final aT₁W images.

The MULTIPLEX-derived cT₁W and aT₁W images have demonstrated clinical potential in the detection and prediction of various neurological disorders (16-19). However, to our knowledge, the utility of MULTIPLEX-cT₁W and -aT₁W imaging in the assessment of brain metastases has not yet been investigated. In the current study, we thus employed this advanced MULTIPLEX technique in contrast-enhanced MRI to investigate its feasibility for detecting small brain metastases. The MULTIPLEX-cT₁W and -aT₁W imaging were compared with T₁W imaging based on a magnetization-prepared rapid gradient-echo (MPRAGE) sequence, which is commonly and widely used to detect metastases and evaluate image quality in clinical brain imaging. We hypothesized that compared with MPRAGE-T₁W imaging, MULTIPLEX-derived imaging would demonstrate superior detection performance for small metastases and overall image quality improvement. Our findings may contribute to the development and optimization of brain tumor imaging protocols, potentially extending to broader neuro-oncological applications. We present this article in accordance with the STARD reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-aw-2474/rc).

Methods

Patients

This prospective study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of Nanjing Drum Tower Hospital (approval No. 2020-379-01, AF/SC-08/03.0), and informed consent was obtained from all individual participants.

Patients diagnosed with malignant lung cancer, who underwent contrast-enhanced MRI assessments for the evaluation of intracranial metastases using both MULTIPLEX and MPRAGE-T₁W imaging between May 2022 and May 2023 were consecutively enrolled in the study. Patients were excluded if they met any of the following criteria: (I) they had excessive motion artifacts on either MULTIPLEX or MPRAGE-T₁W images; and/or (II) they had incomplete imaging sequences. All enrolled patients received the same chemotherapy of Pemetrexed disodium via intravenous administration. After treatment, each patient underwent at least two follow-up MRI examinations. All enrolled patients underwent both MULTIPLEX and MPRAGE-T₁W imaging within 10 minutes. Patients’ demographic characteristics, including age and sex, were collected from clinical records.

MRI acquisition protocol

The imaging data were obtained on a 3.0 T MRI scanner (uMR790; United Imaging HealthCare, Shanghai, China) using a 32-channel head coil. One minute after the intravenous administration of gadobutrol (Gadovist, Bayer Schering Pharma, Berlin, Germany) at a standard dose of 0.1 mL/kg body weight and a rate of 1.5 mL/s, the two post-contrast sequences were scanned consecutively in random order to avoid timing bias from the decreasing concentration of gadolinium-based contrast material over time. The MULTIPLEX sequence was performed with the following parameters: dual FAs (FA₁/FA₂ =4°/16°), dual TRs (TR₁/TR₂ =8.45/35.75 ms), seven echoes with echo times (TEs) ranging from 4.15 to 31.45 ms, and ΔTE = 4.55 ms (bipolar readouts), bandwidth =240 Hz/pixel, field of view =224×190×64 mm³, voxel size =1.03×0.82×2.00 mm³, and total acquisition time =4:48 minutes. MULTIPLEX images were reconstructed and processed online using in-house C++ programs built on the Automatic Differentiation using Expression Template platform (United Imaging HealthCare), as described previously (13). Briefly, MULTIPLEX-cT₁W images are yielded from the combination of all echo images from the medium FA using the sum-of-square method. MULTIPLEX-aT₁W signals are inversely proportional to T₁ values, and are calculated separately from both TRs and then averaged to yield the final aT₁W images (13). The MPRAGE-T₁W sequence was performed using MPRAGE, with the following parameters: TR =7.9 ms, TE =3.1 ms, FA =10°, bandwidth =250 Hz/pixel, field of view =256×232×176 mm³, voxel size =1×1×1 mm³, with the vendor-provided artificial intelligence-assisted compressed sensing (ACS) algorithm dramatically shortening the acquisition time to 1:13 minutes. The imaging parameters are summarized in Table S1. Patient head motion was monitored concurrently and continuously throughout the entire scanning session. The scans were visually inspected for motion artifacts by the radiologic technician retrospectively.

Image analysis

Due to the inconsistent acquisition resolution, all images were first resampled to a resolution of 1×1×2 mm³. However, it should be noted that because lesion visibility and induced partial-volume effects may introduce bias, a simple resampling procedure might not ensure a fully fair comparison. The resampled images were subsequently assessed by two radiologists independently (X.L. and X.H., with 11 and 8 years of neural MRI experience, respectively), who were blinded to the clinical information and the imaging type. All images were randomized and reviewed via a picture archiving and communication system workstation (Neusoft, Shenyang, China). A pre-reading session was conducted to reach an evaluation agreement between the readers.

Subjective image quality was scored on a four-point Likert scale, graded as follows: 1 = poor (uninterpretable images), 2 = fair (interpretable but moderately degraded), 3 = good (interpretable with minimal artifacts), and 4 = excellent (interpretable with no artifacts) (20). Diagnostic confidence for detecting enhanced brain metastases was also assessed using a four-point Likert scale, graded as follows: 1 = poor (lesion border delineation), 2 = fair (definition of lesion extent), 3 = good (visualization of internal lesion morphology), and 4 = excellent (lesion contrast enhancement compared with the surrounding normal tissues) (21).

Objective image quality was quantitatively assessed using the SNR and contrast-to-noise ratio (CNR). Regions of interest (ROIs) of approximately consistent size were placed on the brain metastases, the head of the caudate nucleus [representing grey matter (GM)] and the genu of the corpus callosum [representing white matter (WM)] on the axial view (22,23). Lesions with homogeneous solid enhancement and clear margins were included in the ROIs. Circular ROIs were placed at the center of enhancing lesions to avoid partial-volume effects. The SNR and CNR were defined and calculated according to the following formulas:

SNRROI=μROIσROI[1]

CNRlesion/ROI=μlesion−μROIσROI[2]

where μROI and σROI represent the mean intensity and the standard deviation (SD) from the signal in the ROI, respectively.

Brain metastasis quantification

The diagnosis of brain metastases was determined as described previously (22) and independently confirmed by the two radiologists mentioned above. On baseline MRI, the radiologist identified and annotated enhancing lesions and evaluated their presence, absence, and total number for each imaging sequence. Lesions were considered eligible for analysis if they were: (I) contrast-enhancing structures clearly distinguishable from imaging artifacts or normal enhancing anatomical structures; and (II) contrast-enhancing lesions visible on at least one T₁W imaging sequence. In the current study, small brain metastases were defined as metastases with a diameter of less than 5 mm.

All patients with suspected brain metastases underwent at least two serial follow-up MRI examinations after chemotherapy. The first follow-up examination occurred at around one month after the treatment, and subsequent scans occurred with a minimal interval of three months. Suspected lesions were classified as metastatic or non-metastatic according to the Response Assessment in Neuro-Oncology Brain Metastases (RANO-BM) criteria (24). Specifically, a lesion was labeled as metastatic if it demonstrated a ≥30% reduction in size following treatment, or showed progression within 6 months, defined as a ≥20% increase in size or the appearance of new lesions. Lesions that remained stable without progression during follow-up were considered non-metastatic. For equivocal or indeterminate lesions, the final classification was determined by consensus between two experienced neuroradiologists.

A false-positive lesion was defined as a contrast-enhancing focus detected on a given imaging sequence that was initially interpreted as a metastatic lesion but was subsequently determined to represent a vascular cross-section rather than a true metastasis based on multiplanar reconstruction and longitudinal follow-up. These lesions showed no interval growth or progression on follow-up MRI examinations. In the current study, all false positives were vascular cross-sections. A false-negative lesion was defined as a lesion that was initially interpreted as a vascular structure (e.g., vascular cross-section) or missed on a given imaging sequence at baseline, but that subsequently demonstrated interval enlargement or met the RANO-BM progression or post-treatment size reduction criteria on follow-up imaging, thereby confirming its metastatic nature. For detection efficiency analyses of small metastases, the number of false positives and false negatives was derived separately from the two radiologists and used to calculate sensitivity, the average false-negative rate, the average false-positive rate, and the mean discrepancy (the combined count of false positives and false negatives) (25). It should be noted that in the current study, a clear definition of true negative was not applicable due to the lack of pathological confirmation, which was neither feasible nor clinically justifiable. Therefore, the number of false positives per patient was reported rather than the conventional false-positive rate.

Interobserver agreement for the number of metastatic lesions was assessed using intraclass correlation coefficients (ICCs) between the two radiologists.

Statistical analysis

The statistical analyses were performed using R version 4.4.1. ICC estimates between the two reviewers were calculated based on absolute agreement using a two-way random model. ICC values <0.5, ≥0.5 and <0.75, ≥0.75 and <0.9, and ≥0.90 are indicative of poor, moderate, good, and excellent reliability, respectively (26).

The subjective evaluations of image quality and lesion quantification by the two readers were averaged to yield a single value for each imaging type. Mean values and SDs were calculated for normally distributed continuous data, and median values and interquartile ranges (IQRs) for non-normally distributed data. Differences in lesion quantification, qualitative image quality, and diagnostic confidence among the three imaging types were evaluated using the non-parametric Friedman test. Differences in the quantitative image quality indexes of distinct anatomical structures across the different imaging types were assessed using repeated measures analysis of variance (ANOVA). Log transformation was applied when the dependent variable did not conform to a normal distribution, as assessed by the Shapiro-Wilk method. Post-hoc comparisons were performed using either pairwise Wilcoxon signed-rank tests or t-tests as appropriate.

The P values were adjusted for multiple comparison correction using the Bonferroni method. All P values were two-tailed, with a statistical significance level of P<0.05. Effect sizes were reported as Cohen’s d for parametric post-hoc pairwise comparisons, and as rank-biserial correlation (r) for non-parametric results.

Results

Patient characteristics and brain metastasis detection

In total, 50 patients who met the inclusion criteria were initially identified for inclusion in the study. Two patients were excluded for excessive head motion and one patient for incomplete sequences. Consequently, 47 patients (mean age, 55±7 years; 22 males) were included in the subsequent analyses.

Twenty patients were diagnosed with brain metastases. The inter-reader agreement for detecting metastatic lesions was excellent across all three imaging types (ICCs >0.9), with MULTIPLEX-cT₁W imaging showing the highest evaluation reliability, followed by MULTIPLEX-aT₁W and MPRAGE-T₁W imaging (Table 1). In terms of the metastases with a diameter larger than 5 mm, the number of detected lesions per patient was consistent across the three imaging types (Figure 1A and Table 1). The maximum diameters of the detected lesions did not differ among the three imaging types (χ²_[2]=0.235, P=0.889; Table 1). However, in terms of small metastases, a significant effect of imaging type was observed (χ²_[2]=19.5, P<0.001), with post-hoc comparisons demonstrating that compared with MPRAGE-T₁W imaging, a greater number of small metastases was detected by MULTIPLEX-derived cT₁W (P=0.002, r=0.83) and aT₁W imaging (P=0.002, r=0.83; Figure 1A and Table 1). Meanwhile, the proportion of small metastases to total metastases detected also differed significantly among the three imaging types (χ²_[2]=15.7, P<0.001), with post-hoc comparisons similarly indicating that a greater proportion of small metastases was detected from MULTIPLEX-derived cT₁W (P=0.007, r=0.75) and aT₁W imaging (P=0.01, r=0.63; Figure 1B and Table 1) compared with MPRAGE-T₁W imaging.

Figure 1 Brain metastasis detection across imaging types. Boxplots illustrating the number of small metastases (A), and the proportion of small to total metastases (B) detected across MPRAGE-T₁W, MULTIPLEX-cT₁W, and MULTIPLEX-aT₁W imaging. Stacked bar plot illustrating the true positives, false negatives, and false positives detected for small metastases per imaging type (C). Each circle represents an individual measurement. Boxplots indicate the interquartile range, and the colored diamonds denote the mean values. All the P values in the paired Wilcoxon rank test were adjusted for multiple comparisons with Bonferroni correction. *, P<0.05; **, P<0.01; ***, P<0.001. MPRAGE-T₁W, magnetization prepared rapid gradient-echo T₁-weighted; MULTIPLEX-aT₁W, MULTI-parametric magnetic resonance imaging with fLEXible design derived augmented T₁-weighted; MULTIPLEX-cT₁W, MULTI-parametric magnetic resonance imaging with fLEXible design derived composite T₁-weighted.

The sensitivity, number of true positives, number of false negatives, number of false positives, and number of discrepant lesions for small metastases summed per imaging type by the two readers, as well as averaged at the patient level, are summarized in Table 2 and Figure 1C. Examples of false-positive and false-negative lesions are illustrated in Figure 2. At the patient level, fewer false positives were detected with baseline MULTIPLEX-cT₁W and MULTIPLEX-aT₁W imaging than with MPRAGE-T₁W imaging (Table 2). Overall, the MULTIPLEX-cT₁W and MULTIPLEX-aT₁W imaging demonstrated superior sensitivity in small lesion detection than MPRAGE-T₁W imaging (94.95% and 84.15% respectively vs. 60.65%).

Figure 2 Examples of false-negative (upper plane) and false-positive (lower plane) lesions in a 47-year-old female patient. In the upper plane, the MPRAGE-T₁W images did not show a tiny lesion (dashed circle), which was visible on the MULTIPLEX-derived images (yellow arrows). The follow-up MRI confirmed the metastatic nature of the lesion through post-treatment size reduction. In the lower plane, the MPRAGE-T₁W imaging appeared to show an intra-axial lesion (yellow arrow), which was revealed to be a vessel in multiplanar reconstruction (yellow arrow). The MULTIPLEX-derived images did not show this apparent lesion due to their intrinsic vessel suppression. MPRAGE-T₁W, magnetization prepared rapid gradient-echo T₁-weighted; MRI, magnetic resonance imaging; MULTIPLEX-aT₁W, MULTI-parametric MRI with fLEXible design derived augmented T₁-weighted; MULTIPLEX-cT₁W, MULTI-parametric MRI with fLEXible design derived composite T₁-weighted.

A representative case of a 47-year-old female patient with multiple brain metastases is illustrated in Figure 3. Compared with MPRAGE-T₁W imaging, MULTIPLEX-cT₁W imaging demonstrated superior image quality, enabling clearer visualization of the internal structure of the lesions (Figure 3A). Similarly, MULTIPLEX-aT₁W imaging exhibited higher contrast between the lesions and surrounding tissues, improving boundary delineation and identification. Meanwhile, severe vascular shadowing was observed in MPRAGE-T₁W imaging, interfering with the detection of the lesions. Conversely, vessel signals near the ventricle were effectively suppressed on both MULTIPLEX-derived cT₁W and aT₁W imaging, allowing clear visualization of small metastases (Figure 3B, red arrow). As shown in Figure 3C, small metastasis and leptomeningeal metastasis were not identified on MPRAGE-T₁W imaging but were clearly observable on both MULTIPLEX-cT₁W and -aT₁W imaging. A similar missed detection of small metastases on MPRAGE-T₁W imaging was also demonstrated in the case of a 54-year-old female with multiple brain metastases (Figure 4). Together, the combination of MULTIPLEX-derived cT₁W and aT₁W imaging offered superior detection of small brain metastases.

Figure 3 Images of a 47-year-old female patient with multiple enhancing brain metastases obtained using MPRAGE-T₁W imaging (green arrows; left column), MULTIPLEX-cT₁W imaging (orange arrows; middle column), and MULTIPLEX-aT₁W imaging (blue arrows; right column), presented across three representative slices. The MULTIPLEX-cT₁W images demonstrated better visualization of the lesion’s internal structure (A, middle column), while the MULTIPLEX-aT₁W images demonstrated enhanced lesion boundary definition (A, right column). Conversely, lesion identification on MPRAGE-T₁W images was hindered by insufficient lesion contrast (A,B, left column), and adjacent vascular interference (red arrows; B, left column). A small metastatic lesion and leptomeningeal metastasis were missed on the MPRAGE-T₁W image (C, left column), but were clearly depicted on the MULTIPLEX-cT₁W and -aT₁W images. MPRAGE-T₁W, magnetization prepared rapid gradient-echo T₁-weighted; MULTIPLEX-aT₁W, MULTI-parametric magnetic resonance imaging with fLEXible design derived augmented T₁-weighted; MULTIPLEX-cT₁W, MULTI-parametric magnetic resonance imaging with fLEXible design derived composite T₁-weighted.

Figure 4 Missed detection of small brain metastases on MPRAGE-T₁W imaging in a 54-year-old woman with multiple metastases. Conversely, MULTIPLEX-cT₁W (orange arrows), and -aT₁W (blue arrow) imaging clearly depicted small metastases. MPRAGE-T₁W, magnetization prepared rapid gradient-echo T₁-weighted; MULTIPLEX-aT₁W, MULTI-parametric magnetic resonance imaging with fLEXible design derived augmented T₁-weighted; MULTIPLEX-cT₁W, MULTI-parametric magnetic resonance imaging with fLEXible design derived composite T₁-weighted.

Quantitative image quality

Quantitative image quality metrics (i.e., the SNR and CNR) did not follow a normal distribution (W=0.83, P<0.001 and W=0.77, P<0.001, respectively), and were therefore log transformed prior to analyses.

A two-way ANOVA was first conducted to assess the SNR differences in distinct normal anatomical structures (GM and WM) across the three imaging types (MPRAGE-T₁W, MULTIPLEX-cT₁W, and MULTIPLEX-aT₁W). As expected, a significant main effect of imaging type was observed (F_{[2, 92]}=88.99, P<0.001), with the post-hoc pairwise comparisons revealing that MULTIPLEX-cT₁W achieved the highest SNR, outperforming both MULTIPLEX-aT₁W {t_[93]=13.10, P<0.001, family-wise error (FWE), Cohen’d =1.35} and MPRAGE-T₁W imaging (t_[93]=5.80, P<0.001, FWE, Cohen’d =0.60). Meanwhile, a significant interaction effect between anatomical structure and imaging type was observed (F_{[2, 92]}=3.80, P=0.026), with further post-hoc analyses indicating that the SNR improvement of MULTIPLEX-cT₁W imaging relative to MPRAGE-T₁W imaging was primarily dominant in GM (t_[46]=7.03, P<0.001, FWE, Cohen’d =1.02), rather than in WM (t_[46]=2.27, P=0.084, FWE, Cohen’d =0.33; Figure 5A).

Figure 5 Quantitative measurements of image quality across imaging types and anatomical structures. The log-transformed SNR of normal brain structures (A) of WM and GM, brain metastases (B), and the CNR values of lesion-to-WM and lesion-to-GM (C) were illustrated and compared across MPRAGE-T₁W, MULTIPLEX-cT₁W, and MULTIPLEX-aT₁W imaging. Each circle represents an individual measurement. Boxplots indicate the interquartile range, and the colored diamonds denote the mean values. The colored shade indicates the data distribution. All the P values in the paired t-test were adjusted for multiple comparisons with Bonferroni correction. *, P<0.05; ***, P<0.001; ns, non-significant. CNR, contrast-to-noise ratio; GM, grey matter; MPRAGE-T₁W, magnetization prepared rapid gradient-echo T₁-weighted; MULTIPLEX-aT₁W, MULTI-parametric magnetic resonance imaging with fLEXible design derived augmented T₁-weighted; MULTIPLEX-cT₁W, MULTI-parametric magnetic resonance imaging with fLEXible design derived composite T₁-weighted; SNR, signal-to-noise ratio; WM, white matter.

Subsequently, a one-way ANOVA was conducted to assess SNR differences in brain metastases among the same three imaging types. A significant main effect of imaging type was again observed (F_{[2, 28]}=3.45, P=0.046), with post-hoc pairwise comparisons indicating that MULTIPLEX-cT₁W imaging exhibited a higher SNR than MPRAGE-T₁W imaging (t_[14]=2.48, P=0.027, uncorrected, Cohen’d =0.64; Figure 5B). We further restricted the above analyses to metastatic lesions with a diameter less than 5 mm and found similar results: small brain metastases from MULTIPLEX-cT₁W imaging demonstrated a significantly higher SNR than MPRAGE-T₁W imaging (Table S2; Figure S1A).

Similarly, a two-way ANOVA was conducted to assess CNR differences in brain metastases relative to distinct anatomical structures (lesion-to-GM and lesion-to-WM) across the three imaging types (MPRAGE-T₁W, MULTIPLEX-cT₁W, and MULTIPLEX-aT₁W). A significant main effect of structure type was observed (F_{[1, 14]}=30.23, P<0.001), with post-hoc pairwise comparisons indicating an overall better lesion-to-GM differentiation compared to lesion-to-WM differentiation (t_[44]=5.10, P<0.001, Cohen’d =0.76). A significant main effect of imaging type was also found (F_{[2, 28]}=9.64, P<0.001), with post-hoc comparisons revealing that MULTIPLEX-aT₁W imaging achieved the highest CNR, outperforming both MPRAGE-T₁W (t_[29]=5.38, P<0.001, FWE, Cohen’d =0.98), and MULTIPLEX-cT₁W imaging (t_[29]=2.70, P=0.03, FWE, Cohen’d =0.49). Further, a significant interaction effect between structure and imaging type (F_{[2, 28]}=8.26, P=0.006) was observed, with post-hoc comparisons suggesting that the CNR gain in MULTIPLEX-aT₁W imaging relative to MPRAGE-T₁W imaging was mainly driven by the improvement in lesion-to-GM contrast (t_[29]=5.87, P<0.001, FWE, Cohen’d =1.52), while the lesion-to-WM contrast difference did not reach statistical significance (t_[14]=2.42, P=0.088, FWE, Cohen’d =0.63; Figure 5C). Again, we examined the CNR differences in small lesions across the different imaging types and found similar results: the CNR improvement in MULTIPLEX-aT₁W imaging compared to MPRAGE-T₁W imaging was the most prominent in lesion-to-GM contrast rather than lesion-to-WM contrast (Table S2; Figure S1B).

Together, these results indicated an overall image quality improvement of MULTIPLEX-derived cT₁W and aT₁W imaging compared with MPRAGE-T₁W imaging, in the SNR and CNR, respectively.

Qualitative image quality

Subjective image quality scores did not follow a normal distribution (W=0.23, P<0.001 and W=0.75, P<0.001 for visual image quality and diagnostic confidence, respectively) and were compared across the three imaging types. All three imaging types met the standard image quality for diagnosis, and no significant difference was found in the subjective scoring (χ²_[2]=3.00, P=0.22; Figure 6A). However, diagnostic confidence differed significantly among imaging types (χ²_[2]=6.95, P=0.031), with post-hoc pairwise comparisons revealing that MULTIPLEX-cT₁W imaging provided higher diagnostic confidence relative to MPRAGE-T₁W imaging (P=0.043, FWE, r=0.39; Figure 6B). These results suggested that despite comparable visual image quality, MULTIPLEX-derived imaging outperformed MPRAGE-T₁W imaging in diagnostic confidence in detecting brain metastases.

Figure 6 Qualitative measurements of image quality and diagnostic efficiency. Subjective image quality scores (A) and diagnostic confidence (B) were illustrated and compared across MPRAGE-T₁W, MULTIPLEX-cT₁W, and MULTIPLEX-aT₁W imaging. Each colored circle represents an individual score, while the black diamonds indicate the median score for each imaging type. All the P values in the paired Wilcoxon rank test were adjusted for multiple comparisons with Bonferroni correction. *, P<0.05. MPRAGE-T₁W, magnetization prepared rapid gradient-echo T₁-weighted; MULTIPLEX-aT₁W, MULTI-parametric magnetic resonance imaging with fLEXible design derived augmented T₁-weighted; MULTIPLEX-cT₁W, MULTI-parametric magnetic resonance imaging with fLEXible design derived composite T₁-weighted.

Discussion

This study was the first to investigate the application of contrast-enhanced MULTIPLEX-derived cT₁W and aT₁W imaging in brain metastasis detection, in comparison to the gold-standard T₁W imaging. Our results demonstrated that MULTIPLEX-derived imaging provided superior detection of small brain metastases with a diameter of less than 5 mm, compared with MPRAGE-T₁W imaging, as well as higher detection sensitivity. Quantitative image quality analyses revealed that compared with MPRAGE-T₁W imaging, MULTIPLEX-cT₁W imaging exhibited a higher SNR in GM and metastases, while aT₁W imaging showed better lesion-to-GM contrast. Qualitative analyses indicated comparable subjective image quality among the three T₁W imaging, while MULTIPLEX-cT₁W imaging offered the highest diagnostic confidence. Despite the inconsistent native resolution and the possibility that resampling may not completely ensure a fair comparison, these findings demonstrated the superior performance of contrast-enhanced MULTIPLEX-derived imaging in detecting small brain metastases compared with MPRAGE-T₁W imaging.

Although the absolute increase in the median number of small metastases detected per patient was modest, the clinical relevance of this finding should not be underestimated. In a subset of patients, MULTIPLEX-derived cT₁W and aT₁W imaging revealed one or more additional metastases with a diameter of less than 5mm that were not identified on conventional MPRAGE-T₁W imaging. In clinical practice, even a single additional brain metastasis may affect both radiotherapy planning—including target delineation and the choice between stereotactic radiosurgery and more extensive treatment strategies—and prognostic stratification based on lesion burden. Notably, the present study was not designed or powered to evaluate whether improved lesion detection translates into better clinical outcomes. Therefore, the true clinical benefit of identifying additional small metastases remains to be established in prospective studies that incorporate treatment decisions and longitudinal outcome measures. Nevertheless, more accurate lesion detection, counting, and delineation are inherently relevant to treatment planning and risk stratification in everyday neuro-oncologic practice, supporting the potential clinical value of MULTIPLEX-derived imaging.

A recent study comparing non-contrast-enhanced MULTIPLEX imaging with T₁W imaging for the detection of neonatal punctate WM lesions showed the superior performance of MULTIPLEX, including an improved SNR on cT₁W imaging, and improved lesion-to-WM contrast on aT₁W imaging (16). Building on previous findings, our study further demonstrated that contrast-enhanced MULTIPLEX imaging enhances the detection of small metastases with limited visibility on MPRAGE-T₁W imaging. We also demonstrated for the first time that the SNR and CNR improvements in MULTIPLEX-derived imaging were predominantly observed in GM. This finding appears to differ from previous reports that primarily emphasized WM lesions (16). Nevertheless, due to brain tissue immaturity and higher water content, neonatal WM presents lower intensity than GM on T₁W imaging (27). Conversely, the punctate WM lesions exhibit relatively higher intensity than the surrounding WM (28), a pattern opposite to that observed on adult T₁W imaging (1,5). Therefore, rather than contradicting previous findings in neonatal imaging, our results extend these observations to adult brain metastases, highlighting the broader applicability of MULTIPLEX imaging for lesion detection across developmental stages.

Other techniques optimizing T1W images, such as high-resolution three-dimensional (3D) volumetric interpolated breath-hold examination (VIBE) combined with maximum intensity projection (MIP) reconstructions, have been proposed to enhance the visualization of small brain metastases. However, recent research has shown that MIP reconstructions do not provide a significant increase in sensitivity compared with native 3D VIBE images in metastasis detection, suggesting that projection-based post-processing alone may be insufficient. Conversely, MULTIPLEX-derived cT₁W and aT₁W imaging improves lesion conspicuity through contrast-mechanism-driven optimization and background suppression rather than post-hoc signal projection. These results are consistent with other contrast-based optimization approaches (20,23), and may explain their ability to reveal additional small lesions that remain inconspicuous on conventional MPRAGE-T₁W images.

Traditionally, intracranial metastases are thought to arise at the GM-WM junction due to the vascular branching and tapering of cortical arterioles as they enter the subcortical WM. However, accumulating evidence from high-resolution MRI studies indicates that small brain metastases, particularly melanoma metastases, often localize to the cortico-meningeal interface, suggesting an early leptomeningeal or superficial cortical origin. Lasocki et al. demonstrated that most intracranial melanoma metastases with a diameter between 2 and 9 mm presented as cortico-meningeal nodules, with secondary extension into the deeper brain parenchyma (29). Therefore, the precise detection of small lesions near cortical GM is essential for the early diagnosis and timely treatment of brain metastases, which further raises concerns about the relatively poor efficiency of current contrast-enhanced T₁W imaging in detecting small metastases.

In this anatomical context, the lesion-to-GM CNR is clinically more relevant than the lesion-to-WM CNR in the detection of small metastases. Small cortico-meningeal or superficial cortical lesions are typically embedded within or adjacent to cortical GM, where background signal intensity and contrast uptake differ substantially from that of WM. Improved lesion-to-GM CNR therefore directly enhances lesion conspicuity at the cortical surface, where partial-volume effects, vascular enhancement, and meningeal signal can otherwise obscure small metastases. Given its superior detection performance of small brain metastases, MULTIPLEX imaging may serve as an advanced scanning sequence compared with current T₁W imaging for brain metastases.

Despite its improved SNR and CNR, the superior diagnostic performance of MULTIPLEX-derived imaging compared with MPRAGE-T₁W imaging may also be attributable to the suppression of vascular signals, which are relatively enhanced on standard T₁W imaging and can thus interfere with the diagnosis of small metastases (20,23,30-32). Due to its dual-TR, dual-FA, and multi-echo compositing strategy, MULTIPLEX-derived images diminish the inflow-related vascular hyperintensity typically seen on contrast-enhanced T₁W images, while strengthening the signal in small vessels. Additionally, the aT₁W contrast, calculated using T₁W-to-PDW signal ratios, intrinsically attenuates non-T₁-dependent signals, further reducing vascular signal contamination and enhancing the conspicuity of small metastases (13). Therefore, these technical advantages likely contribute to the enhanced lesion conspicuity and diagnostic confidence observed in MULTIPLEX-derived imaging, particularly in the detection of small metastases. However, due to its multi-echo and synthetic nature, MULTIPLEX is more prone to motion artifacts than standard T₁W imaging. Thus, careful motion management and monitoring is warranted when applying this technique clinically.

Despite the overall improvements in small metastasis detection and image quality, the acquisition times of MULTIPLEX were considerably longer than those for MPRAGE-T₁W imaging. Nevertheless, we believe that MULTIPLEX remains a valuable tool for clinical applications. The MPRAGE-T₁W sequence uses the advanced ACS acceleration algorithm (33), which has been reported to reduce the acquisition time by 29.2% (34) to over 50% (35), while maintaining or even enhancing image quality and diagnostic confidence. Standard whole brain T₁W imaging with partially parallel acceleration takes approximately 5 minutes (20,23,31). Therefore, the acquisition time of MULTIPLEX remains comparable to that of T₁W imaging in major clinical workflows. Additionally, MULTIPLEX inherently generates a series of qualitative images and parametric maps beyond cT₁W and aT₁W imaging, which may provide additional diagnostic and prognostic value, but these were not assessed in the current study. For example, during treatment follow-up, transient lesion enlargement may occur as part of treatment response. Relying solely on lesion size comparisons on structural imaging may lead to inaccurate assessments of therapeutic efficacy. MULTIPLEX-derived quantitative susceptibility mapping enables the detection of intralesional calcification or hemorrhage, adding to the differentiation between treatment-induced regression and tumor progression (Figure S2). Thus, the broad capabilities of MULTIPLEX may enhance prognostic stratification and inform treatment decisions, optimizing patient care. Nevertheless, future research should seek to develop advanced acceleration algorithms for MULTIPLEX. Additionally, comprehensive studies are warranted to evaluate the full range of imaging derived from MULTIPLEX in various neuro-oncologic scenarios.

An additional factor that warrants consideration is the timing of post-contrast image acquisition. In the present study, all contrast-enhanced images were acquired during the early post-contrast phase, reflecting routine clinical practice and enabling a controlled comparison between imaging sequences. However, delayed post-contrast imaging has been shown to improve lesion conspicuity by allowing greater contrast agent accumulation and retention in metastatic tissue, particularly for small lesions (36). Future studies should explore the combined effect of MULTIPLEX-derived T1W contrasts and optimized post-contrast acquisition timing (37), potentially incorporating elements of late gadolinium enhancement paradigms to further enhance detection sensitivity or lesion characterization.

This study had several limitations. First, the sample size was relatively small, and the study was conducted at a single center, which may have led to the overestimation of effects and the undersampling of lesion heterogeneity. Future studies with larger cohorts and multicenter data are needed for further validation. Second, all the included patients had lung cancer as the primary tumor; however, metastases from melanoma, breast, and other origins may display different enhancement patterns and distributions. Thus, the generalizability of the results to brain metastases from other malignancies remains limited. Given that the intracranial distribution patterns of brain metastases vary by primary tumor type, potentially influencing lesion detection and localization, further research should include patients with various primary cancers, such as breast cancer or melanoma, to assess the broader applicability of the MULTIPLEX technique. Third, though image resampling was applied for fair comparison, the acquisition resolution was not consistent between the MULTIPLEX and MPRAGE-T₁W sequences. It should be acknowledged that any advantages of MULTIPLEX might only partly reflect its contrast or sequence design. Conversely, the MPRAGE-T₁W might benefit from its isotropic resolution, but this advantage was not used in the current study. Future studies would benefit from more closely aligned acquisition to ensure methodological rigor. Finally, in the present study, brain metastases were confirmed based on follow-up MRI rather than histopathologic verification for all lesions. This approach may introduce a certain degree of lesion misclassification, including potential false-positive or false-negative findings. However, pathological confirmation of every suspected brain metastasis is neither feasible nor ethically justifiable in routine clinical practice, particularly for small or asymptomatic lesions. Accordingly, serial imaging follow-up is widely accepted as a pragmatic reference standard in brain metastasis imaging studies and has been extensively adopted in previous neuro-oncologic research (24). By applying standardized response and progression criteria and requiring longitudinal confirmation, our approach balances methodological rigor with clinical feasibility, thereby maintaining real-world relevance while minimizing classification bias.

Conclusions

Our study demonstrated that contrast-enhanced MULTIPLEX-derived cT₁W and aT₁W imaging offer superior diagnostic efficacy compared to MPRAGE-T₁W imaging, particularly in the detection of small brain metastases. The enhanced SNR and CNR, combined with the partial vascular signal suppression, contribute to improved lesion conspicuity and diagnostic confidence. These findings show the clinical utility of MULTIPLEX as a promising tool for comprehensive brain metastasis assessment. Further validation in larger, multicenter studies and across diverse primary tumor types is warranted to confirm the generalizability of the current results.

Acknowledgments

We would like to thank the patients who took part.

Footnote

Reporting Checklist: The authors have completed the STARD reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-aw-2474/rc

Data Sharing Statement: Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-aw-2474/dss

Funding: This work was supported by the National Science and Technology Innovation 2030—Major Program of “Brain Science and Brain-Like Research” (No. 2022ZD0211800) to B.Z.; Jiangsu Provincial Health Commission General Science Program (No. M2024057) to X.Z.; the Key Project Funded by the Medical Science and Technology Development Foundation of the Nanjing Department of Health (No. ZKX21031), Medical Imaging Artificial Intelligence Special Research Fund of the Nanjing Medical Association Radiology Branch (Key Project 17), and fundings for clinical trials from the Affiliated Drum Tower Hospital, Medical School of Nanjing University (No. 2021-LCYJ-PY-36) to Z.W.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-aw-2474/coif). Y.Z. was the employee of United Imaging Healthcare and received salary from United Imaging Healthcare within the past 36 months. S.D. and Y.W. are employees of United Imaging Healthcare and receiving salary from United Imaging Healthcare. 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 prospective study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of Nanjing Drum Tower Hospital (approval No. 2020-379-01, AF/SC-08/03.0) and 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. Achrol AS, Rennert RC, Anders C, Soffietti R, Ahluwalia MS, Nayak L, Peters S, Arvold ND, Harsh GR, Steeg PS, Chang SD. Brain metastases. Nat Rev Dis Primers 2019;5:5. [Crossref] [PubMed]
2. DiLuna ML, King JT Jr, Knisely JP, Chiang VL. Prognostic factors for survival after stereotactic radiosurgery vary with the number of cerebral metastases. Cancer 2007;109:135-45. [Crossref] [PubMed]
3. Seute T, Leffers P, ten Velde GP, Twijnstra A. Detection of brain metastases from small cell lung cancer: consequences of changing imaging techniques (CT versus MRI). Cancer 2008;112:1827-34. [Crossref] [PubMed]
4. Brindle KM, Izquierdo-García JL, Lewis DY, Mair RJ, Wright AJ. Brain Tumor Imaging. J Clin Oncol 2017;35:2432-8. [Crossref] [PubMed]
5. Derks SHAE, van der Veldt AAM, Smits M. Brain metastases: the role of clinical imaging. Br J Radiol 2022;95:20210944. [Crossref] [PubMed]
6. Kaufmann TJ, Smits M, Boxerman J, Huang R, Barboriak DP, Weller M, Chung C, Tsien C, Brown PD, Shankar L, Galanis E, Gerstner E, van den Bent MJ, Burns TC, Parney IF, Dunn G, Brastianos PK, Lin NU, Wen PY, Ellingson BM. Consensus recommendations for a standardized brain tumor imaging protocol for clinical trials in brain metastases. Neuro Oncol 2020;22:757-72. [Crossref] [PubMed]
7. Langen KJ, Galldiks N, Hattingen E, Shah NJ. Advances in neuro-oncology imaging. Nat Rev Neurol 2017;13:279-89. [Crossref] [PubMed]
8. Sze G, Milano E, Johnson C, Heier L. Detection of brain metastases: comparison of contrast-enhanced MR with unenhanced MR and enhanced CT. AJNR Am J Neuroradiol 1990;11:785-91.
9. Kaal EC, Taphoorn MJ, Vecht CJ. Symptomatic management and imaging of brain metastases. J Neurooncol 2005;75:15-20. [Crossref] [PubMed]
10. Mehrabian H, Detsky J, Soliman H, Sahgal A, Stanisz GJ. Advanced Magnetic Resonance Imaging Techniques in Management of Brain Metastases. Front Oncol 2019;9:440. [Crossref] [PubMed]
11. Lignelli A, Khandji AG. Review of imaging techniques in the diagnosis and management of brain metastases. Neurosurg Clin N Am 2011;22:15-25. v. [Crossref] [PubMed]
12. Pope WB. Brain metastases: neuroimaging. Handb Clin Neurol 2018;149:89-112. [Crossref] [PubMed]
13. Ye Y, Lyu J, Hu Y, Zhang Z, Xu J, Zhang W. MULTI-parametric MR imaging with fLEXible design (MULTIPLEX). Magn Reson Med 2022;87:658-73. [Crossref] [PubMed]
14. Denk C, Rauscher A. Susceptibility weighted imaging with multiple echoes. J Magn Reson Imaging 2010;31:185-91. [Crossref] [PubMed]
15. Ye Y, Lyu J, Hu Y, Zhang Z, Xu J, Zhang W, Yuan J, Zhou C, Fan W, Zhang X. Augmented T(1) -weighted steady state magnetic resonance imaging. NMR Biomed 2022;35:e4729. [Crossref] [PubMed]
16. Ji L, Kong L, Zhang X, Chen X, Lu C, Wu F, Tang R, Zhao M. Application of T1-weighted and augmented T1-weighted images of multi-parametric MR sequence in detecting neonatal punctate white matter lesions. Magn Reson Imaging 2025;117:110317. [Crossref] [PubMed]
17. Li W, Sha L, Zhu J, Long F, Chen L. Prediction of epileptogenicity in patients with tuberous sclerosis complex using multimodal cerebral MRI. Eur J Radiol 2024;181:111800. [Crossref] [PubMed]
18. Duan M, Pan R, Gao Q, Wu X, Lin H, Yuan J, Zhang Y, Liu L, Tian Y, Fu T. A rapid multi-parametric quantitative MR imaging method to assess Parkinson's disease: a feasibility study. BMC Med Imaging 2024;24:58. [Crossref] [PubMed]
19. Zhang S, Zhong J, Huang X, Peng R. The application value of MTP synthetic sequence in the diagnosis of acute ischemic stroke. Chinese Journal of Magnetic Resonance Imaging 2022;13:107-11.
20. Kammer NN, Coppenrath E, Treitl KM, Kooijman H, Dietrich O, Saam T. Comparison of contrast-enhanced modified T1-weighted 3D TSE black-blood and 3D MP-RAGE sequences for the detection of cerebral metastases and brain tumours. Eur Radiol 2016;26:1818-25. [Crossref] [PubMed]
21. Rowley HA, Scialfa G, Gao PY, Maldjian JA, Hassell D, Kuhn MJ, Wippold FJ 2nd, Gallucci M, Bowen BC, Schmalfuss IM, Ruscalleda J, Bastianello S, Colosimo C. Contrast-enhanced MR imaging of brain lesions: a large-scale intraindividual crossover comparison of gadobenate dimeglumine versus gadodiamide. AJNR Am J Neuroradiol 2008;29:1684-91. [Crossref] [PubMed]
22. Kim D, Heo YJ, Jeong HW, Baek JW, Han JY, Lee JY, Jin SC, Baek HJ. Usefulness of the Delay Alternating with Nutation for Tailored Excitation Pulse with T1-Weighted Sampling Perfection with Application-Optimized Contrasts Using Different Flip Angle Evolution in the Detection of Cerebral Metastases: Comparison with MPRAGE Imaging. AJNR Am J Neuroradiol 2019;40:1469-75. [Crossref] [PubMed]
23. Fu Q, Cheng QG, Kong XC, Liu DX, Guo YH, Grinstead J, Zhang XY, Lei ZQ, Zheng CS. Comparison of contrast-enhanced T1-weighted imaging using DANTE-SPACE, PETRA, and MPRAGE: a clinical evaluation of brain tumors at 3 Tesla. Quant Imaging Med Surg 2022;12:592-607. [Crossref] [PubMed]
24. Lin NU, Lee EQ, Aoyama H, Barani IJ, Barboriak DP, Baumert BG, et al. Response assessment criteria for brain metastases: proposal from the RANO group. Lancet Oncol 2015;16:e270-8. [Crossref] [PubMed]
25. Parillo M, Vertulli D, Vaccarino F, Mallio CA, Beomonte Zobel B, Quattrocchi CC. The sensitivity of MIPs of 3D contrast-enhanced VIBE T1-weighted imaging for the detection of small brain metastases (≤ 5 mm) on 1.5 tesla MRI. Neuroradiol J 2024;37:744-750. [Crossref] [PubMed]
26. Koo TK, Li MY. A Guideline of Selecting and Reporting Intraclass Correlation Coefficients for Reliability Research. J Chiropr Med 2016;15:155-63. [Crossref] [PubMed]
27. Dubois J, Dehaene-Lambertz G, Kulikova S, Poupon C, Hüppi PS, Hertz-Pannier L. The early development of brain white matter: a review of imaging studies in fetuses, newborns and infants. Neuroscience 2014;276:48-71. [Crossref] [PubMed]
28. Ramenghi LA, Fumagalli M, Righini A, Bassi L, Groppo M, Parazzini C, Bianchini E, Triulzi F, Mosca F. Magnetic resonance imaging assessment of brain maturation in preterm neonates with punctate white matter lesions. Neuroradiology 2007;49:161-7. [Crossref] [PubMed]
29. Lasocki A, Khoo C, Lau PKH, Kok DL, Mcarthur GA. High-resolution MRI demonstrates that more than 90% of small intracranial melanoma metastases develop in close relationship to the leptomeninges. Neuro Oncol 2020;22:423-32. [Crossref] [PubMed]
30. Jeevanandham B, Kalyanpur T, Gupta P, Cherian M. Comparison of post-contrast 3D-T(1)-MPRAGE, 3D-T(1)-SPACE and 3D-T(2)-FLAIR MR images in evaluation of meningeal abnormalities at 3-T MRI. Br J Radiol 2017;90:20160834. [Crossref] [PubMed]
31. Edelman RR, Walker M, Ankenbrandt WJ, Leloudas N, Pang J, Bailes J, Bobustuc G, Koktzoglou I. Improved Brain Tumor Conspicuity at 3 T Using Dark Blood, Fat-Suppressed, Dixon Unbalanced T1 Relaxation-Enhanced Steady-State MRI. Invest Radiol 2023;58:641-8. [Crossref] [PubMed]
32. Baek TW, Kang Y, Lee HJ. Improved Lesion Conspicuity with Contrast-Enhanced 3D T1 TSE Black-Blood Imaging in Cranial Neuritis: A Comparative Study of Contrast-Enhanced 3D T1 TSE, 3D T1 Fast-Spoiled Gradient Echo, and 3D T2 FLAIR. AJNR Am J Neuroradiol 2021;42:945-50. [Crossref] [PubMed]
33. Wang S, Cao G, Wang Y, Liao S, Wang Q, Shi J, Li C, Shen D. Review and Prospect: Artificial Intelligence in Advanced Medical Imaging. Front Radiol 2021;1:781868. [Crossref] [PubMed]
34. Gu W, Yang C, Wang Y, Hu W, Wu D, Cai S, Hong G, Hu P, Zhang Q, Dai Y. AI-Assisted Compressed Sensing Enables Faster Brain MRI for the Elderly: Image Quality and Diagnostic Equivalence with Conventional Imaging. Int J Gen Med 2025;18:379-90. [Crossref] [PubMed]
35. Karthik A, Aggarwal K, Kapoor A, Singh D, Hu L, Gandhamal A, Kumar D. Comprehensive assessment of imaging quality of artificial intelligence-assisted compressed sensing-based MR images in routine clinical settings. BMC Med Imaging 2024;24:284. [Crossref] [PubMed]
36. Parillo M, Mallio CA, Dekkers IA, Rovira À, van der Molen AJ, Quattrocchi CCESMRMB-GREC Working Group. Late/delayed gadolinium enhancement in MRI after intravenous administration of extracellular gadolinium-based contrast agents: is it worth waiting? MAGMA 2024;37:151-68. [Crossref] [PubMed]
37. Mallio CA, Messina L, Parillo M, Lo Vullo G, Beomonte Zobel B, Parizel PM, Quattrocchi CC. Anterior pituitary gland T1 signal intensity is influenced by time delay after injection of gadodiamide. Sci Rep 2020;10:14967. [Crossref] [PubMed]