AI-assisted compressed sensing and gadoxetic acid-enhanced MRI for evaluating colorectal liver metastases in complex hepatic backgrounds: a prospective 5T MRI study

Introduction

Colorectal cancer (CRC) is the third most common malignancy worldwide (1-3). According to the latest epidemiological data, the global incidence of CRC was 18.7 per 100,000 individuals in 2022 (2), yet China exhibited an even higher incidence of 20.2 per 100,000 individuals (4). Over 50% of CRC patients will develop metastatic disease; among whom, 70% will experience spread to the liver (5), also referred to as colorectal liver metastasis (CRLM). Moreover, the intrahepatic recurrence rate remains high, with approximately 60% of patients developing recurrent lesions regardless of whether the initial treatment involves resection, resection plus radiofrequency ablation (RFA), or RFA alone (6).

Chemotherapy, including neoadjuvant chemotherapy (NAC) and adjuvant chemotherapy (AC), is a key therapeutic strategy for patients with CRC and CRLMs (7). Specifically, NAC can increase the resectability of initially unresectable CRLMs and improve overall survival (8), whereas AC can significantly reduce the risk of recurrence following radical resection of metastatic lesions, particularly in patients with liver metastases (9). Although chemotherapy can provide therapeutic benefits for patients with liver metastases, it can also lead to chemotherapy-associated liver injury (CALI) (10), including steatosis, sinusoidal obstruction syndrome (SOS) (11,12), or intracellular accumulation of substances such as hepatic iron, leading to overload (12,13). Moreover, in patients with recurrent CRLM following local treatment (such as surgical resection or ablation), the liver (e.g., the nontumor-bearing hepatic parenchyma) may display ischemic changes (14), hepatic regeneration, or local fibrotic scarring (15).

Both CALIs and pathological hepatic tissue changes following local treatments for CRLMs may obscure metastatic lesions or produce nodular regenerative hyperplasias that can mimic metastases, making accurate imaging-based CRLM detection difficult (10,16-18). Surgical resection is vital for achieving long-term disease-free survival in CRLM patients regardless of initial or postconversion resectability, whereas preoperative CALI assessment is essential for those receiving chemotherapy (19). Therefore, advanced imaging techniques that can precisely localize and characterize CRLMs while assessing the severity of CALI are essential, as they could provide crucial data for optimizing surgical planning.

Multimodal magnetic resonance imaging (MRI) has become an important tool for assessing the CRLM-related liver backgrounds described above (20-23). For example, the proton density fat fraction (PDFF) sequence enables quantitative assessments of hepatic iron overload (HIO) and steatosis through R2* and PDFF values (20,21). Furthermore, heterogeneity in parenchymal signals during the portal venous phase and reticular hypointensities on hepatobiliary phase (HBP) gadoxetic acid-enhanced MRI (EOB-MRI) are useful markers for identifying SOS (22,23). Additionally, the combination of EOB-MRI with diffusion-weighted imaging (DWI) yields greater sensitivity for detecting and characterizing CRLM lesions (24).

Recently, technological advances have led to the development of 5 Tesla (T) MRI, which has been applied with increasing frequency in abdominal imaging (21,25-27), in part because of its excellent signal-to-noise ratio (SNR) and tissue contrast. Additionally, the development of artificial intelligence (AI)-assisted compressed sensing (ACS) technology has further optimized the clinical utility of high-field MRI (28-30). By integrating AI, partial Fourier, parallel imaging, and compressed sensing (CS) techniques, ACS significantly reduces the scan time while achieving 1.0 mm³ isotropic, high-resolution three-dimensional (3D) HBP imaging (31). This technology not only enables high-resolution multiplanar reconstruction but also improves the detection of small lesions and the visualization of the biliary tract (31).

Although multimodal, high-field MRI has demonstrated potential benefits for CRLM imaging, studies in this area remain limited. Therefore, this study investigated the diagnostic efficacy of 5T multimodal MRI with ACS and EOB-MRI for the qualitative and quantitative assessment of CALI and the detection of CRLMs in complex hepatic backgrounds. We present this article in accordance with the STARD reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1425/rc).

Methods

Study design and participants

This prospective, single-center study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study protocol was approved by the Medical Research Ethics Committee of The First Affiliated Hospital of University of Science and Technology of China (No. XJS2023-1-30[YJ]). Written informed consent was provided by all participants. Between May 2023 and June 2024, consecutive patients with suspected CRLMs were enrolled at the First Affiliated Hospital of the University of Science and Technology of China. The inclusion criteria were as follows: pathologically confirmed CRC treated with oxaliplatin-based chemotherapy within 6 months before MRI examination; suspicions of newly developed or recurrent CRLM according to ultrasound and/or CT imaging, and no contraindications to MRI. All patients underwent a 5T multimodal MRI scan on the same machine to evaluate the liver background and determine the presence of CRLMs. The exclusion criteria were as follows: (I) significant MRI motion artifacts (respiratory, cardiac, or patient movement) impairing diagnostic assessment; (II) unclear CRLM, defined by the absence of both histological confirmation and adequate follow-up imaging/clinical data; and (III) insufficient liver biochemistry data, defined as missing liver function tests that include alanine transaminase (ALT) and aspartate aminotransferase (AST). CRLM was confirmed with a combination of methods, including pathology, intraoperative contrast-enhanced ultrasound with surgical exploration, or imaging follow-up (≥6 months) with chemotherapy response evaluation.

Image acquisition and MRI protocol

All participants underwent abdominal multiparametric MRI with a 5.0 T scanner (uMR Jupiter 5T, United Imaging Healthcare, Shanghai, China). The protocol and scanning sequences included the following: coronal single-shot T2-weighted imaging (T2WI); dual-echo T1-weighted (T1W) imaging; axial T1W 3D fast spoiled gradient-recalled echo (FSPGR) dynamic contrast-enhanced imaging; coronal T1W 3D delayed-phase imaging; fat-suppressed axial T2WI; DWI; PDFF mapping; susceptibility-weighted imaging with fast technique (uSWIFT); and 3D isotropic HBP imaging using ACS (ACS-HBP, transverse and coronal orientations, acquisition voxel size 1.2 mm³, reconstructed voxel size 0.6 mm³, 300 slices). For contrast-enhanced imaging, EOB-DTPA (Primovist, Bayer, Leverkusen, Germany) was intravenously administered at 0.1 mL/kg and 1 mL/s, followed by a 15 mL saline flush at the same rate. 3D T1W images were acquired during the arterial phase (20–30 s), portal-venous phase (70 s), equilibrium phase (180 s), and HBP (20 min) after injection. Apparent diffusion coefficient (ADC) maps were generated using b values of 50 and 800 s/mm². The complete MRI protocol parameters are detailed in Table S1 in the Appendix 1.

Image analysis

Two radiologists with more than 10 years of experience independently reviewed all MR images. Although they were aware that the purpose was to evaluate the images for CRLMs, they were blinded to the other clinical information of the patients. The images were analyzed in random order with a 4-week interval between sessions using a picture archiving and communication system (PACS) workstation.

• Both radiologists performed qualitative and quantitative assessments of background liver characteristics—namely CALI—on multimodal MRI (detailed methodologies for evaluating the images for fatty liver, HIO, and SOS are described in Appendix 1).
• For the CRLM evaluations, each radiologist independently analyzed the images from three MRI datasets, all of which included precontrast T1- and T2-weighted sequences, in a randomized order: (I) DWI: DWI acquired with b values of 50 and 800 s/mm² along with the corresponding ADC maps; (II) HBP imaging: arterial, portal venous, equilibrium, and HBP images; and (III) combined images: HBP imaging plus DWI. The HBP CRLM assessments were performed as follows: (I) regions of interest (ROIs) were placed on the area of the lesion demonstrating the most hyperintense signal and on the adjacent liver parenchyma to calculate the tumor-to-parenchyma signal intensity ratio (RatioT/P), classified as (i) intense (RatioT/P ≥1); (ii) residual (0.5≤ RatioT/P <1); or (iii) deficient (RatioT/P < 0.5); (II) the HBP signal patterns of the CRLMs were evaluated and classified into three types: (i) homogeneous defects; (ii) target sign (central hyperintensity with a hypointense rim); and (iii) reversed target sign (central hypointensity with a hyperintense rim). According to the literature (32,33), on Gd-EOB-MRI, a lesion was considered a metastasis if it showed intermediate signal intensity on T2WI, hypointense signal on T1WI, high signal intensity on DWI at b=800 s/mm² with an ADC equal to or lower than that of the adjacent liver parenchyma, faint peripheral ring-like enhancement on the arterial and portal phases, and hypointense signal on HBP. Although these criteria were provided to the observers as reference standards, the final determination was made based on each observer’s subjective judgment. Each reader recorded the presence, size (the maximum diameter), and location of the lesion and assigned confidence levels: (i) not metastatic; (ii) possibly not metastatic; (iii) indeterminate; (iv) probably metastatic; and (v) definitely metastatic. Lesions that scored 4 or 5 points were classified as metastatic for subsequent analyses. Benign lesions (e.g., hemangiomas, cysts), considered negative findings, were typically straightforward to diagnose on MRI and were outside the primary objective.

Statistical analysis

Interrater agreement in the detection of hepatic steatosis, HIO, and SOS was assessed with the kappa (κ) statistic. Diagnostic performance, including for all lesions and for lesions smaller than 10 mm, was evaluated using the sensitivity, positive predictive value (PPV), and the area under the receiver operating characteristic (ROC) curve (AUC, Az). Factors affecting lesion detection were analyzed using generalized estimating equations (GEEs), and the results were reported as odds ratios (ORs) and 95% confidence intervals (CIs). The McNemar test and GEEs were used to assess the significance of differences in diagnostic performance for CRLMs among the MR sequences (DWI set, HBP set, and combined set). A P value <0.05 was considered indicative of statistical significance. In the pairwise comparison of sensitivity and PPV, the statistical results were considered significant if the P value was less than 0.01671 (i.e., following adjustment with the Bonferroni method). Statistical analyses were performed using the software SPSS 25.0 (IBM Corp., Armonk, NY, USA).

Results

Cohort and lesion characteristics

A total of 35 patients (26 men; mean age, 61.6±1.7 years) with suspected CRLMs, all postchemotherapy, were included, including 10 who experienced recurrence after prior treatment and 25 with newly diagnosed CRLM (Figure 1). MRI identified 118 suspected liver metastases among the patients (median 3 per patient; range, 1–5; median size, 12.7 mm, 44.1% ≤10 mm). In all 35 patients, the interval from the index MRI to CRLM confirmation by histopathology or treatment response was ≤3 months. There were 10 false-positive lesions in 4 patients (4 atypical hemangiomas, 1 fat necrosis with calcification, 2 posttreatment changes lacking epithelial components, 1 inflammatory infiltration, and 2 sinusoidal congestions) and 108 true-positive lesions in 31 patients, of which 53 were confirmed via resection, 2 via biopsy, 26 via intraoperative contrast-enhanced ultrasound (treated with microwave ablation), and 27 via their clinical responses to chemotherapy. The hepatic background was normal in 5 patients (14.3%) and abnormal in 30 (85.7%). Interrater agreement was high in the identification of hepatic steatosis (κ=0.802), HIO (κ=0.813), and SOS (κ=0.881). On HBP imaging, 86.1% (93/108) of all lesions and 71.7% (33/46) of lesions ≤10 mm presented with a target or reverse target sign (Figure 1I). The baseline characteristics of the patients and lesions are shown in Table 1.

Figure 1 Gd-EOB-DTPA MRI (EOB-MRI, 5T) of CRLM in a 58-year-old male with moderate hepatic steatosis following adjuvant chemotherapy for rectal cancer (patient 1). (A,B) T1W in-/opposed-phase images show signal drop in the lesion (white circles) and adjacent liver, indicating steatosis. (C) Fat fraction map (11.67%) indicates moderate hepatic steatosis. (D) Fat-suppressed T2WI shows a mildly hyperintense lesion (white circle). (E) DWI (b=1,000 s/mm²) demonstrates a marked hyperintensity (white circle). (F) ADC map confirms restricted diffusion (white circle). (G,H) Arterial and PVP images show peripheral rim enhancement (white circles). (I) HBP image shows the target sign in smaller lesions (hyperintense center, hypointense rim, white arrow) and the reverse target sign in larger lesions (hyperintense rim, hypointense center, arrowhead). Histopathological examination of surgically resected specimens CRLM. 5T, 5 Tesla; ADC, apparent diffusion coefficient; CRLM, colorectal liver metastases; DWI, diffusion-weighted imaging; Gd-EOB-DTPA, gadolinium-ethoxybenzyl-diethylenetriamine pentaacetic acid; HBP, hepatobiliary phase; MRI, magnetic resonance imaging; PVP, portal-venous phase; T1W, T1-weighted; T2WI, T2-weighted imaging.

Factors influencing CRLM detection performance according to GEE analysis

GEE analysis (Table 2) demonstrated that both combined imaging (Reader 1: β=1.70, 95% CI: 0.60–2.79, P=0.002; Reader 2: β=1.61, 0.67–2.55, P=0.001) and HBP imaging (Reader 1: β=1.55, 0.56–2.53, P=0.002; Reader 2: β=1.48, 0.61–2.35, P=0.001) significantly outperformed DWI alone; compared with HBP imaging, combined imaging showing slightly greater effect magnitudes. Larger lesions (>10 mm) were associated with increased detectability (Reader 1: β=1.25, 0.10–2.40, P=0.033; Reader 2: β=1.31, 0.24–2.39, P=0.017), whereas lesion location had no significant effect on detectability (P>0.36). Hepatic background factors, including SOS, HIO, and steatosis, tended to nonsignificantly reduce diagnostic efficacy (all P>0.05; Figures 1,2). Similarly, clinical factors such as prior surgery or a history of RFA had no significant impact on detection performance (P>0.12; Figure 3).

Figure 2 EOB-MRI (5T) findings in a 70-year-old male with CRLMs against a complex hepatic background of HID and SOS after rectal cancer surgery and adjuvant chemotherapy (patient 2). (A,B) SWI and phase images show a heterogeneous hypointensity in the liver parenchyma, suggesting HID. (C) R2* map (mean value: 90.44) indicates moderate HID. A heterogenous reticular hypointense signal in the liver background on PVP (F) and HBP images (H), suggesting SOS. (D) Small-FOV-DWI (b=1,000 s/mm²) reveals marked hyperintensity in a small lesion in segment IV (white arrow). (E) Corresponding ADC map shows a slightly hyperintense signal without obviously restricted diffusion (white arrow). The diagnostic confidence score was 3, suggesting uncertainty in diagnosing metastasis. (F,G) PVP and TP contrast-enhanced T1W images demonstrate peripheral rim enhancement in the lesion (white arrows). (H) HBP images depict the segment IV lesion with a target sign-like appearance (white arrow). On the HBP and combined sequences, the diagnostic confidence score was 4, strongly suggesting metastasis. Histopathological examination confirmed colorectal liver metastases. 5T, 5 Tesla; ADC, apparent diffusion coefficient; CRLM, colorectal liver metastases; EOB-MRI, gadoxetic acid-enhanced magnetic resonance imaging; FOV, field of view; HBP, hepatobiliary phase; HID, hepatic iron deposition; PVP, portal-venous phase; SOS, sinusoidal obstruction syndrome; SWI, susceptibility weighted imaging; T1W, T1-weighted; TP, transitional phase.

Figure 3 EOB-MRI (5T) of a 57-year-old man (patient 3) demonstrating recurrent CRLM, and marked iron deposition in the background liver after chemotherapy and local microwave ablation. (A) SWI images show significantly reduced signal intensity of the liver parenchyma. (B) R2* map demonstrates an average R2* value of approximately 237.9, combined with SWI findings, indicating significant hepatic iron deposition. (C) T2WI shows slightly hyperintense signal in the segment VIII lesion (white circle). (D) DWI (b=1,000 s/mm²) reveals marked hyperintensity in the lesion (white circle). (E) Corresponding ADC map shows slightly hyperintense signal without obvious restricted diffusion (white circle). The diagnostic confidence score was 3, suggesting uncertain metastasis. (F,G) Arterial phase and transitional phase contrast-enhanced T1WI demonstrate peripheral rim enhancement in the lesion (white circles). (H) HBP shows a target-like appearance (white circle). The diagnostic confidence score was 4 on both HBP and combined series, strongly suggesting metastasis. (I,J) Follow-up T2WI and HBP imaging at 3 months demonstrate significant lesion enlargement (white circles), confirming the diagnosis of CRLM. 5T, 5 Tesla; ADC, apparent diffusion coefficient; CRLM, colorectal liver metastases; DWI, diffusion-weighted imaging; EOB-MRI, gadoxetic acid-enhanced magnetic resonance imaging; HBP, hepatobiliary phase; SWI, susceptibility weighted imaging; T1WI, T1-weighted imaging; T2WI, T2-weighted imaging.

These findings demonstrate the superiority of combined imaging and HBP imaging over DWI in detecting lesions, particularly larger lesions, while highlighting that diagnostic performance may not be affected by background liver changes or previous treatments.

Diagnostic performance and ROC curve analysis of DWI, HBP imaging, and combined imaging for CRLM

For all lesions (n=118), HBP imaging was significantly more sensitive (Table 3, Reader 1: 96.3%, Reader 2: 95.4%, both P<0.001) than DWI was (81.5%, 80.6%, respectively). Combined imaging achieved the highest sensitivity (97.2%), which was not significantly different from that of HBP imaging (P>0.05). For small lesions (≤10 mm; n=52), HBP (93.6%, 91.5%) and combined imaging (95.7%) demonstrated significantly greater sensitivities than did DWI (66.0%, P=0.001–0.002), whereas consistently high PPVs were achieved with all imaging sequences (Table 3, all P>0.05). Figures 2-4 illustrate the superior diagnostic confidence achieved with combined and HBP imaging (score of 4) for small lesions compared to DWI (score 3).

Figure 4 EOB-MRI (5T) findings in a 57-year-old male with CRLM and severe SOS after radical resection of right colon cancer and 7 cycles of adjuvant chemotherapy (patient 4). (A) HBP show diffuse reticular hypointensity of the hepatic parenchyma, indicative of severe SOS. (B,C) DWI (b=1,000 s/mm²) and the corresponding ADC map, respectively, show a lesion in segment VIII with ring-shaped hyperintensity on DWI and marked hyperintensity on ADC (confidence =3; metastasis indeterminate). (D) On T2WI the lesion is mildly hyperintense. (E) PVP depicts peripheral ring enhancement. (F) On HBP, the lesion displays a target-sign appearance (white circle); in light of the HBP and combined sequences, the diagnostic-confidence score rose to 4, strongly suggesting metastasis. Contrast-enhanced intra-operative ultrasound confirmed CRLM, and ultrasound-guided microwave ablation was performed. Six-month post-treatment MRI: (G) T1WI shows slight central hyperintensity with corresponding T2 hypointensity (not shown), compatible with coagulative necrosis; (H) PVP shows no enhancement, indicating no recurrence. 5T, 5 Tesla; ADC, apparent diffusion coefficient; CRLM, colorectal liver metastases; DWI, diffusion-weighted imaging; EOB-MRI, gadoxetic acid-enhanced magnetic resonance imaging; HBP, hepatobiliary phase; MRI, magnetic resonance imaging; PVP, portal-venous phase; SOS, sinusoidal obstruction syndrome; T1WI, T1-weighted imaging; T2WI, T2-weighted imaging.

In ROC curve analysis, combined imaging had the highest Az values for identifying all lesions (Table 4, Reader 1: 0.880, Reader 2: 0.917), which were comparable to those of HBP imaging (0.890, 0.885) but superior to those of DWI (0.843, 0.848). For small lesions, combined imaging achieved greater Az values (Table 4, Reader 1: 0.768, Reader 2: 0.860) than both HBP imaging (both readers: 0.785) and DWI (0.628, 0.668).

These findings demonstrate the diagnostic advantages of combined imaging, particularly for small lesions, effectively addressing the limitations of the individual modalities.

Discussion

In this prospective single-center study, the efficacy of 5T EOB-MRI with ACS for detecting CRLMs under CALI-induced hepatic background changes (including prior local CRLM treatment) was evaluated. The results demonstrated that the combination of ACS-HBP imaging and DWI achieved 97.2% sensitivity in detecting CRLMs, significantly outperforming DWI alone (66.0%, P=0.001–0.002). GEE analysis confirmed that neither hepatic background factors (SOS, HID, steatosis) nor a history of local CRLM treatment had a significant impact on diagnostic performance (all P>0.05), indicating that this imaging combination maintains stable diagnostic performance across various forms of complex hepatic background.

Our findings strongly align with those of previous studies, confirming the diagnostic efficacy of EOB-MRI with DWI for CRLMs (24,34-36). For small lesions (≤10 mm), Schulz et al. (35) reported 87% sensitivity with 1.5T EOB-MRI, whereas Lee et al. (36) demonstrated 81.0% sensitivity with 3 T MRI, both in combination with HBP imaging. Comparatively, the imaging modalities employed in our 5T study achieved superior sensitivity (95.7%) and PPV (93.8–95.7%). These increases in performance can be attributed to the higher SNR and improved soft tissue resolution of the 5T system as well as the ability of the ACS-HBP technique to deliver high-resolution 3D isotropic T1W FSPGR imaging, generating 300 whole-liver images capable of multiplanar reconstruction, significantly improving small lesion detection. These results corroborate the findings of Ihara et al. (30), who showed that the use of CS techniques in HBP EOB-MRI improves image quality and lesion conspicuity over parallel imaging without extending the acquisition time.

Notably, although previous EOB-MRI diagnostic performance studies have typically examined untreated CRLM patients, our study specifically involved the data of patients who had undergone chemotherapy (including those with prior local treatments), which better reflects the clinical complexity of CRLM and provides more practical guidance for treatment planning. Although some researchers have addressed the diagnostic performance of EOB-MRI in detecting lesions in patients who have undergone chemotherapy—for example, Sivesgaard et al. (37), who demonstrated that EOB-MRI can achieve superior detection rates compared with contrast-enhanced CT and FDG-PET/CT regardless of chemotherapy status, and Yu et al. (18) , who reported >90% sensitivity in detecting postchemotherapy CRLMs using EOB-MRI with DWI—our study’s novelty lies in the comprehensive assessment of the liver background. We implemented an integrated qualitative-quantitative evaluation approach, including qualitative assessments of fatty liver using dual-echo T1WI with PDFF value-based grading; evaluation of HIO using SWI and phase mapping with R2* value-based grading; and assessment of SOS severity on the basis of signal heterogeneity on portal venous-phase and HBP images. This comprehensive methodology laid a foundation for objectively evaluating the diagnostic performance of EOB-MRI for CRLMs across different liver backgrounds.

According to prior studies using 3 T and lower-field MRI systems, CRLMs typically exhibit imaging features such as the “target” or “reverse-target” sign on HBP imaging (32,33). Interestingly, our study demonstrated that these imaging patterns can also be clearly visualized on 5T ACS-HBP imaging, even in microlesions measuring ≤10 mm. These patterns were observed in 86.1% (93/108) of all CRLMs and in 71.7% (33/46) of CRLMs ≤10 mm. This finding can be attributed to the superior SNR and high spatial resolution of 5T ACS-HBP imaging. Moreover, this finding not only demonstrates the improved detection of small CRLMs but also provides valuable differential diagnostic information, allowing the differentiation of these lesions from hepatic cysts, which typically appear as markedly hypointense lesions on HBP imaging (38).

This study has several limitations. First, although it yielded statistically significant results from 35 patients (118 lesions), its single-center design and lack of a chemotherapy-naive CRLM control group mean that validation through larger multicenter comparative studies is needed. Second, a non-chemotherapy control was lacking, and the CRLM chemotherapy response was not assessed; in our cohort, 97.1% (34/35) received NAC/AC with no CRLM at the time, and new or recurrent CRLM were detected only on follow-up MRI, indicating minimal direct chemotherapeutic effects on lesions (mainly background parenchymal injury), yet the technique remains applicable to patients without prior chemotherapy. Third, CALIs were assessed only by imaging, and pathological confirmation was lacking; nevertheless, other studies have confirmed the high accuracy and reproducibility of CALI imaging-based assessments. Fourth, we lacked a true-negative control group, making it impossible to calculate specificity and the negative predictive value, and the PPV may be overestimated under high disease prevalence (spectrum bias); therefore, future studies should include a true-negative control group to enable unbiased estimation of sensitivity, specificity, PPV, and negative predictive value, and to quantify misclassification of benign/normal liver findings.

Conclusions

Our results show that ACS combined with gadolinium-ethoxybenzyl-diethylenetriamine pentaacetic acid (Gd-EOB-DTPA)-enhanced 5T MRI provides superior diagnostic performance in the detection of CRLMs across diverse hepatic backgrounds, maintaining high sensitivity and PPV (both >90%) for subcentimeter lesions (≤10 mm). On ACS-HBP imaging, CRLMs characteristically display target and reverse-target signs, especially in lesions ≤10 mm, facilitating differentiation from hepatic cysts.

Acknowledgments

The authors extend their sincere gratitude to Dr. Xiaojun Xu (Department of Radiology, The First Affiliated Hospital of USTC, Division of Life Sciences and Medicine, University of Science and Technology of China), whose generous resources and encouragement were instrumental in the completion and publication of this manuscript. Her thoughtful recommendations and critical insights have significantly improved the quality and depth of this research.

Footnote

Reporting Checklist: The authors have completed the STARD reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1425/rc

Data Sharing Statement: Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1425/dss

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1425/coif). The 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 protocol was approved by the Medical Research Ethics Committee of The First Affiliated Hospital of University of Science and Technology of China (No. XJS2023-1-30[YJ]). Written informed consent was obtained from all 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. Li Q, Xia C, Li H, Yan X, Yang F, Cao M, Zhang S, Teng Y, He S, Cao M, Chen W. Disparities in 36 cancers across 185 countries: secondary analysis of global cancer statistics. Front Med 2024;18:911-20. [Crossref] [PubMed]
2. Bizuayehu HM, Ahmed KY, Kibret GD, Dadi AF, Belachew SA, Bagade T, Tegegne TK, Venchiarutti RL, Kibret KT, Hailegebireal AH, Assefa Y, Khan MN, Abajobir A, Alene KA, Mengesha Z, Erku D, Enquobahrie DA, Minas TZ, Misgan E, Ross AG. Global Disparities of Cancer and Its Projected Burden in 2050. JAMA Netw Open 2024;7:e2443198. [Crossref] [PubMed]
3. Siegel RL, Wagle NS, Cercek A, Smith RA, Jemal A. Colorectal cancer statistics, 2023. CA Cancer J Clin 2023;73:233-54. [Crossref] [PubMed]
4. Zheng RS, Chen R, Han BF, Wang SM, Li L, Sun KX, Zeng HM, Wei WQ, He J. Cancer incidence and mortality in China, 2022. Chinese Journal of Oncology 2024;46:221-31. [Crossref] [PubMed]
5. Patel RK, Rahman S, Schwantes IR, Bartlett A, Eil R, Farsad K, Fowler K, Goodyear SM, Hansen L, Kardosh A, Nabavizadeh N, Rocha FG, Tsikitis VL, Wong MH, Mayo SC. Updated Management of Colorectal Cancer Liver Metastases: Scientific Advances Driving Modern Therapeutic Innovations. Cell Mol Gastroenterol Hepatol 2023;16:881-94. [Crossref] [PubMed]
6. de Jong MC, Pulitano C, Ribero D, Strub J, Mentha G, Schulick RD, Choti MA, Aldrighetti L, Capussotti L, Pawlik TM. Rates and patterns of recurrence following curative intent surgery for colorectal liver metastasis: an international multi-institutional analysis of 1669 patients. Ann Surg 2009;250:440-8. [Crossref] [PubMed]
7. Glynne-Jones R, Wyrwicz L, Tiret E, Brown G, Rödel C, Cervantes A, Arnold D. Rectal cancer: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol 2017;28:iv22-40. [Crossref] [PubMed]
8. Adam R, Delvart V, Pascal G, Valeanu A, Castaing D, Azoulay D, Giacchetti S, Paule B, Kunstlinger F, Ghémard O, Levi F, Bismuth H. Rescue surgery for unresectable colorectal liver metastases downstaged by chemotherapy: a model to predict long-term survival. Ann Surg 2004;240:644-57; discussion 657-8. [Crossref] [PubMed]
9. Tatsuta K, Sakata M, Kojima T, Booka E, Kurachi K, Takeuchi H. Updated insights into the impact of adjuvant chemotherapy on recurrence and survival after curative resection of liver or lung metastases in colorectal cancer: a rapid review and meta-analysis. World J Surg Oncol 2025;23:56. [Crossref] [PubMed]
10. Robinson PJ. The effects of cancer chemotherapy on liver imaging. Eur Radiol 2009;19:1752-62. [Crossref] [PubMed]
11. Viganò L, Rubbia-Brandt L, De Rosa G, Majno P, Langella S, Toso C, Mentha G, Capussotti L. Nodular Regenerative Hyperplasia in Patients Undergoing Liver Resection for Colorectal Metastases After Chemotherapy: Risk Factors, Preoperative Assessment and Clinical Impact. Ann Surg Oncol 2015;22:4149-57. [Crossref] [PubMed]
12. Sharma A, Houshyar R, Bhosale P, Choi JI, Gulati R, Lall C. Chemotherapy induced liver abnormalities: an imaging perspective. Clin Mol Hepatol 2014;20:317-26. [Crossref] [PubMed]
13. Calistri L, Rastrelli V, Nardi C, Maraghelli D, Vidali S, Pietragalla M, Colagrande S. Imaging of the chemotherapy-induced hepatic damage: Yellow liver, blue liver, and pseudocirrhosis. World J Gastroenterol 2021;27:7866-93. [Crossref] [PubMed]
14. Yamashita S, Venkatesan AM, Mizuno T, Aloia TA, Chun YS, Lee JE, Vauthey JN, Conrad C. Remnant Liver Ischemia as a Prognostic Factor for Cancer-Specific Survival After Resection of Colorectal Liver Metastases. JAMA Surg 2017;152:e172986. [Crossref] [PubMed]
15. Kim YS, Rhim H, Lim HK, Choi D, Lee MW, Park MJ. Coagulation necrosis induced by radiofrequency ablation in the liver: histopathologic and radiologic review of usual to extremely rare changes. Radiographics 2011;31:377-90. [Crossref] [PubMed]
16. You SH, Park BJ, Kim YH. Hepatic Lesions that Mimic Metastasis on Radiological Imaging during Chemotherapy for Gastrointestinal Malignancy: Recent Updates. Korean J Radiol 2017;18:413-26. [Crossref] [PubMed]
17. Carnaghi C, Tronconi MC, Rimassa L, Tondulli L, Zuradelli M, Rodari M, Doci R, Luttmann F, Torzilli G, Rubello D, Al-Nahhas A, Santoro A, Chiti A. Utility of 18F-FDG PET and contrast-enhanced CT scan in the assessment of residual liver metastasis from colorectal cancer following adjuvant chemotherapy. Nucl Med Rev Cent East Eur 2007;10:12-5.
18. Yu MH, Lee JM, Hur BY, Kim TY, Jeong SY, Yi NJ, Suh KS, Han JK, Choi BI. Gadoxetic acid-enhanced MRI and diffusion-weighted imaging for the detection of colorectal liver metastases after neoadjuvant chemotherapy. Eur Radiol 2015;25:2428-36. [Crossref] [PubMed]
19. Cervantes A, Adam R, Roselló S, Arnold D, Normanno N, Taïeb J, Seligmann J, De Baere T, Osterlund P, Yoshino T, Martinelli EESMO Guidelines Committee. Electronic address: clinicalguidelines@esmo. Metastatic colorectal cancer: ESMO Clinical Practice Guideline for diagnosis, treatment and follow-up. Ann Oncol 2023;34:10-32. [Crossref] [PubMed]
20. Imajo K, Kessoku T, Honda Y, Hasegawa S, Tomeno W, Ogawa Y, Motosugi U, Saigusa Y, Yoneda M, Kirikoshi H, Yamanaka S, Utsunomiya D, Saito S, Nakajima A. MRI-Based Quantitative R2(*) Mapping at 3 Tesla Reflects Hepatic Iron Overload and Pathogenesis in Nonalcoholic Fatty Liver Disease Patients. J Magn Reson Imaging 2022;55:111-25. [Crossref] [PubMed]
21. Liu J, Wang Z, Yu D, Yang Y, Li Z, Wang X, Yang Y, Cheng C, Zou C, Gan J. Comparative analysis of hepatic fat quantification across 5 T, 3 T and 1.5 T: A study on consistency and feasibility. Eur J Radiol 2024;180:111709. [Crossref] [PubMed]
22. Han NY, Park BJ, Yang KS, Kim MJ, Sung DJ, Sim KC, Cho SB. Hepatic Parenchymal Heterogeneity as a Marker for Oxaliplatin-Induced Sinusoidal Obstruction Syndrome: Correlation With Treatment Response of Colorectal Cancer Liver Metastases. AJR Am J Roentgenol 2017;209:1039-45. [Crossref] [PubMed]
23. Shin NY, Kim MJ, Lim JS, Park MS, Chung YE, Choi JY, Kim KW, Park YN. Accuracy of gadoxetic acid-enhanced magnetic resonance imaging for the diagnosis of sinusoidal obstruction syndrome in patients with chemotherapy-treated colorectal liver metastases. Eur Radiol 2012;22:864-71. [Crossref] [PubMed]
24. Görgec B, Hansen IS, Kemmerich G, Syversveen T, Abu Hilal M, Belt EJT, et al. MRI in addition to CT in patients scheduled for local therapy of colorectal liver metastases (CAMINO): an international, multicentre, prospective, diagnostic accuracy trial. Lancet Oncol 2024;25:137-46. [Crossref] [PubMed]
25. Xu Q, Zhao H, Gao R, Wang X, Xu J, Sun G, Xue K, Yang Y, Li E, Zhu L, Wu W, Feng F. Insulinoma detection and surgery planning: a comparative study of 5.0T MRI versus 3.0T MRI and MDCT. Abdom Radiol (NY) 2025;50:4148-59. [Crossref] [PubMed]
26. Yin L, Li Z, Shang M, Li Z, Tang B, Yu D, Gan J. Magnetic resonance cholangiopancreatography at 5.0 T: quantitative and qualitative comparison with 3.0 T. BMC Med Imaging 2024;24:331. [Crossref] [PubMed]
27. Zheng L, Yang C, Sheng R, Dai Y, Zeng M. Renal imaging at 5 T versus 3 T: a comparison study. Insights Imaging 2022;13:155. [Crossref] [PubMed]
28. 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]
29. Sheng RF, Zheng LY, Jin KP, Sun W, Liao S, Zeng MS, Dai YM. Single-breath-hold T2WI liver MRI with deep learning-based reconstruction: A clinical feasibility study in comparison to conventional multi-breath-hold T2WI liver MRI. Magn Reson Imaging 2021;81:75-81. [Crossref] [PubMed]
30. Ihara K, Onoda H, Tanabe M, Iida E, Ueda T, Kobayashi T, Higashi M, Nickel MD, Imai H, Ito K. Breath-hold High-resolution T1-weighted Gradient Echo Liver MR Imaging with Compressed Sensing Obtained during the Gadoxetic Acid-enhanced Hepatobiliary Phase: Image Quality and Lesion Visibility Compared with a Standard T1-weighted Sequence. Magn Reson Med Sci 2024;23:146-52. [Crossref] [PubMed]
31. Kang HJ, Lee JM, Ahn SJ, Bae JS, Kannengiesser S, Kiefer B, Suh KS. Clinical Feasibility of Gadoxetic Acid-Enhanced Isotropic High-Resolution 3-Dimensional Magnetic Resonance Cholangiography Using an Iterative Denoising Algorithm for Evaluation of the Biliary Anatomy of Living Liver Donors. Invest Radiol 2019;54:103-9. [Crossref] [PubMed]
32. Kim A, Lee CH, Kim BH, Lee J, Choi JW, Park YS, Kim KA, Park CM. Gadoxetic acid-enhanced 3.0T MRI for the evaluation of hepatic metastasis from colorectal cancer: metastasis is not always seen as a "defect" on the hepatobiliary phase. Eur J Radiol 2012;81:3998-4004. [Crossref] [PubMed]
33. Granata V, Catalano O, Fusco R, Tatangelo F, Rega D, Nasti G, Avallone A, Piccirillo M, Izzo F, Petrillo A. The target sign in colorectal liver metastases: an atypical Gd-EOB-DTPA "uptake" on the hepatobiliary phase of MR imaging. Abdom Imaging 2015;40:2364-71. [Crossref] [PubMed]
34. Zech CJ, Korpraphong P, Huppertz A, Denecke T, Kim MJ, Tanomkiat W, Jonas E, Ba-Ssalamah AVALUE study group. Randomized multicentre trial of gadoxetic acid-enhanced MRI versus conventional MRI or CT in the staging of colorectal cancer liver metastases. Br J Surg 2014;101:613-21. [Crossref] [PubMed]
35. Schulz A, Joelsen-Hatlehol ES, Brudvik KW, Aasand KK, Hanekamp B, Viktil E, Johansen CK, Dormagen JB. Preoperative detection of colorectal liver metastases: DWI alone or combined with MDCT is no substitute for Gd-EOB-DTPA-enhanced MRI. Acta Radiol 2020;61:302-11. [Crossref] [PubMed]
36. Lee KH, Lee JM, Park JH, Kim JH, Park HS, Yu MH, Yoon JH, Han JK, Choi BI. MR imaging in patients with suspected liver metastases: value of liver-specific contrast agent gadoxetic acid. Korean J Radiol 2013;14:894-904. [Crossref] [PubMed]
37. Sivesgaard K, Larsen LP, Sørensen M, Kramer S, Schlander S, Amanavicius N, Bharadwaz A, Tønner Nielsen D, Viborg Mortensen F, Morre Pedersen E. Diagnostic accuracy of CE-CT, MRI and FDG PET/CT for detecting colorectal cancer liver metastases in patients considered eligible for hepatic resection and/or local ablation. Eur Radiol 2018;28:4735-47. [Crossref] [PubMed]
38. Lee JE, Kim SH, Lee S, Choi SY, Hwang JA, Woo SY. Differentiating metastatic mucinous colorectal adenocarcinomas from simple cysts of the liver using contrast-enhanced and diffusion-weighted MRI. Br J Radiol 2018;91:20180303. [Crossref] [PubMed]