Initial quantitative analysis of gemstone spectral imaging CT: Analysis and discussion of the application value for the differentiation of small liver metastases (<3 cm)

Abstract

Background

Hepatic metastases are the most common malignant tumors in the liver. Conventional contrast-enhanced CT examinations face challenges in distinguishing between benign and malignant atypical metastatic liver lesions with a diameter <3 cm, and evaluating their therapeutic efficacy remains particularly difficult.

Objective

To assess the clinical value of quantitative iodine analysis and spectral curves for diagnosing and differentiating liver metastases using gemstone spectral CT.

Methods

Among 915 patients with suspected hepatic metastases, 140 cases (87 males, 53 females) were pathologically confirmed. Primary malignancies included colorectal cancer (41 cases), gastric cancer (21 cases), lung cancer (35 cases), pancreatic cancer (31 cases), and breast cancer (12 cases). A total of 425 small lesions (<3 cm) were detected. CT values at lesion centers and peripheries were measured and compared against normal liver parenchyma. Quantitative iodine concentrations and spectral curve slopes were analyzed to evaluate differences in small hepatic metastases originating from distinct primary malignancies

Results

In the differentiation of hepatic metastatic subtypes, mean CT values demonstrated diagnostic utility in distinguishing colorectal cancer from gastric in AP (Arterial Phase, P = 0.001) in the Center of lesions.On the contrary, the quantitative analysis of focus edge iodine reliably distinguished the AP (P < 0.000), lung cancer (P = 0.023) and pancreatic cancer (P < 0.000) of colorectal cancer and gastric cancer. The results have statistical significance.

Conclusions

GSI(Gemstone spectral Imaging)-derived spectral curve slope and quantitative iodine analysis may facilitate differential diagnosis of small hepatic metastatic lesions with diverse primary origins, especially the source of colorectal cancer.

Keywords

Small liver metastases iodine curve GSI CT

1. Introduction

Liver metastases represent the most prevalent malignant liver tumors.¹ However, differentiating small hepatic metastases (<3 cm) from benign lesions remains challenging. These small metastases often lack typical enhancement patterns, making them difficult to distinguish visually from benign lesions such as atypical cavernous hemangiomas. Moreover, detecting metastatic liver tumors with atypical imaging features and assessing treatment response pose significant clinical difficulties.

While angiography, PET, MRI, and Doppler ultrasonography can aid in diagnosing and differentiating metastatic liver tumors, CT remains the primary imaging modality for clinical diagnosis, differential diagnosis, and follow-up of hepatic metastases.² In recent years, gemstone spectral CT has been validated as an effective modality for detecting small hepatic lesions.^3,4 Prior studies indicate that spectral CT may improve diagnostic accuracy in hepatic tumor characterization.^5–7 Studies demonstrate that spectral CT parameters - including the energy spectrum curve, optimal monoenergetic images, and iodine quantification - can improve metastatic lymph node detection and lesion characterization.^8,9 The technique, introduced by P.J.Lv et al.,¹⁰ is applied to distinguish liver hemangioma, focal nodule hyperplasia (FNH), or hepatic vascular smooth muscle tumors from hepatocellular carcinoma.^6,11,12

Given gemstone spectral CT's unique tissue characterization capabilities, its diagnostic value for hepatic tumors warrants further investigation.^12,13 However, to our knowledge, no comparative study has evaluated different gemstone spectral imaging (GSI) parameters—such as iodine concentration and spectral curve slope—for differentiating metastatic liver tumors.This study aims to detect early-stage small hepatic metastases and characterize gemstone spectral imaging (GSI) features of metastases originating from different primary tumors.We present the following article in accordance with the MDAR reporting checklist.

Gemstone spectral CT enables scanning at 40–140 keV, allowing extraction of monoenergetic images at any energy level within this range. This optimizes lesion-to-background contrast while reducing beam-hardening artifacts, ensuring superior image quality.Gemstone spectral CT acquires mixed-energy images, monoenergetic images, and material decomposition images simultaneously through a single volumetric scan. This facilitates high-resolution spectral analysis and enables a multi-parameter diagnostic approach. The technology effectively minimizes hardening artifacts and enhances lesion detection in parenchymal organs, significantly improving tumor localization, characterization, and staging accuracy.

2. Materials and methods

2.1. Study population

The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). This study was approved by the institutional review board of Tongji hospital (ID:TJ-IRB20181123), and the requirement for formal informed consent was obtained from all patients. A total of 915 consecutive patients were enrolled in this study, and all patients underwent a CT scan (Discovery CT 750 HD; GE Healthcare, Waukesha, Wis) to evaluate metastatic hepatic tumors between December 2018 and December 2021. The inclusion criteria were as follows: (a) Solid hepatic tumors with a cancer history or suspected metastatic tumors; (b) CT GSI was performed in patients before the treatment of hepatic tumors; (c) histopathologic diagnosis or clinical follow-up was confirmed after the completion of the CT imaging examination according to current WHO guidelines. The exclusion criteria were as follows: (a) CT data were not available owing to patients’ movement artifacts; (b) the maximum diameter of the solid tumor was >3 cm.

The study excluded 192 patients without pathological confirmation, 536 patients with solid tumors exceeding 3 cm, 32 patients with rare metastasis subtypes (including 7 esophageal, 6 papillary RCC, 4 cervical/nasopharyngeal/ovarian cancers each, 3 renal carcinomas, and single cases of bladder cancer, melanoma, duodenal cancer, and sinus cancer), and 15 patients lacking histopathological diagnosis, ultimately enrolling 140 eligible patients for analysis (see Figure 1 for the enrollment flowchart).

Figure 1.

Flow diagram showing the lesion exclusion criteria and final patient cohort.

2.2. CT technique

All examinations were performed on a dual-energy 64 slice multi-detector CT system (Discovery CT 750 High Definition; GE Healthcare Technologies, Waukesha, Wisconsin, USA). The hepatic arterial phase (AP) was scanned by bolus tracking (Smartprep; GE Healthcare Technologies) when the threshold enhancement of 120 HU was reached in the abdominal aorta. The scanning of the portal venous phase (PVP, Portal Venous Phase;) was started 30 s after completing the AP scanning. Non-ionic iodinated contrast agent (Iopromide 370 mg I/mL) was injected via the antecubital vein using an 18-gauge intravenous cannula and a double tube high-pressure syringe (Stellant D; Medred Co., Pittsburgh, Pennsylvania, USA) followed by 20 mL of physiological saline, each at an injection rate of 3–4 mL/s. The patients were in the supine position. The total contrast volume was 0.8 mL/kg. AP and PVP scanning was performed in the spectral imaging mode with fast tube voltage switching between 140 kVp and 40 kVp. Other CT imaging parameters were as follows: rotation time 0.6 s; matrix 512 × 512; heli-cal pitch 0.984:1; slice thickness/interval 0.625 mm; and table speed 39.37 mm/rotation. Three kinds of images were reconstructed from the single spectral CT acquisition for analysis: conventional polychromatic images were obtained at 140 kVp, iodine-based and water-based, monochromatic images, and material-decomposition images were obtained at energies ranging from 40 to 140 keV.

2.3. Image processing and analysis

All CT data was transferred from the picture archiving and communication system (PACS) to a workstation (AW4.6; GE Healthcare). Two independent blinded CT scan technologists with 6 and 16 years of experience reviewed the quality of all images. Regions of interest (ROI) were drawn in the central and peripheral regions of hepatic metastatic lesions in AP and PVP by three experienced imaging diagnosicians. Normal liver tissue was analyzed as the control at the same time. The parameters derived from the analysis included: CT value, iodine value, and energy spectrum curve slope characteristics (λHU). We found that the slope of the spectral curve was higher at HU = 60. In general, the curve between 100 Kev and 140 Kev was almost flat, so we chose a series of high-slope 60 and 100 Kev sensitivity increases representing the spectral attenuation curves of different slopes over the entire range (details are provided in Figure 2). Receiver operating characteristic (ROC, Receiver Operating Characteristic) curves were obtained based on iodine quantitative analysis. According to the formula λHU = [CT60-kev1-CT100 KeV] / 60, the slope of the curve is calculated as the CT attenuation difference of two energy levels (60 KeV and 100 KeV) divided by the energy difference of the spectral HU curve (60 KeV).

Figure 2.

Illustration of the calculation of the ratio. A. The primary imaging; To draw two vertical lines: x = 60 and x = 100, get two intersections A and B; To get the straight line which runs through intersections A and B. The slope of straight line AB served as the slope of the primary curve in imaging A.

2.4. Statistical analysis

Statistical analysis was performed with statistical software (PASW version 19.0, SPSS). One-way ANOVA was utilized to compare the CT value, the iodine value, and λHU between the 5 different hepatic metastatic tumor subtypes. Multiple comparisons between subtypes were accomplished using receiver operating characteristic curve. Sensitivity curve was used to compare the CT value, the iodine value, and λHU between the central regions and peripheral regions of hepatic metastatic lesions in the AP and PVP. P < 0.05 was considered statistically significantand with Bonferroni correction applied for multiple comparisons.

3. Results

3.1. Patient and lesion characteristics

This prospective study enrolled 140 patients (mean age 55.2 ± 12.3 years; range 26–81 years; 76 males, 64 females) with histologically confirmed hepatic metastases, comprising 109 lesions from 41 colorectal cancers, 73 from 31 pancreatic cancers, 58 from 21 gastric cancers, 31 from 35 lung cancers, and 26 from 12 breast cancers. Quantitative analysis (Figure 3) revealed significant differences (p < 0.01) in CT attenuation, iodine density, and λHU values between lesion cores, margins, and adjacent parenchyma during both arterial and portal venous phases, with pancreatic metastases demonstrating characteristic spectral curve patterns (Figures 4–6). The spectral CT parameters provided superior discriminative capacity for small (<3 cm) metastases compared to conventional enhancement CT, particularly through iodine quantification. Figure 4A–B, Figure 5A–C, and Figure 6A–F show the features of the colorectal, pancreatic, gastric, lung, and breast metastases on CT values, iodine values, and λHU, ROC curve, respectively.

Figure 3.

A 56-year-old male patient with liver metastases from pancreatic cancer with abdominal pain. A: Routine computed tomography (CT) images showed that the CT values of the center, the marginal part of the lesion, and the normal liver tissue in the arterial phase were 19.88 HU, 49.56 HU, and 59.84 HU, respectively. B: Iodine base chart showed that the iodine values in the center of the lesion, the marginal part, and the normal liver tissue were 1.32 mg/mL, 5.08 mg/mL, and 4.48 mg/mL, respectively. C: The energy spectrum curve showed that the slopes of the energy spectrum curve of the center of the lesion, the edge part, and the normal liver tissue in the arterial phase were 0.181, 0.809, and 0.382, respectively. D: CT values of the center, the marginal part of the lesion, and the normal liver tissue in the venous phase were 20.2 HU, 48.72 HU, and 67.64 HU, respectively. E: Iodine values in the center, the edge of the lesion, and the normal liver tissue at the venous stage were 1.64 mg/mL, 4.6 mg/mL, and 6.28 mg/mL, respectively. F: The slopes of the energy spectrum curve of the center, the edge, and the normal liver tissue in the F venous stage were 0.423, 0.634, and 0.058, respectively. Note:The arrows indicate the measured hepatic metastatic mass from pancreatic cancer.

Figure 4.

Receiver operating characteristic (ROC) curve analysis of computed tomography (CT) values in 5 hepatic metastatic tumor subtypes. CT values could only distinguish colorectal cancer from gastric cancer hepatic metastasis in the centers of the lesions (Figure 4A: Blue - (lesion center) - Colorectal cancer vs gastric cancer (AP); Green - (center of lesion) - Colorectal cancer vs gastric cancer (PVP)), and colorectal cancer from lung cancer hepatic metastasis in peripheral regions (Figure 4B: Blue (edge of lesion) - Colorectal cancer vs lung cancer (PVP)).

Figure 5.

Receiver operating characteristic (ROC) curve analysis of iodine values in 5 hepatic metastatic tumor subtypes. Iodine values could distinguish colorectal cancer from gastric cancer hepatic metastasis in the arterial phase, both in peripheral regions and lesion centers (Figure 5A: Blue - (focus center) - colorectal cancer vs gastric cancer (AP); Green - (focus edge) - colorectal cancer vs gastric cancer (AP)). It also distinguished colorectal cancer from pancreatic cancer (Figure 5B: Blue - (focus edge) - colorectal cancer vs pancreatic cancer (AP) Green - (focus edge) - colorectal cancer vs pancreatic cancer (PVP)) and colorectal cancer from lung cancer, mainly in the peripheral regions (Figure 5C: Blue - (center of lesion) - Colorectal cancer vs lung cancer (PVP) Green - (edge of lesion) - Colorectal cancer vs lung cancer (AP) Yellow - (edge of lesion) - Colorectal cancer vs lung cancer (PVP)).

Figure 6.

Receiver operating characteristic (ROC) curve analysis of the energy spectrum curve slope (λHU) in 5 hepatic metastatic tumor subtypes. The λHU could distinguish almost all the pairs of hepatic metastases, including colorectal cancer vs. gastric cancer (Figure 6A: Blue - (lesion center) - Colorectal cancer vs gastric cancer (AP); Green - (lesion edge) - Colorectal cancer vs lung cancer (AP)), colorectal cancer vs. lung cancer (Figure 6B: Blue - (lesion center) - Colorectal cancer vs lung cancer (AP); Green - (lesion center) - Colorectal cancer vs lung cancer (PVP); Yellow - (lesion edge) - Colorectal cancer vs lung cancer (AP); Purple - (focus edge) - colorectal cancer vs lung cancer (AP)), gastric cancer vs. pancreatic cancer (Figure 6C: Blue - (focus edge) - gastric cancer vs pancreatic cancer (AP); Green - (focus edge) - gastric cancer vs pancreatic cancer ((PVP)),), pancreatic cancer vs. lung cancer (Figure 6D: Blue - (focus center) - colorectal cancer vs lung cancer (PVP)), and gastric cancer vs. lung cancer (Figure 6E: Blue - (focus edge) - gastric cancer vs lung cancer (PVP)).

3.2. Subtype comparison

The CT values in the center and the periphery of the metastatic lesions in the AP and PVP in the GSI images were measured in 5 metastatic liver tumors with the LSD test (Table 1). In the center of the lesions, only the differences between colorectal cancer and gastric cancer were statistically significant in AP (P < 0.001) and PVP (P < 0.000). In the peripheral region of the lesions, only colorectal cancer and lung cancer showed a significant difference in CT value in PVP (P = 0.04). All the other groups were not statistically significant (all P > 0.05). Quantitative analysis of iodine concentrations in central and peripheral tumor regions across arterial (AP) and portal venous (PVP) phases (Table 2) demonstrated distinct patterns: In central regions, colorectal cancer metastases showed significantly higher iodine uptake than gastric cancer during AP (P < 0.001) and lung cancer during PVP (P < 0.000), with no other intergroup differences reaching significance (all P > 0.05). Peripheral region analysis revealed more pronounced differentiation, with colorectal cancer exhibiting statistically greater iodine values versus gastric cancer in AP (P = 0.006), pancreatic cancer in both AP (P < 0.000) and PVP (P = 0.03), and lung cancer in both AP (P = 0.023) and PVP (P = 0.02). No other comparative analyses achieved statistical significance (all P > 0.05), suggesting these specific iodine uptake patterns may serve as discriminative features for metastasis origin identification.

Table 1.
Comparison results of the CT values in 5 hepatic metastatic tumor subtypes.

Subtype Comparison Center lesions A P Peripheral regions A P Center lesions V P Peripheral regions V P

Colorectal Pancreatic 44.29 ± 9.90 40.53 ± 10.63 0.163 59.68 ± 11.10 61.85 ± 10.86 0.220 51.85 ± 14.68 47.56 ± 14.93 0.450 76.48 ± 16.05 76.49 ± 14.39 1.000

Gastric 37.41 ± 10.75 <0.05 59.68 ± 11.10 0.976 41.91 ± 12.89 <0.05 70.08 ± 14.63 0.054

Lung 40.89 ± 15.59 0.937 61.27 ± 14.57 0.501 48.33 ± 17.96 0.975 81.05 ± 17.98 0.889

Breast 44.05 ± 7.78 1.000 57.84 ± 9.93 0.589 49.94 ± 14.32 1.000 75.83 ± 13.83 1.000

Pancreatic Gastric 40.53 ± 10.63 37.41 ± 10.75 0.641 61.85 ± 10.86 59.68 ± 11.10 0.276 47.56 ± 14.93 41.91 ± 12.89 0.196 76.49 ± 14.39 70.08 ± 14.63 0.073

Lung 40.89 ± 15.59 1.000 61.27 ± 14.57 0.817 48.33 ± 17.96 1.000 81.05 ± 17.98 0.900

Breast 44.05 ± 7.78 0.811 57.84 ± 9.93 0.251 49.94 ± 14.32 1.000 75.83 ± 13.83 1.000

Gastric Lung 37.41 ± 10.75 40.89 ± 15.59 0.949 59.68 ± 11.10 61.27 ± 14.57 0.523 41.91 ± 12.89 48.33 ± 17.96 0.565 70.08 ± 14.63 81.05 ± 17.98 <0.05

Breast 44.05 ± 7.78 0.145 57.84 ± 9.93 0.616 49.94 ± 14.32 0.510 75.83 ± 13.83 0.826

Lung Breast 40.89 ± 15.59 44.05 ± 7.78 0.989 61.27 ± 14.57 57.84 ± 9.93 0.370 48.33 ± 17.96 49.94 ± 14.32 1.000 81.05 ± 17.98 75.83 ± 13.83 1.000

Subtype	Comparison	Center lesions A	P	Peripheral regions A	P	Center lesions V	P	Peripheral regions V	P
Colorectal	Pancreatic	44.29 ± 9.90	40.53 ± 10.63	0.163	59.68 ± 11.10	61.85 ± 10.86	0.220	51.85 ± 14.68	47.56 ± 14.93	0.450	76.48 ± 16.05	76.49 ± 14.39	1.000
	Gastric		37.41 ± 10.75	<0.05		59.68 ± 11.10	0.976		41.91 ± 12.89	<0.05		70.08 ± 14.63	0.054
	Lung		40.89 ± 15.59	0.937		61.27 ± 14.57	0.501		48.33 ± 17.96	0.975		81.05 ± 17.98	0.889
	Breast		44.05 ± 7.78	1.000		57.84 ± 9.93	0.589		49.94 ± 14.32	1.000		75.83 ± 13.83	1.000
Pancreatic	Gastric	40.53 ± 10.63	37.41 ± 10.75	0.641	61.85 ± 10.86	59.68 ± 11.10	0.276	47.56 ± 14.93	41.91 ± 12.89	0.196	76.49 ± 14.39	70.08 ± 14.63	0.073
	Lung		40.89 ± 15.59	1.000		61.27 ± 14.57	0.817		48.33 ± 17.96	1.000		81.05 ± 17.98	0.900
	Breast		44.05 ± 7.78	0.811		57.84 ± 9.93	0.251		49.94 ± 14.32	1.000		75.83 ± 13.83	1.000
Gastric	Lung	37.41 ± 10.75	40.89 ± 15.59	0.949	59.68 ± 11.10	61.27 ± 14.57	0.523	41.91 ± 12.89	48.33 ± 17.96	0.565	70.08 ± 14.63	81.05 ± 17.98	<0.05
	Breast		44.05 ± 7.78	0.145		57.84 ± 9.93	0.616		49.94 ± 14.32	0.510		75.83 ± 13.83	0.826
Lung	Breast	40.89 ± 15.59	44.05 ± 7.78	0.989	61.27 ± 14.57	57.84 ± 9.93	0.370	48.33 ± 17.96	49.94 ± 14.32	1.000	81.05 ± 17.98	75.83 ± 13.83	1.000

Table 2.

Comparison results of the iodine value in five hepatic metastatic tumor subtyes.

subtype	comparison	Center lesions A		P	peripheral regions A		P	Center lesions V		P	peripheral regions V		P
Colorectal	Pancreatic	4.75 ± 2.39	5.89 ± 3.57	0.169	7.21 ± 3.41	10.22 ± 3.98	<0.05	8.86 ± 3.89	10.35 ± 7.65	0.343	14.66 ± 4.77	17.73 ± 10.23	<0.05
	Gastric		6.11 ± 2.41	<0.05		10.61 ± 3.37	<0.05		9.22 ± 8.32	0.318		16.38 ± 8.42	0.146
	Lung		6.36 ± 3.82	0.269		10.32 ± 4.36	<0.05		10.86 ± 4.60	<0.05		18.58 ± 6.11	<0.05
	Breast		6.11 ± 4.57	0.953		9.22 ± 5.13	0.821		10.53 ± 4.79	0.193		17.76 ± 3.81	0.132
Pancreatic	Gastric	5.89 ± 3.57	6.11 ± 2.41	1.000	10.22 ± 3.98	10.61 ± 3.37	1.000	10.35 ± 7.65	9.22 ± 8.32	0.914	17.73 ± 10.23	16.38 ± 8.42	0.528
	Lung		6.36 ± 3.82	1.000		10.32 ± 4.36	1.000		10.86 ± 4.60	0.148		18.58 ± 6.11	0.999
	Breast		6.11 ± 4.57	1.000		9.22 ± 5.13	0.998		10.53 ± 4.79	0.428		17.76 ± 3.81	1.000
Gastric	Lung	6.11 ± 2.41	6.36 ± 3.82	1.000	10.61 ± 3.37	10.32 ± 4.36	1.000	9.22 ± 8.32	10.86 ± 4.60	0.191	16.38 ± 8.42	18.58 ± 6.11	0.422
	Breast		6.11 ± 4.57	1.000		9.22 ± 5.13	0.978		10.53 ± 4.79	0.474		17.76 ± 3.81	0.857
Lung	Breast	6.36 ± 3.82	6.11 ± 4.57	1.000	10.32 ± 4.36	9.22 ± 5.13	0.999	10.86 ± 4.60	10.53 ± 4.79	0.828	18.58 ± 6.11	17.76 ± 3.81	1.000

4. Discussion

Multiphase contrast-enhanced CT is the primary imaging modality for liver tumor differentiation, but its ability to identify primary origins of hepatic metastases remains unclear; our results show limited diagnostic value of mean CT attenuation values in determining metastatic lesion origins due to (1) heterogeneous signal integration from tumor tissue, vascular enhancement, necrosis, and peritumoral inflammation, and (2) resultant obscuring of histology-specific imaging signatures, indicating that advanced quantitative imaging techniques are needed to characterize distinct radiomorphological patterns of metastases from different primary sites.

Recent studies have found that the iodine value of GSI CT can detect slight iodine concentration changes in tumor tissues and reflect the histopathological features more efficiently.^6,14,15 For example, P.J.Lv reported that quantitative analysis of the iodine value could distinguish liver cancer from FNH, liver cancer from hepatic angiomyolipoma, and small hepatocellular carcinoma from small hepatic hemangioma. They also reported that CT spectral imaging enables material decomposition and analysis of several additional quantitative CT imaging parameters—namely, the normalized iodine concentration (NIC) and the lesion-to-normal hepatic tissue ratio (LNR)—that may be used to improve the sensitivity for differentiating hydropericardium hepatitis syndrome (HHs) from hepatocellular carcinoma (HCC) in two-phase scanning. At the same time, it may increase the sensitivity for differentiating small hemangiomas from small HCCs in two-phase scanning.^6,10,11 Iodine value analysis can also be employed to identify stomach, lung, pancreas, mediastinum, cervix, adrenal, neck, prostate, and other benign and malignant tumors.^2,16–24

The spectral curve is the attenuation of matter or structure with the X-ray energy curve. It has been reported that the spectral curve can reveal the characteristics of the material energy attenuation; thus, it reflects the characteristics of the lesion.^12,24 However, at this point, no extensive research has been reported on the role of the spectral curve in the differentiation of hepatic metastases. The slope of the energy spectrum curve can identify the origins of the tumors more accurately. It can semi-quantitatively determine the similarity of lesion components, such as the similarity between metastases and primary lesions. In a limited disease classification, similar spectral curves may reflect the same or similar pathological class. This feature may help to identify the primary sources of the tumors.^2,25–30

Compared with the CT value, the iodine value is a more valuable parameter and can better identify tumors of different origins. The possible reason is that the iodine value can better reflect the tumor's blood supply status than the CT value. During our study, compared with metastases from other sources, the iodine value of liver metastases from colon cancer was significantly different from the iodine value of most other metastatic tumors (except breast cancer). This may be because colon cancers are mainly the epithelium adenocarcinomas of the digestive duct, and most of them enhance with typical features (target signs and bull's eye sign). However, other tumor origins lack such a characteristic. Therefore, the iodine value does not help diagnose other cancer types, such as lung cancer, gastric cancer, pancreatic cancer, and breast cancer. Further investigations may be required.

We found that some atypical metastatic liver tumors were often difficult to distinguish from small metastatic liver tumors (diameter <3 cm). Although the hepatic artery supplies most hepatic metastases, a few small hepatic metastases (diameter <3 cm) are partially supplied by the Portal vein system. As a result, some small metastatic liver tumors do not have the enhancement signs of typical metastatic liver tumors, and metastatic liver tumors of different sources, such as breast cancer and RCC of the same rich blood supply origin, have the best imaging in the AP, showing peripheral ring or mixed enhancement. Colon cancer and gastric cancer were the sources of blood deficiency, and PVP imaging was the best. Metastatic liver tumors of this origin type have uniform enhancement characteristics and are difficult to distinguish. Therefore, traditional CT enhancement still has some defects. In this paper, by comparing and analyzing the characteristics of the CT value, iodine value, and λHU of small metastatic liver tumors from different sources, it was found that quantitative analysis of the CT iodine value could distinguish colorectal cancer versus gastric cancer, rectal cancer versus pancreatic cancer, and rectal cancer versus lung cancer.

In contrast, general CT could only distinguish colorectal cancer versus gastric cancer and gastric cancer versus lung cancer. Comparative analysis with Table 2 results demonstrates the superior diagnostic performance of spectral CT iodine quantification.The λHU is highly consistent with the iodine value (Tables 2 and 3), which further proves its credibility. With the help of the area under the ROC curve, we can conclude that colorectal cancer has significant specificity compared with several other sources of liver metastases. In primary tumor is not clear or no biopsy cases, or those who are not typical, lack of blood supply small metastatic liver tumors conventional CT enhancement is difficult to distinguish.

Table 3.
Comparison results of the energy spectrum curve slope (λHU) in 5 hepatic metastatic tumor subtypes.

Subtype Comparison Center lesions A P Peripheral regions A P Center lesions V P Peripheral regions V P

Colorectal Pancreatic 0.33 ± 0.22 0.43 ± 0.28 0.250 0.42 ± 0.23 0.72 ± 0.31 <0.05 0.53 ± 0.31 0.57 ± 0.36 0.996 0.91 ± 0.38 1.28 ± 1.01 <0.05

Gastric 0.47 ± 0.24 <0.05 0.66 ± 0.30 <0.05 0.61 ± 0.28 0.779 0.98 ± 0.30 0.976

Lung 0.55 ± 0.30 <0.05 0.80 ± 0.37 <0.05 0.79 ± 0.36 <0.05 1.36 ± 0.45 <0.05

Breast 0.31 ± 0.19 1.000 0.59 ± 0.36 0.724 0.73 ± 0.24 0.162 1.16 ± 0.37 0.338

Pancreatic Gastric 0.43 ± 0.28 0.47 ± 0.24 0.989 0.72 ± 0.31 0.66 ± 0.30 0.986 0.57 ± 0.36 0.61 ± 0.28 0.999 1.28 ± 1.01 0.98 ± 0.30 0.211

Lung 0.55 ± 0.30 0.139 0.80 ± 0.37 0.852 0.79 ± 0.36 <0.05 1.36 ± 0.45 1.000

Breast 0.31 ± 0.19 0.478 0.59 ± 0.36 0.912 0.73 ± 0.24 0.462 1.16 ± 0.37 0.997

Gastric Lung 0.47 ± 0.24 0.55 ± 0.30 0.782 0.66 ± 0.30 0.80 ± 0.37 0.360 0.61 ± 0.28 0.79 ± 0.36 0.057 0.98 ± 0.30 1.36 ± 0.45 <0.05

Breast 0.31 ± 0.19 0.165 0.59 ± 0.36 0.999 0.73 ± 0.24 0.808 1.16 ± 0.37 0.717

Lung Breast 0.55 ± 0.30 0.31 ± 0.19 <0.05 0.80 ± 0.37 0.59 ± 0.36 0.515 0.79 ± 0.36 0.73 ± 0.24 0.997 1.36 ± 0.45 1.16 ± 0.37 0.649

Subtype	Comparison	Center lesions A	P	Peripheral regions A	P	Center lesions V	P	Peripheral regions V	P
Colorectal	Pancreatic	0.33 ± 0.22	0.43 ± 0.28	0.250	0.42 ± 0.23	0.72 ± 0.31	<0.05	0.53 ± 0.31	0.57 ± 0.36	0.996	0.91 ± 0.38	1.28 ± 1.01	<0.05
	Gastric		0.47 ± 0.24	<0.05		0.66 ± 0.30	<0.05		0.61 ± 0.28	0.779		0.98 ± 0.30	0.976
	Lung		0.55 ± 0.30	<0.05		0.80 ± 0.37	<0.05		0.79 ± 0.36	<0.05		1.36 ± 0.45	<0.05
	Breast		0.31 ± 0.19	1.000		0.59 ± 0.36	0.724		0.73 ± 0.24	0.162		1.16 ± 0.37	0.338
Pancreatic	Gastric	0.43 ± 0.28	0.47 ± 0.24	0.989	0.72 ± 0.31	0.66 ± 0.30	0.986	0.57 ± 0.36	0.61 ± 0.28	0.999	1.28 ± 1.01	0.98 ± 0.30	0.211
	Lung		0.55 ± 0.30	0.139		0.80 ± 0.37	0.852		0.79 ± 0.36	<0.05		1.36 ± 0.45	1.000
	Breast		0.31 ± 0.19	0.478		0.59 ± 0.36	0.912		0.73 ± 0.24	0.462		1.16 ± 0.37	0.997
Gastric	Lung	0.47 ± 0.24	0.55 ± 0.30	0.782	0.66 ± 0.30	0.80 ± 0.37	0.360	0.61 ± 0.28	0.79 ± 0.36	0.057	0.98 ± 0.30	1.36 ± 0.45	<0.05
	Breast		0.31 ± 0.19	0.165		0.59 ± 0.36	0.999		0.73 ± 0.24	0.808		1.16 ± 0.37	0.717
Lung	Breast	0.55 ± 0.30	0.31 ± 0.19	<0.05	0.80 ± 0.37	0.59 ± 0.36	0.515	0.79 ± 0.36	0.73 ± 0.24	0.997	1.36 ± 0.45	1.16 ± 0.37	0.649

Another noteworthy finding in our study was that when comparing the diagnostic efficacy of hepatic metastases from different primary origins, colorectal cancer liver metastases consistently demonstrated the largest area under the ROC curve (AUC) compared to gastric and lung cancer metastases (Table 4 [0.697, 0.777, and 0.804]). This observation indicates that a larger AUC corresponds to greater diagnostic significance.The possible reason is that compared with other tumors, colorectal cancer liver metastases have different imaging characteristics. The classical “bull eye sign” and “target sign” may reflect these imaging features. The λHU may make it easier to quantitatively distinguish liver metastases from colorectal cancer and other liver metastases. Overall, energy spectrum analysis of metastatic liver tumors with distinct sources may provide some guidance in identifying the nature and origin of the lesions.

Table 4.

ROC analysis of CT value, iodine value, and spectrum curve slope (λHU) in 5 hepatic metastatic tumor subtypes.

Subtype comparison	Parameter	AUC	Cut-off values	Sensitivity (%)	Specificity (%)
CT values
Colorectal cancer vs. gastric cancer	A, center lesions	0.683	0.285	90.6	37.9
	V, center lesions	0.697	0.361	76.4	55.2
Gastric cancer vs. lung cancer	V, peripheral regions	0.692	0.415	93.1	48.4
Iodine values
Colorectal cancer vs. gastric cancer	A, center lesions	0.681	0.368	57.5	79.3
	A, peripheral regions	0.777	0.474	69.8	77.6
Colorectal cancer vs. pancreatic cancer	A, peripheral regions	0.721	0.356	72.6	63.1
	V, peripheral regions	0.653	0.239	62.3	61.6
Colorectal cancer vs. lung cancer	V, center lesions	0.629	0.277	85.8	41.9
	A, peripheral regions	0.703	0.371	72.6	64.5
	V, peripheral regions	0.678	0.328	65.1	67.7
The value of the slope of the spectral curve
Colorectal cancer vs. gastric cancer	A, center lesions	0.671	0.328	80.2	52.6
	A, peripheral regions	0.734	0.452	66.3	78.9
Colorectal cancer vs. lung cancer	A, center lesions	0.717	0.371	75.6	61.5
	V, center lesions	0.723	0.411	82.6	58.5
	A, peripheral regions	0.804	0.502	73.3	76.9
	V, peripheral regions	0.768	0.467	69.8	76.9
Gastric cancer vs. pancreatic cancer	A, peripheral regions	0.565	0.178	42.1	75.7
	V, peripheral regions	0.644	0.292	76.3	52.9
Lung cancer vs. breast cancer	A, center lesions	0.755	0.458	70.8	75.1
Pancreatic cancer vs. lung cancer	V, center lesions	0.684	0.354	60.1	75.4
Gastric cancer vs. lung cancer	V, peripheral regions	0.753	0.532	76.3	76.9

Our study had several limitations. First, the small sample size requires future validation with larger cohorts to strengthen statistical power. Second, the retrospective design introduced inherent selection bias. Third, measurements were influenced by region-of-interest size, necessitating methodological refinement. Fourth, partial lesion sampling rather than comprehensive ROI analysis may have caused sampling error. Additionally, primary tumor classification was limited by insufficient pathological data. Finally, histopathology-CT correlations were excluded as they fell beyond our study scope.

5. Conclusions

In this study, using hepatic metastatic lesion subtype analysis, the mean CT value only detected colorectal from gastric metastasis and lung from gastric metastasis. However, there was no significant difference between other pairs regarding the mean CT value (Table 1). In this study, through comparative iodine value analysis of the 5 common metastatic liver tumors, it was observed that iodine values could be used to identify colorectal cancer versus gastric cancer, colorectal cancer versus lung cancer, and colorectal cancer versus pancreatic cancer (Table 2). And the detailed statistics about the different iodine values across different cancer sites: We concluded that Colorectal cancer VS Cancer of the stomach (Center lesions A Iodine value with :37.41 ± 10.75 mg/mL.), Colorectal cancer VS Pancreatic cancer (peripheral regions A and peripheral regions V, Iodine value with :10.22 ± 3.98 mg/mL and 17.73 ± 10.23 mg/mL.), Colorectal cancer VS Lung cancer (peripheral regions A and peripheral regions V, Iodine value with :10.32 ± 4.36 mg/mL and 18.58 ± 6.11 mg/mL.). Our findings provide a reliable basis for avoiding invasive procedures in tumor patients through spectral CT iodine quantification of small hepatic metastases (<3 cm), particularly facilitating early diagnosis and treatment of metastases from different primary origins, especially the source of colorectal cancer. This study highlights the need to develop more advanced techniques with expanded sample sizes to differentiate small metastases from various primary sites and even distinct pathological subtypes of the same origin. Due to limited sample size, small hepatic metastases of rare subtypes (lacking statistical representativeness and characteristic imaging features) were excluded. Future studies with expanded cohorts and optimized techniques to systematically characterize these lesions will be of significant clinical value.These advancements will establish a critical foundation for breakthroughs in clinical management and early-stage stratification, ultimately improving patient survival outcomes.

In summary, our quantitative analysis of GSI iodine values combined with auxiliary parameters may facilitate the diagnosis of small lesions and help determine the origin of metastatic liver tumors, especially the source of colorectal cancer.

Footnotes

Abbreviations

Ethical considerations and consent to participate

We confirm that this study complies with all regulations. This retrospective study was approved by the institutional review board of Tongji hospital (ID:TJ-IRB20181123), and the requirement for formal informed consent was obtained from all patients.

Consent for publication

All the listed authors have seen and approved the submitted manuscript.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The study was supported by The Chinese National Natural Science Funds (No. 81501447, 81771801) and Hubei Provincial Natural Science Foundation (No. 2025AFD863).

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data availability statement

The datasets analyzed during the current study are available from the corresponding author on reasonable request.

References

Zhou

Meng

Zhang

, et al. Diagnostic value of delayed PET/MR in liver metastasis in comparison with PET/CT. Front Oncol 2021; 11: 717687.

Mahmoudi

Bernatz

Althoff

, et al. Dual-energy CT based material decomposition to differentiate intrahepatic cholangiocarcinoma from hepatocellular carcinoma. Eur J Radiol 2022; 156: 110556.

Wesdorp

Kemna

Bolhuis

, et al. Interobserver variability in CT-based morphologic tumor response assessment of colorectal liver metastases. Radiol Imaging Cancer 2022; 4: e210105.

Lewin

Laurent-Bellue

Desterke

, et al. Evaluation of perfusion CT and dual-energy CT for predicting microvascular invasion of hepatocellular carcinoma. Abdom Radiol (NY) 2022; 47: 2115–2127.

Zhao

Ling

, et al. Dual-Energy computed tomography imaging in early-stage hepatocellular carcinoma: a preliminary study. Contrast Media Mol Imaging 2022; 2022: 2146343.

Foti

Ascenti

Agostini

, et al. Dual-Energy CT in oncologic imaging. Tomography 2024; 10: 299–319.

Sun

, et al. Differentiating hepatocellular carcinoma from angiomyolipoma of the liver with CT spectral imaging: a preliminary study. Clin Radiol 2013; 68: e491–e497.

Suh

Baek

Choi

, et al. Performance of CT in the preoperative diagnosis of cervical lymph node metastasis in patients with papillary thyroid cancer: a systematic review and meta-analysis. AJNR Am J Neuroradiol 2017; 38: 154–161.

Guo

, et al. Spectral CT imaging in the differential diagnosis of necrotic hepatocellular carcinoma and hepatic abscess. Clin Radiol 2014; 69: e517–e524.

10.

Elbanna

Mansoori

Mileto

, et al. Dual-energy CT in diffuse liver disease: is there a role. Abdom Radiol (NY) 2020; 45: 3413–3424.

11.

Lin

, et al. Differentiation of small hepatic hemangioma from small hepatocellular carcinoma: recently introduced spectral CT method. Radiology 2011; 259: 720–729.

12.

Wang

Shi

, et al. Quantitative analysis of the dual-energy CT virtual spectral curve for focal liver lesions characterization. Eur J Radiol 2014; 83: 1759–1764.

13.

Lin

Chen

, et al. Hepatocellular carcinoma and focal nodular hyperplasia of the liver: differentiation with CT spectral imaging. Eur Radiol 2013; 23: 1660–1668.

14.

Chamroukhi

Brivet

Savadjiev

, et al. DECT-CLUST: dual-energy CT image clustering and application to head and neck squamous cell carcinoma segmentation. Diagnostics (Basel) 2022; 12: 3072.

15.

Han

Sun

, et al. Application of contrast-enhanced dual-energy spectral CT for differentiating borderline from malignant epithelial ovarian tumours. Clin Radiol 2021; 76: 585–592.

16.

Zhao

Huang

, et al. Dual-Energy CT-based nomogram for decoding HER2 Status in patients with gastric cancer. AJR Am J Roentgenol 2021; 216: 1539–1548.

17.

Luo

Sha

, et al. Differentiation of malignant from benign orbital tumours using dual-energy CT. Clin Radiol 2022; 77: 307–313.

18.

Taudorf

Ulriksen

, et al. Gastrointestinal applications of iodine quantification using dual-energy CT: a systematic review. Diagnostics (Basel) 2020; 10: 814.

19.

Ren

Yang

. Value of dual-source CT dual-energy parameters combined with serum detection of VEGF and CEA in the diagnosis of early lung cancer. Biotechnol Genet Eng Rev 2023; 39: 1000–1011.

20.

Yuan

Quan

Che

, et al. Preoperative prediction of the lymphovascular tumor thrombus of colorectal cancer with the iodine concentrations from dual-energy spectral CT. BMC Med Imaging 2023; 23: 103.

21.

Kim

Park

, et al. Application of dual-energy spectral computed tomography to thoracic oncology imaging. Korean J Radiol 2020; 21: 838–850.

22.

Meng

Shen

, et al. Differentiating malignant from benign gastric mucosal lesions with quantitative analysis in dual energy spectral computed tomography: initial experience. Medicine (Baltimore) 2017; 96: e5878.

23.

Zhu

Zheng

Jiang

, et al. Evaluation of dual-energy and perfusion CT parameters for diagnosing solitary pulmonary nodules. Thorac Cancer 2021; 12: 2691–2697.

24.

Lyu

Chen

, et al. Deep learning reconstruction CT for liver metastases: low-dose dual-energy vs standard-dose single-energy. Eur Radiol 2024; 34: 28–38.

25.

Wang

Zhang

, et al. Progress in the application of dual-energy CT in pancreatic diseases. Eur J Radiol 2023; 168: 111090.

26.

Kim

Cha

, et al. Differentiating pulmonary metastasis from benign lung nodules in thyroid cancer patients using dual-energy CT parameters. Eur Radiol 2022; 32: 1902–1911.

27.

Liang

Zhou

Zheng

, et al. Dual-energy CT with virtual monoenergetic images and iodine maps improves tumor conspicuity in patients with pancreatic ductal adenocarcinoma. Insights Imaging 2022; 13: 153.

28.

Badnjević

Cifrek

Koruga

. Integrated software suite for diagnosis of respiratory diseases[C]//eurocon 2013. IEEE 2013; 564–568.

29.

Veljović

Špirtović-Halilović

Muratović

, et al. Artificial neural network and docking study in design and synthesis of xanthenes as antimicrobial agents[C]. In: CMBEBIH 2017: proceedings of the international conference on medical and biological engineering 2017. Singapore: Springer, 2017, pp.617–626.

30.

Badnjevic

Cifrek

Koruga

. Classification of chronic obstructive pulmonary disease (COPD) using integrated software suite[C]. In: XIII Mediterranean conference on medical and biological engineering and computing 2013: MEDICON 2013, 25–28 September 2013. Seville, Spain: Springer International Publishing, 2014, pp.911–914.