Clinical Benefit of Structured Reporting Using the Liver Imaging Reporting and Data System (LI-RADS) in MRI in Patients at Risk for Hepatocellular Carcinoma

Abstract

Purpose

The aim of this study was to evaluate the quality of structured reporting of MRI examinations in patients at risk for HCC using the Liver Imaging Reporting and Data System (LI-RADS) created during the daily clinical routine.

Materials and Methods

In this retrospective study, MRI examinations of 163 patients under HCC surveillance and HCC follow-up were analyzed. The study cohort included a group of 76 patients whose examinations were performed using free-text reporting and a group of 87 patients with structured template reporting based on the LI-RADS criteria. The diagnostic accuracy of the specified LI-RADS category was analyzed for both groups. The impact of structured reporting and the frequency of reporting of the major and ancillary LI-RADS features in free-text reports and structured reports were compared using the χ²-test.

Results

Liver lesions were classified according to LI-RADS significantly more often in the structured reports (91.4%) than in the unstructured free-text reports (82.6%) (p < 0.01). Most relevant major LI-RADS criteria were described significantly more frequently when using the structured report template compared to free text, e.g., arterial hyperenhancement 100% vs. 92.5% (p < 0.01) and washout 93.4% vs. 79.7% (p < 0.01). Only the documentation of threshold growth was not significantly improved, but absolute lesion size was reported significantly more often within the structured reports. The diagnostic accuracy of the LI-RADS category was higher for the structured reports than for the free-text reports (82.3% vs. 63.9%).

Conclusion

Structured reporting improves LI-RADS documentation in MRI reports of patients at risk for HCC and may help with multidisciplinary communication and patient management.

Key Points

• Structured reporting showed improved documentation of key features for LI-RADS classification compared to nonstructured MRI reports.

• Structured reports improve the categorization of liver lesions according to LI-RADS in patients at risk for HCC.

Citation Format

• Nelles C, Wagner A, Lennartz S et al. Clinical Benefit of Structured Reporting Using the Liver Imaging Reporting and Data System (LI-RADS) in MRI in Patients at Risk for Hepatocellular Carcinoma. Rofo 2025; DOI 10.1055/a-2553-1392

Zusammenfassung

Ziel

Ziel dieser Studie war es, die Qualität der strukturierten Befundung in der klinischen Routine von MRT-Untersuchungen bei Patienten mit Risiko für ein hepatozelluläres Karzinom (HCC) unter Verwendung des Liver Imaging Reporting and Data System (LI-RADS) zu bewerten.

Material und Methoden

In dieser retrospektiven Studie wurden MRT-Untersuchungen von 163 Patienten in der HCC-Überwachung und der HCC-Nachbeobachtung analysiert. Die Studienkohorte umfasste eine Gruppe von 76 Patienten, deren Untersuchungen mit Freitextbefunden beurteilt wurden, und eine Gruppe von 87 Patienten, deren Untersuchungen mit strukturierten Befundvorlagen auf der Grundlage der LI-RADS-Kriterien bewertet wurden. Die diagnostische Genauigkeit der angegebenen LI-RADS-Kategorie wurde für beide Gruppen berechnet. Die Auswirkungen der strukturierten Befundung und die Häufigkeit der Angabe der Haupt- und Nebenmerkmale der LI-RADS-Kriterien in Freitextbefunden und strukturierten Befunden wurden mithilfe des χ²-Tests verglichen.

Ergebnisse

Die Klassifizierung von Leberläsionen nach LI-RADS wurde in strukturierten Befunden mit 91,4% signifikant häufiger vorgenommen als in Freitextbefunden mit 82,6% (p<0,01). Die meisten relevanten LI-RADS-Hauptkriterien wurden bei Verwendung der strukturierten Befundvorlage signifikant häufiger beschrieben als in Freitextbefunden, z.B. arterielles Hyperenhancement (100% gegenüber 92,5%, p<0,01) und Wash-Out (93,4% gegenüber 79,7%, p<0,01). Lediglich die Dokumentation des Schwellenwachstums wurde nicht signifikant verbessert, jedoch wurde die absolute Größe der Läsionen signifikant häufiger in den strukturierten Befunden angegeben. Die diagnostische Genauigkeit war bei den strukturierten Befunden höher als bei den Freitextbefunden (82,3% vs. 63,9%).

Schlussfolgerung

Eine strukturierte Befundung verbessert die LI-RADS-Dokumentation in MRT-Befunden von Patienten mit Risiko für ein HCC und kann die interdisziplinäre Kommunikation und das Patientenmanagement unterstützen.

Kernaussagen

• Strukturierte Befundung zeigte eine verbesserte Dokumentation der Hauptmerkmale für die LI-RADS-Klassifizierung im Vergleich zu nicht strukturierten MRT-Freitextbefunden.

• Strukturierte Befunde verbessern die Kategorisierung von Leberläsionen gemäß LI-RADS bei Patienten mit Risiko für ein HCC.

Keywords

structured reporting - oncology - hepatocellular cancer (HCC) - Liver Imaging Reporting and Data System (LI-RADS)

Abbreviations

ACR: American College of Radiology

APHE: Arterial phase hyperenhancement

BCLC: Barcelona-Clinic Liver Cancer Staging System

CT: Computed tomography

DCE: Dynamic contrast enhancement

DWI: Diffusion-weighted imaging

EASL: European Association for the Study of the Liver

HBV: Hepatitis B virus

HCC: Hepatocellular cancer

HCV: Hepatitis C virus

LI-RADS: Liver Imaging Reporting and Data System

MRI: Magnetic resonance imaging

OLT: Orthotopic liver transplant

RIS: Radiological information system

TACE: Transarterial chemoembolization

TARE: Transarterial radioembolization

TG: Threshold growth

TIV: Tumor in vein

US: Ultrasound

WO: Washout

Introduction

Hepatocellular carcinoma (HCC) is the fifth most frequent malignancy and the fourth leading cause of cancer-related deaths worldwide [1] [2]. Despite advancements in treatment options, the overall prognosis of patients with advanced HCC remains poor, while patients who are diagnosed in the early stages of HCC have a better probability of survival due to eligibility for curative treatments such as surgical resection, local ablation, transarterial embolization, and liver transplantation [2]. Therefore, accurate noninvasive detection and characterization of early-stage HCC is prognostically pertinent and an important challenge for liver imaging using magnetic resonance imaging (MRI), computed tomography (CT), and ultrasound (US).

HCC is most frequent in patients with chronic inflammation of the liver or liver cirrhosis. Liver injury can be caused by chronic infections such as hepatitis B virus and hepatitis C virus infection (HBV and HCV), chronic intoxication, e.g. alcohol abuse, or other conditions like fatty liver disease or autoimmune diseases. US is the mainstay of surveillance in screening programs for patients at risk for HCC. Currently, the European Association for the Study of the Liver (EASL) recommends US examinations at 6-month intervals for patients at risk for HCC based on cost analysis [2]. However, US is examiner-dependent and is less effective in obese patients or in patients with severe liver steatosis. MRI is the most sensitive imaging modality for screening the entire liver and detecting early HCC with reported sensitivities ranging from 85–100% [2] [3] [4].

The Liver Imaging Reporting and Data System (LI-RADS) was introduced in 2011 by the American College of Radiology (ACR). It aims to standardize radiology review, thereby improving noninvasive lesion characterization and optimizing reporting of MRI, CT and US liver examinations in patients at high risk for HCC. Since its conception in 2011, several updates of LI-RADS have been published, the latest version having been published in 2018 [5]. LI-RADS defines terms for lesion characterization and provides an algorithm to categorize liver lesions according to their likelihood of HCC. Variation regarding the interpretation of liver lesions is thus supposed to be reduced and the communication between radiologists and clinicians is supposed to be improved. LI-RADS uses five major criteria to categorize liver lesions. Arterial phase hyperenhancement (APHE), observation size (in mm), enhancing capsule, non-peripheral washout (WO), and threshold growth (TG). Findings are subdivided into definitely benign (LR-1), probably benign (LR-2), intermediate probability of malignancy (LR-3), probably HCC (LR-4), and definitely HCC (LR-5). In addition, ancillary features can be used for further differentiation. Due to this distinct classification system, LI-RADS helps to reduce ambiguity in imaging reports [5].

The radiological report is the key component in the communication between radiologists and referring clinicians [6]. Traditionally, reports are written as free text. Several studies have shown that structured reporting using dedicated report templates has a number of advantages compared to conventional reporting and is accepted by radiologists [7] [8] [9] [10] [11]. Therefore, many radiological societies have recommended clinical and technical implementation of structured reporting in the daily routine [6] [12] [13]. Based on these results, we hypothesized that structured reporting may be beneficial for assessing liver MRI examinations of patients at risk for HCC, particularly with regard to the documentation of LI-RADS features.

Therefore, the aim of this study was to evaluate the quality of structured reporting for MRI examinations in patients at risk for HCC by comparing LI-RADS to free-text reports in the daily routine.

Materials and methods

Patient selection

This retrospective, single-center study was approved by the institutional review board (IRB) and informed patient consent was waived due to the retrospective character of the study. The clinical database of our radiological information system (Orbis RIS, Dedalus, Belgium) was used to screen for patients at risk for HCC who underwent liver MRI scan at 1.5 and 3.0 Tesla (Achieva and Ingenia, Philips, The Netherlands) with extracellular contrast agent (Gd-DOTA; 0,1 mmol/kg body weight) from August 2013 until November 2017 that yielded definite and probable HCC or lesions that required follow-up. However, avital liver lesions after treatment, such as after microwave ablation, transarterial chemoembolization (TACE), or transarterial radioembolization (TARE), were excluded. Balanced distribution of the senior physicians was ensured to avoid a bias regarding liver imaging experience.

Free-text report

In structured and free-text reports, lesions were determined as definite HCC if they were described with the following terms: imaging-based HCC, definite HCC, typical HCC = LI-RADS 5, and they were determined as “probable” when described as “suspicious for HCC, possible HCC = LI-RADS 4”. Lesions that needed follow-up were included if they were described with the following terms: dysplastic nodule, inflammatory nodule, regenerative nodule, lesion requiring monitoring, or unclear lesion = LI-RADS 3.

Structured template

LI-RADS was introduced to our department in 2013 and was used between August 2013 and November 2017 in versions LI-RADS v2013.1 and v2014. After November 2017, the structured report template was updated to LI-RADS v2017. Thus, the inclusion time was set from 2013 to 2017.

The structured reporting template for HCC diagnosis on MRI that was analyzed in this study was introduced to clinical routine reporting in our institution in June 2016 ([Fig. 1]). It is accessible online for residents and staff radiologists of our institution, who were instructed on how to use the template and attended a special teaching course on liver imaging. On the internal institutional homepage, the structured reporting template is accompanied by the most recent update of LI-RADS criteria for lesion characterization as well as additional information regarding the definition of the Milan criteria, treatment response algorithm, visualization of liver segments, and a link to the LI-RADS manual on the ACR homepage for further information.

Fig. 1 Structured reporting template for MR imaging in patients at risk for HCC in German a and English b. *Please note: for liver lesions, threshold growth was not mentioned during the study period and was included as a consequence of this analysis.

The structured MRI report is consistently used in patients that are at high risk for HCC, such as patients with chronic HBV/HCV, alcohol-induced liver cirrhosis, or autoimmune diseases, starting from June 2016.

Each suspicious lesion is listed separately in the structured template report and is categorized by the following features: liver segment, diameter, series and image number in which it was observed, signal intensity in T2w, T1w, and diffusion-weighted imaging (DWI), dynamic contrast enhancement (DCE), washout, pseudo-capsule, macrovascular infiltration and LI-RADS categorization. Next, liver vessels, gall bladder and intra- and extrahepatic bile ducts, ascites, and lymph nodes are assessed. All other abdominal organs as well as bones are then also evaluated.

Analysis of free-text vs. structured reports

We divided HCC free-text reports and HCC structured template reports into two separate groups. It was observed whether the following criteria were reported unequivocally: Couinaud segment, major LI-RADS criteria (diameter, presence/absence of the following: non-rim arterial phase hyperenhancement (APHE), threshold growth (TG), non-peripheral washout (WO), enhancing capsule), presence/absence of macrovascular infiltration, presence/absence of ancillary features, LI-RADS category, and definite diagnosis ([Fig. 2]).

Fig. 2 Typical hepatocellular carcinoma (HCC) on MRI. Upper row: Major features on dynamic contrast-enhanced (DCE) T1wFS MRI with non-rim arterial phase hyperenhancement (arrow), washout (asterisk; compared to the adjacent liver tissue), and delayed pseudo-capsule (arrow). Lower row, left images: Threshold growth >50% within 6 months in arterial phase (arrow head). Lower row, right image: Exemplary ancillary feature such as moderate T2w signal hyperintensity.

Consistency between LI-RADS category and ground truth

For both groups, the correspondence between the classified LI-RADS category and the ground truth was analyzed. As corresponding histology was only rarely available (for LI-RADS 4 and 5 lesions: 45.5% in the structured reports group and only 24.3% in the free-text reports group, which confirmed HCC in all cases; for LI-RADS 3 lesions, in both groups, there were no findings that could be correlated unequivocally), the imaging-based ground truth was determined independently by two board-certified radiologists, each with 6 years of experience in oncologic imaging. In the event of differing LI-RADS scores, the final LI-RADS score was determined by a senior consultant radiologist with more than 20 years of experience in oncologic imaging. Additionally, for both groups, the correspondence between the reported features for LI-RADS categorization and the final classified LI-RADS score was analyzed for each reported lesion. For reports in which the explicit LI-RADS category was not given, it was determined on the basis of the linguistic coding described earlier (see the paragraph “free-text report”).

Statistical analysis

Medcalc (GraphPad Software, San Diego, California, USA) was used for statistical analysis. Statistical significance was set at p-value < 0.05. The frequencies of reporting the defined criteria in free-text reports and structured LI-RADS template reports were compared using the X²-test [14] [15]. Continuous variables were compared between the groups using the Wilcoxon rank sum test.

Results

Patients

The study included MRI examinations of a total of 163 patients at risk for HCC; a group of 76 patients with free-text reporting (mean age: 66.6 ± 11.2 years) and a group of 87 patients with clinical structured template reporting (mean age: 64.7 ± 11.9 years). For both groups, the gender distribution favored male patients, with 80.3% of male patients in the free-text reporting subgroup and 79.3% in the structured template subgroup. Apart from that, the patient characteristics are balanced and are shown in [Table 1]. The cause of liver disease is given in [Table 2].

Table 1 Demographic information of the included patients at risk for HCC.

	Free text	structured Template	p-value
Sex
male	61/76 (80.3%)	69/87 (79.3%)	0.8744
Female	15/76 (19.7%)	18/87 (20.7%)	0.8744
Age	66.6 ± 11.2	64.7 ± 11.9	0.31

Table 2 Cause of liver disease in the study participants.

Cause of HCC	Free text	structured Template	p-value
HBV	14/76 (18.4%)	9/87 (10.3%)	0.1771
HCV	23/76 (30.3%)	25/87 (28.7%)	0.8644
HBV & HCV	1/76 (1.3%)	1/87 (1.1%)	1
alcoholic cirrhosis	11/76 (14.5%)	12/87 (13.8%)	1
alcoholic cirrhosis & HCV	1/76 (1.3%)	3/87 (3.4%)	0.6238
non-alcoholic fatty liver disease	7/76 (9.2%)	8/87 (9.2%)	1
autoimmune hepatitis	1/76 (1.3%)	0/87 (0%)	0.4663
porphyria	1/76 (1.3%)	0/87 (0%)	0.4663
hemochromatosis	0/76 (0%)	3/87 (3.4%)	0.2488
overlap syndrome	0/76 (0%)	1/87 (1.1%)	1
not otherwise specified	17/76 (22.4%)	25/87 (28.7%)	0.3752

Cohorts of 90 patients were selected for the free-text and structured reporting cohort, but 14 patients were excluded from the free-text cohort and 3 patients were excluded from the structured reporting group due to the fact that the reports were not focused on HCC, e.g., in the case of synchronous renal cancer or unclear findings, or in the case of relevant errors of the report, e.g., errors made by the speech recognition system.

Analysis of structured vs. free-text reporting

Lesion location and size

The bivariate comparison of all assessed items of free-text reporting and structured template reporting is summarized in [Table 3]. The Couinaud segment was reported in 264 of 266 lesions in structured template reporting and in 230 of 241 lesions for free-text reporting (99.2% vs 95.4%; p < 0.05).

Table 3 Overview of the clinically relevant features reported in free-text and structured MRI reports.

	Free text	structured Template	p-value
Segment reported	230/241 (95.4%)	264/266 (99.2%)	0.0074
Size reported	183/241 (75.9%)	261/266 (98.1%)	< 0.0001
Threshold growth reported	134/156 (85.9%)	162/202 (80.2%)	0.1581
AP enhancement reported	223/241 (92.5%)	266/266 (100%)	< 0.0001
Washout reported	177/222 (79.7%)	227/243 (93.4%)	< 0.0001
Capsule reported	39/222 (17.6%)	90/241 (37.3%)	< 0.0001
Vascular invasion reported	148/241 (61.4%)	264/266 (99.2%)	< 0.0001
Ancillary features reported (all lesions)	120/241 (49.8%)	158/256 (61.7%)	0.0076
Ancillary features reported (LI-RADS 3/4 lesions)	52/104 (50%)	73/142 (51.4%)	0.8286
LI-RADS reported	53/241 (22%)	243/266 (91.4%)	< 0.0001
Unambiguous diagnosis	199/241 (82,6%)	263/266 (98.9%)	< 0.0001

LI-RADS major features

Regarding the major LI-RADS features, lesion size was reported significantly more often in structured reports (261 of 266 lesions) than in free-text reporting (183 of 241 lesions) (98.1% vs. 75.9%; p < 0.0001). In contrast, threshold growth was reported more often in free-text reports with a total of 134 of 156 lesions (85.9%) whereas structured templates reported only 162 of 202 lesions (80.2%) with threshold growth, yet this difference was nonsignificant (p = 0.15). Non-rim arterial phase hyperenhancement (APHE) was reported more frequently in structured template reporting (266/266) than in free-text reporting (223 of 241 lesions) (100% vs. 92.5%; p < 0.0001). Non-peripheral washout (WO) was unequivocally mentioned within the structured template in 227 of 243 lesions compared to 177 of 222 lesions in the free-text reports (93.4% vs. 79.7%; p < 0.0001). The major feature “presence/absence of enhancing capsule” was the least reported feature overall. However, the mentioning of “enhancing capsule” improved considerably from only 39 of 222 lesions (17.6%) in the free-text reports to 90 of 241 lesions (37.3%) in the structured template (p < 0.0001). As an additional feature, the presence or absence of macrovascular invasion “tumor in vein” (TIV) was reported significantly more often using structured reports (264/266 lesions) than in the free-text reports (148/241 lesions) (99.2% vs. 61.4%; p < 0.0001).

LI-RADS ancillary features

The presence or absence of ancillary LI-RADS features was reported less frequently in free-text reports with a total of 120 of 241 lesions (49.8%) compared to structured template reports with a total of 158 of 256 lesions (61.7%; p < 0.05). However, considering that ancillary features can only up- and downgrade lesions with a LI-RADS category from 2 to 4, there was no statistically significant difference between free-text and structured template reporting (p > 0.05) for the reporting of ancillary features in liver lesions with an intermediate probability of malignancy (LI-RADS 3) and probable HCC (LI-RADS 4). In this subgroup, ancillary features were reported in 52 of 104 lesions (50%) with free text and in 73 of 142 lesions (51.4%) with the structured template.

LI-RADS category and clarity of diagnosis

Based on the major LI-RADS criteria as well as the ancillary features, the final LI-RADS category was only specified in 53 of 241 lesions (17.4%) in free-text reporting compared to 243 of 266 lesions (91.4%) in structured template reports (p < 0.0001). However, due to the fact that the LI-RADS categorization was only used by some radiologists specializing in the liver at the beginning of the implementation of LI-RADS into the clinical routine and before the introduction of structured template reporting, the study also aimed to examine the frequency of distinct diagnostic statements in the final radiologist report. The LI-RADS structured template led to an unambiguous diagnosis in 263 of 266 lesions (98.9%) compared to 199 of 241 lesions (82.6%) using free-text reporting (p < 0.0001).

Consistency between LI-RADS category and ground truth

During this independent second read of the MRI examinations, an additional 8 lesions had to be excluded from the free-text group, as the review of the MRI examinations revealed that one or more contrast-enhanced sequences could not be adequately assessed due to insufficient image quality or artifacts (e.g., breathing artifacts in the arterial phase with regard to the assessment of arterial hyperenhancement or in the delayed phase with regard to the diagnosis of washout). No lesion had to be additionally excluded from the structured reports group. The reported LI-RADS category was consistent with the ground truth in 82.3% of cases (219/266) for the structured reports, but only in 63.9% of cases (149/233) for the free-text reports. The LI-RADS category was more frequently underestimated in the free-text group compared to the structured reports group (25.3% (59/233) vs. 9.8% (26/266)) ([Fig. 3]a). Moreover, the reported features in the findings section were in accordance with the LI-RADS score in the impression section in 80.8% of the structured reports (215/266), but only in 55.8% of the free-text reports (130/233) ([Fig. 3]b).

Fig. 3 a Correspondence between specified LI-RADS category in the impression section and imaging-based ground truth for free-text and structured reports. b Correspondence between reported features for LI-RADS classification and specified LI-RADS category in the impression section for free-text and structured reports.

Discussion

Our study aimed to compare structured reporting with free-text reports in the assessment of patients at risk for HCC. We found a significantly higher proportion of reported HCC features when using structured templates, leading to a significantly higher percentage of reported final LI-RADS categories and unambiguous diagnoses. The results we found support the existing scientific base of evidence that structured LI-RADS template reporting improves the accuracy and comprehensibility of radiological MRI reports in patients under HCC surveillance or HCC follow-up examinations compared to free-text reporting.

The major HCC criteria size, non-rim APHE, non-peripheral WO and enhancing capsule as well as the Couinaud segment, and final LI-RADS category were mentioned significantly more frequently in structured LI-RADS template reports in comparison to free-text reports in the daily routine. The documentation of ancillary LI-RADS features improved as well but was included equally for relevant lesions with up- or downgrading (LI-RADS 3 and 4). This discrepancy is explained by the fact that ancillary features were more frequently included in structured reports than in free-text reports for LI-RADS 5 lesions. Interestingly, TG was the only major LI-RADS feature mentioned less frequently in the structured LI-RADS template reports, most likely due to the fact that TG was not queried separately in the structured report template used in the study inclusion interval. However, the measured size of the lesions was mentioned significantly more frequently in the structured reports. Hence, changes in lesion size were in fact reported in the structured reports yet were not attributed to the term “threshold growth”. In a recent study, van der Pol et al. performed a systematic review and an individual patient data meta-analysis to analyze the probability of HCC for each LI-RADS major feature on CT and MRI [16]. In this multivariable analysis of CT/MRI LI-RADS, all major features were associated with HCC except for threshold growth (OR, 1.6; 95% CI: 0.7, 3.6; p = 0.07). APHE and non-peripheral WO demonstrated a strong association with HCC and these major features were significantly more frequently reported in our structured reports. Ultimately, structured reporting provided definite characterization based on LI-RADS and diagnostic statements in the final assessment of all findings more consistently than free-text reporting, which is in line with the results we found.

The highest discrepancy between structured and free-text reporting was observed in the final LI-RADS category, which was mentioned in the majority of the cases for structured template reporting (91.4%), yet was only reported by name in a fraction of the cases in free-text reporting (22.0%). This difference could be partially due to the fact that most radiologists were not familiar with and confident using LI-RADS before the introduction of the structured reporting for HCC. In this context, the LI-RADS category was mostly observed in free-text reports which were signed by radiologists specialized in liver imaging. Nevertheless, unambiguous descriptive characterization of the liver lesions in patients at risk for HCC was given within free-text reporting in 82.6% of cases but was significantly lower compared to structured reporting.

Recent publications have shown that structured reports are superior to traditional free-text reports in terms of completeness, clarity, comprehensibility, and quality [8] [10] [17] [18] [19] [20] [21] [22] [23]. In a clinical evaluation of rectal cancer, Sahni et al. showed that the introduction of structured report templates increased the proportion of reports rated as good or optimal from 38% to 70% [17]. Similarly, Nörenberg et al. found that 98% of structured reports contained all the information necessary for surgical planning in rectal cancer, whereas only 38% of free-text reports did [18]. Analogous results were also found for pancreatic cancer, with structured reports containing significantly more relevant information than free-text reports [10] [23]. In a study by Brook et al., the clarity of findings was independently assessed by 3 surgeons, with structured findings being rated as clearer by all 3 surgeons at 94%, 60%, and 98%, respectively, in contrast to nonstructured free-text reports at 47%, 54%, and 32%, respectively [10]. When asked whether the surgeon had sufficient information to plan surgery and assess resectability, the structured findings yielded significant advantages with 96%, 69%, and 98% compared with only 31%, 43%, and 25% for the free-text findings. Thus, structured reporting of findings is clearly preferred by referring clinicians for differential diagnosis and the staging of tumors, which additionally contributes to improved interdisciplinary communication [24].

In agreement with this, our results demonstrate that structured reporting using LI-RADS enhances the completeness, clarity, and comprehensibility of reports when assessing HCC compared to free-text reporting. We observed a significant improvement in the documentation of the relevant imaging features for LI-RADS classification and diagnosis of HCC in structured reports. Furthermore, consistency between the reported features and the final LI-RADS category was much higher for the structured reports (80.8%) compared to the free-text reports (only 55.8%). Flusberg et al. investigated the quality of structured template reports for LI-RADS in CT and MRI [7]. They included 269 patients with 306 definite or probable HCCs (LR-5 and LR-4). The LI-RADS category was reported in only 18.4% of HCCs when using free-text report, but in 98.3% of HCCs within the structured template (p < 0.001), which is in concordance with our findings. Similarly, they reported that all major features for the diagnosis of HCC improved when comparing free-text reporting to structured reporting: APHE 80.8% vs. 97.8%, WO 74.4% vs. 98.3%, enhancing capsule 19.2% vs. 97.2%, diameter 94.4 vs. 100%, and TG 28.8% vs. 93.4% (p < 0.001). These findings were in agreement with our findings, except for in the case of TG, for which we did not find statistically significant differences. This demonstrates that the inclusion of all relevant findings as a checkbox in the structured reporting template is beneficial and directly improves the quality of the radiological report.

Moreover, while the aforementioned study only focused on the comparison of the content of the reports and the imaging features mentioned therein, we were able to demonstrate the clinical impact by showing that the reported LI-RADS score was much more accurate for structured reports than for free-text reports when compared to the imaging-based ground truth (82.3% vs. 63.9%). This result clearly highlights the benefits and importance of structured reporting in clinical practice. However, it must be pointed out that the focus of our study, just like that of the study by Flusberg et al., was the evaluation of the question of how best to report liver MRI examinations and not the diagnostic accuracy of the LI-RADS system itself. This was beyond the aim of our study. There are already a large number of recent studies regarding this issue that performed an in-depth analysis of the accuracy of the LI-RADS categories and the influence of imaging features on the LI-RADS scoring system [25] [26] [27] [28] [29]. In this study, the presence of histopathological findings was not an inclusion criterion, as they were only available for a small number of lesions (for LI-RADS 4 and 5 lesions: 45.5% in the structured reports group and only 24.3% in the free-text reports group; no unequivocal findings for LI-RADS 3 lesions in both groups). However, HCC was confirmed by histopathological findings for all LI-RADS 4 and 5 lesions for which they were available. This could correspond to the high proportion of HCCs in LI-RADS 4 (74%) and of LI-RADS 5 lesions (95%) reported in the literature and the LI-RADS guidelines [30].

To the best of our knowledge, no previous study has synoptically analyzed reporting of all major LI-RADS criteria, ancillary LI-RADS features, and vascular infiltration (TIV) in structured template reports and free-text reports of HCC surveillance on MRI. Moreover, besides definite and probable HCCs (LR-5 and LR-4), we also assessed lesions with an intermediate probability of malignancy (LR-3) on MRI. Recently, Kierans et al. confirmed the high diagnostic performance of LI-RADS v2017 and v2018 on MRI with a constant specificity of 97.3% for LR-5 and LR-TIV+5, thus highlighting the clinical relevance of LI-RADS for the noninvasive diagnosis of HCC on MRI [31].

In addition, a study by Kabadi et al. showed that different interpretations of free-text reports can occur in approximately 12% of cases when individually different terms are used [32]. The use of uniform language terms and classification systems, such as LI-RADS, with defined imaging criteria could reduce these limitations. Recently, Shenoy-Bhangle et al. reviewed the improvements using LI-RADS for lesion characterization and treatment response assessment at HCC multidisciplinary conferences [33]. Structured template reporting simplifies information extraction for the clinician and facilitates further therapeutic planning, which, for instance, has been proven for the determination of patient suitability to undergo liver transplantation according to the Milan criteria or the resectability of pancreatic cancer [10] [34]. Poullos et al. evaluated the use of a standardized radiology report template to improve the ability of liver transplant surgeons to diagnosis HCC stage T2 and to determine patient suitability to undergo orthotopic liver transplant (OLT) [34]. Structured reporting increased the percentage of reports documenting the presence or absence of portal hypertension from 74% to 88% for surgeon 1 and 86% to 87% for surgeon 2 (p = 0.042), lesion number from 76% to 88% for surgeon 2 (p = 0.038), lesion size from 95% to 96% for surgeon 1 and 82% to 93% for surgeon 2 (p = 0.03), and enhancement from 93% to 94% for surgeon 1 and 80% to 91% for surgeon 2 (p = 0.049). Moreover, in this study, the Organ Procurement and Transplantation Network (OPTN) class for OLT dramatically increased from 8% to 82% for surgeon 1 and from 2% to 81% for surgeon 2 (p < 0.001). Hence, surgeons were better able to determine the presence of T2 disease and qualification for exception points after implementation of the structured template. In line with an earlier study by Schwartz et al. [21], satisfaction with the reports clearly improved as well (p < 0.0001). Thus, they concluded that structured reporting improved determination of patient suitability to undergo transplant according to the Milan criteria. Moreover, according to the internationally accepted Barcelona-Clinic Liver Cancer staging system (BCLC), an individual treatment recommendation requires clear information regarding the number of HCC lesions, size, portal invasion, and/or extrahepatic spread [35], which will be provided by using the structured LI-RADS report.

There are limitations of our study that need to be addressed. First, the structured reporting template was developed by our structured reporting team and was not based on an interdisciplinary consensus. This might have led to the lack of integration of a query for threshold growth (TG). Second, in our radiological information system (RIS), the structured reporting template could be modified freely, it was not mandatory to enter the requested information, and deletion of specific information was easily possible. We assume that the completeness and quality could be further optimized by using dedicated oncological software solutions with required inputs. Finally, at the time of implementation of LI-RADS, only experienced liver radiologists used it consistently in their clinical reports, whereas non-specialized radiologists tended to avoid using LI-RADS, which might have skewed our results. However, the introduction of changes in the clinical routine and personal styles of reporting typically needs time and training. A recent survey showed that structured reporting with LI-RADS is not currently commonly used in the daily routine in Germany (e.g., only in 20.6% of participants from university hospitals), although there is a widespread wish for more widespread implementation of structured reporting [36]. Therefore, the results of our study might currently have limited applicability to non-specialized centers and radiologists. However, analogous to BI-RADS and PI-RADS, widespread implementation in the clinical routine seems possible without the need for dedicated software solutions [36]. This could directly help to improve interdisciplinary communication and clinical decision making towards an earlier start of treatment and, as a consequence, lead to better overall survival. The impact of a sub-specialization in radiology was not analyzed in this study. This was beyond the scope of this evaluation, which focused on the improvements of structured reporting compared to free-text reports in the clinical routine with a broader range of radiologists including non-specialized radiologists.

In conclusion, in line with previous studies for different tumor entities, our study demonstrates that structured reports are superior to traditional free-text reports in terms of completeness, clarity, comprehensibility, and quality. Thus, structured reporting of HCCs based on the Liver Imaging Reporting and Data System (LI-RADS) on MRI improves clinical radiological reports and could enhance multidisciplinary communication and management of patients at risk for HCC.

Clinical relevance

• Structured reports of liver MRI using LI-RADS are superior to free-text reports in patients at risk for HCC in terms of completeness and quality.

• Structured reports improve the documentation of key features for LI-RADS classification.

• Structured reporting improves interdisciplinary communication und management of patients at risk for HCC.

Conflict of Interest

Simon Lennartz: Authorship and speaker fees, Amboss GmbH

• References

• 1 Bray F, Ferlay J, Soerjomataram I. et al. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2018; 68: 394-424
• 2 European Association for the Study of the Liver. EASL Clinical Practice Guidelines: Management of hepatocellular carcinoma. J Hepatol 2018; 69: 182-236
• 3 Kim DH, Choi SH, Shim JH. et al. Magnetic Resonance Imaging for Surveillance of Hepatocellular Carcinoma: A Systematic Review and Meta-Analysis. Diagnostics 2021; 11: 1665
• 4 Leitlinienprogramm Onkologie (Deutsche Krebsgesellschaft, Deutsche Krebshilfe, AWMF). German S3 Guideline for the management of hepatocellular cancer, AWMF-Registernummer: 032–053OL (August 2023). Accessed January 26, 2024 at: https://www.leitlinienprogramm-onkologie.de/leitlinien/hcc-und-biliaere-karzinome/
• 5 Chernyak V, Fowler KJ, Kamaya A. et al. Liver Imaging Reporting and Data System (LI-RADS) Version 2018: Imaging of Hepatocellular Carcinoma in At-Risk Patients. Radiology 2018; 289: 816-830
• 6 Pinto Dos Santos D, Hempel JM, Mildenberger P. et al. Structured Reporting in Clinical Routine. Rofo 2019; 191: 33-39
• 7 Flusberg M, Ganeles J, Ekinci T. et al. Impact of a Structured Report Template on the Quality of CT and MRI Reports for Hepatocellular Carcinoma Diagnosis. J Am Coll Radiol 2017; 14: 1206-1211
• 8 Marcal LP, Fox PS, Evans DB. et al. Analysis of free-form radiology dictations for completeness and clarity for pancreatic cancer staging. Abdom Imaging 2015; 40: 2391-2397
• 9 Rosenkrantz AB, Pujara AC, Taneja SS. Use of a Quality Improvement Initiative to Achieve Consistent Reporting of Level of Suspicion for Tumor on Multiparametric Prostate MRI. AJR Am J Roentgenol 2016; 206: 1040-1044
• 10 Brook OR, Brook A, Vollmer CM. et al. Structured reporting of multiphasic CT for pancreatic cancer: potential effect on staging and surgical planning. Radiology 2015; 274: 464-472
• 11 Faggioni L, Coppola F, Ferrari R. et al. Usage of structured reporting in radiological practice: results from an Italian online survey. Eur Radiol 2017; 27: 1934-1943
• 12 Kotter E, Pinto dos Santos D. Strukturierte Befundung in der Radiologie. Radiologe 2021; 61: 979-985
• 13 European Society of Radiology (ESR). ESR paper on structured reporting in radiology. Insights Imaging 2018; 9: 1-7
• 14 Campbell I. Chi-squared and Fisher-Irwin tests of two-by-two tables with small sample recommendations. Stat Med 2007; 26: 3661-3675
• 15 Richardson JTE. The analysis of 2 × 2 contingency tables--yet again. Stat Med 2011; 30: 890
• 16 van der Pol CB, McInnes MDF, Salameh JP. et al. CT/MRI and CEUS LI-RADS Major Features Association with Hepatocellular Carcinoma: Individual Patient Data Meta-Analysis. Radiology 2022; 302: 326-335
• 17 Sahni VA, Silveira PC, Sainani NI. et al. Impact of a Structured Report Template on the Quality of MRI Reports for Rectal Cancer Staging. AJR Am J Roentgenol 2015; 205: 584-588
• 18 Nörenberg D, Sommer WH, Thasler W. et al. Structured Reporting of Rectal Magnetic Resonance Imaging in Suspected Primary Rectal Cancer: Potential Benefits for Surgical Planning and Interdisciplinary Communication. Invest Radiol 2017; 52: 232-239
• 19 Marcovici PA, Taylor GA. Journal Club: Structured radiology reports are more complete and more effective than unstructured reports. AJR Am J Roentgenol 2014; 203: 1265-1271
• 20 Bosmans JML, Neri E, Ratib O. et al. Structured reporting: a fusion reactor hungry for fuel. Insights Imaging 2015; 6: 129-132
• 21 Schwartz LH, Panicek DM, Berk AR. et al. Improving communication of diagnostic radiology findings through structured reporting. Radiology 2011; 260: 174-181
• 22 Baeßler B, Maintz D, Persigehl T. Imaging Procedures for Colorectal Cancer. Visc Med 2016; 32: 166-171
• 23 Dimarco M, Cannella R, Pellegrino S. et al. Impact of structured report on the quality of preoperative CT staging of pancreatic ductal adenocarcinoma: assessment of intra- and inter-reader variability. Abdom Radiol (NY) 2020; 45: 437-448
• 24 Spînu-Popa EV, Cioni D, Neri E. Radiology reporting in oncology—oncologists’ perspective. Cancer Imaging 2021; 21: 63
• 25 Singh R, Wilson MP, Manolea F. et al. Diagnostic accuracy and inter-reader reliability of the MRI Liver Imaging Reporting and Data System (version 2018) risk stratification and management system. SA J Radiol 2022; 26: 2386
• 26 Fraum TJ, Tsai R, Rohe E. et al. Differentiation of Hepatocellular Carcinoma from Other Hepatic Malignancies in Patients at Risk: Diagnostic Performance of the Liver Imaging Reporting and Data System Version 2014. Radiology 2018; 286: 158-172
• 27 Cerny M, Bergeron C, Billiard JS. et al. LI-RADS for MR Imaging Diagnosis of Hepatocellular Carcinoma: Performance of Major and Ancillary Features. Radiology 2018; 288: 118-128
• 28 Liu W, Qin J, Guo R. et al. Accuracy of the diagnostic evaluation of hepatocellular carcinoma with LI-RADS. Acta Radiol 2018; 59: 140-146
• 29 Liang Y, Xu F, Guo Y. et al. Diagnostic performance of LI-RADS for MRI and CT detection of HCC: A systematic review and diagnostic meta-analysis. Eur J Radiol 2021; 134
• 30 van der Pol CB, Lim CS, Sirlin CB. et al. Accuracy of the Liver Imaging Reporting and Data System in Computed Tomography and Magnetic Resonance Image Analysis of Hepatocellular Carcinoma or Overall Malignancy-A Systematic Review. Gastroenterology 2019; 156: 976-986
• 31 Kierans AS, Song C, Gavlin A. et al. Diagnostic Performance of LI-RADS Version 2018, LI-RADS Version 2017, and OPTN Criteria for Hepatocellular Carcinoma. AJR Am J Roentgenol 2020; 215: 1085-1092
• 32 Kabadi SJ, Krishnaraj A. Strategies for Improving the Value of the Radiology Report: A Retrospective Analysis of Errors in Formally Over-read Studies. J Am Coll Radiol 2017; 14: 459-466
• 33 Shenoy-Bhangle AS, Tsai LL, Masciocchi M. et al. Role of the radiologist at HCC multidisciplinary conference and use of the LR-TR algorithm for improving workflow. Abdom Radiol (NY) 2021; 46: 3558-3564
• 34 Poullos PD, Tseng JJ, Melcher ML. et al. Structured Reporting of Multiphasic CT for Hepatocellular Carcinoma: Effect on Staging and Suitability for Transplant. AJR Am J Roentgenol 2018; 210: 766-774
• 35 Reig M, Forner A, Rimola J. et al. BCLC strategy for prognosis prediction and treatment recommendation: The 2022 update. J Hepatol 2022; 76: 681-693
• 36 Nelles C, Ristow I, Juchems M. et al. Standardized Reporting of HCC with LI-RADS and mRECIST: Update on the Situation in Germany. Rofo 2024;

Publication History

Received: 26 October 2024

Accepted after revision: 28 February 2025

Article published online:
25 April 2025