This retrospective study was approved by our Institutional Review Board, which waived the requirement for obtaining written informed consent.

This study included all consecutive patients who underwent abdominal dynamic contrast-enhanced CT scan to detect HCC. There was an overlap in patients between the current study and a previous study [20]. However, the theme of the current study (impact of window setting on HCC detection) was different from the previous study (impact of reconstruction algorithm on HCC detection).

Patients who underwent CT from October 2021 to March 2022 with one or more HCCs were assorted to the HCC group. Patients who had four or more HCCs were excluded because these patients are not subject to local therapies, according to the BCLC guideline [4]. There were 26 patients with 42 HCCs (14 patients with 1 lesion, 8 patients with 2 lesions, and 4 patients with 3 lesions) included. The size of the lesions was as follows: < 10 mm (15 lesions), 10–20 mm (15 lesions), and ≥ 20 mm (12 lesions). Two radiologists (X and Y with experience of 5 and 12 years in diagnostic radiology, respectively) created the standard for the diagnosis of HCC, referencing the following: histopathology diagnosis (8 lesions), follow-up examinations including MRI within 6 months (22 lesions), new or increasing in size with CT examinations within 6 months (11 lesions), and a single CT examination (i.e., index test) (1 lesion).

Nine patients without HCC on abdominal dynamic contrast-enhanced CT in February and March 2022 were randomly selected and were included to the non-HCC group. The absence of HCC was established based on the following: histopathology diagnosis who underwent liver transplantation (1 patient) and follow-up CT examinations (interval between the evaluated CT and the follow-up CT; more than 12 months [2 patients], more than 8 months [5 patients], and more than 5 months [1 patient]).

Thirty-five patients (26 patients in the HCC group and 9 patients in the non-HCC group) were assigned to the final analyses. No significant difference was observed in age, sex, hepatitis B viral status, and hepatitis C viral status between the HCC and non-HCC groups (Table 1). Figure 1 shows the patient inclusion process.

Flowchart of patient inclusion process. HCC hepatocellular carcinoma

A multidetector row CT (Aquilion ONE; Canon Medical Systems, Otawara, Japan) was used for all the patients. The following CT scanning parameters were used in their study: tube voltage, 120 kVp; helical pitch, 0.8125:1; gantry rotation time, 0.5 s; and tube current, automatic tube current modulation with standard deviation (SD) set at 13.0. Based on body weight, the concentration and dose of the contrast media were determined as follows: for those weighing < 50 kg, 300 mgI/mL and body weight × 2 mL, respectively; for those weighing between 50 and 60 kg, 350 mgI/mL and 100 mL, respectively; and for those weighing > 60 kg, 370 mgI/mL and 100 mL, respectively. Contrast media was injected via the antecubital peripheral vein in 30 s. The arterial, portal, and delayed phase images were scanned with the following delays: arterial phase, using a bolus tracking system (threshold attenuation of 200 HU in the descending aorta at the level of the diaphragm; portal phase, 40 s after arterial phase; and delayed phase, 180 s after the beginning of contrast agent injection). Images were reconstructed with DLR from the source data (AiCE body sharp standard, Canon Medical Systems). The following image reconstruction parameters were used: field of view, 35–40 cm (adjusted to body size), and slice thickness/interval, 3/3 mm.

CT images were anonymized and exported from the picture archiving and communication system in Digital Imaging and Communications in Medicine format.

All the 35 image sets were randomly ordered. To investigate the optimal WW and WL, two blinded radiologists (A and B with 12 and 9 years of imaging experience, respectively) independently evaluated image sets on OsiriX (https://www.osirix-viewer.com/). They were instructed to manually modify the WW and WL in the arterial, portal, and delayed phase images to detect HCC as clearly as possible. In this investigation, WW and WL values were not displayed on the monitor (i.e., the two radiologists modified the WW and WL without knowing the WW and WL values). The determined WW and WL settings for each patient were preserved automatically. After the completion of investigation for all the patients, the preserved WW and WL settings were reviewed, and WW and WL values were recorded. Visual evaluation described in this subsection and the following two subsections were performed on a single color monitor (DELL U2718Q; 3840 × 2160, 60 Hz, 350 cd/m2). Brightness and contrast were set at 30 and 50%, respectively.

Five other radiologists (readers 1, 2, 3, 4, and 5, with 6, 5, 4, 1, and 1 years of experience in diagnostic radiology, respectively) were involved in HCC-detection test under two different WW. WW120 was the optimal liver WW (WW = 120 HU) based on the previous subsection’s result, which was the rounded value for the average of adjusted WW by the radiologists in the arterial, portal, and delayed phases. For WW150, the conventional liver WW (WW = 150 HU) was adopted based on a previous report.9 The five readers identified HCCs and scored diagnostic confidence (4, definitely present; 3, probably present; 2, possibly present but uncertain; 1, not present) on Image J (https://imagej.nih.gov/ij/index.html). In scoring the diagnostic confidence, several image features such as early enhancement, delayed washout, size, enhancing capsule, etc. were taken into consideration comprehensively. The WW and patient information were concealed from them. They were not informed of the purpose of the study either. All the image sets (= 35 × 2) were randomly ordered. In this section, WW was fixed. However, the readers were allowed to modify WL because CT attenuation of the normal liver parenchyma can be variable based on the patient’s background liver disease (e.g., fatty liver, iron deposition, etc.) and on CT scan phase.

After the HCC-detection test, readers 1–5 were asked to evaluate image quality on Image J. They independently evaluated all the images in terms of the following using 5-point scale (5, clear depiction; 4, clearer than standard; 3, standard; 2, blurred than standard; and 1, unrecognizable): depiction of arterial phase hyper enhancement (APHE) and depiction of washout of HCC. In this part, only images in HCC group were included. All image sets (including both WW120 and WW150) were randomized before the evaluation. As for the WL, based on the previous subsection’s result and considering that the window level of 88 HU was suggested in a previous report [12], 90 HU was adopted as default WL for both the WW 120 and WW150. The five readers were also blinded to the WS.

Statistical analyses were performed using EZR version 4.0.0 (https://www.jichi.ac.jp/saitama-sct/SaitamaHP.files/statmed.html) [21], which is a graphical user interface of R version 4.2.0 (https://www.r-project.org/) (R Foundation for Statistical Computing, Vienna, Austria).

The Mann–Whitney U test and the Fisher’s exact test were used to compare the demographic and clinical characteristics between the HCC and non-HCC groups. To evaluate the diagnostic performance for detecting HCCs with the diagnostic confidence score, jackknife alternative free-response receiver operating characteristic analysis was performed with R package of “RJafroc,” and the figure of merit (FOM), which is an analog to the area under the curve in the conventional receiver operating characteristic analysis, was obtained. To analyze the sensitivity for the detection test, diagnostic confidence scores of 2 or more were considered as positive for the presence of lesions. The sensitivities were compared between WW120 and WW150 with McNemar’s test. The Wilcoxon signed-rank test was used for the comparison of image-quality scores between WW120 and WW150. For these comparisons, p < 0.05 was considered to indicate statistical significance.