Automated differentiation of malignant and benign primary solid liver lesions on MRI: an externally validated radiomics model

Abstract

Background & Aims: Distinguishing malignant from benign primary solid liver lesions is highly important for treatment planning. However, diagnosis on radiological imaging is challenging. In this study, we developed a radiomics model based on magnetic resonance imaging (MRI) to distinguish the most common malignant and benign primary solid liver lesions, and externally validated the model in two centers. Approach & Results: Datasets were retrospectively collected from three tertiary referral centers (A, B and C) including data from affiliated hospitals sent for revision. Patients with malignant (hepatocellular carcinoma and intrahepatic cholangiocarcinoma) and benign (hepatocellular adenoma and focal nodular hyperplasia) lesions were included. For each patient, only a T2-weighted MRI was included. A radiomics model was developed on dataset A using a combination of machine learning approaches, and internally evaluated on dataset A through cross-validation. Next, the model was externally validated on datasets B and C, and compared to scoring by two experienced abdominal radiologists on dataset C. In the resulting dataset, in total, 486 patients were included (A: 187, B: 98 and C: 201). Despite substantial MRI acquisition heterogeneity, the radiomics model developed on dataset A had a mean area under the receiver operating characteristic curve (AUC) of 0.78 in the internal validation on dataset A, and a similar AUC in the external validations (B: 0.74, C: 0.76). In dataset C, the two radiologists showed moderate agreement (Cohens κ: 0.61) and achieved AUCs of 0.86 and 0.82, respectively. Conclusions: Our radiomics model using T2-weighted MRI only can non-invasively distinguish malignant from benign primary solid liver lesions. External validation indicated that our model is generalizable despite substantial differences in the acquisition protocols.

Reproducible radiomics through automated machine learning validated on twelve clinical applications

“…Jacob J. Visser is a medical advisor at Contextflow. (Timbergen et al, 2020); c. primary solid liver tumors (Starmans et al, 2021b); d. gastrointestinal stromal tumors (Starmans et al, 2020b); e. colorectal liver metastases (Starmans et al, 2021a); f. melanoma (Angus et al, 2021); g. hepatocellular carcinoma (Starmans et al, 2020a); h. mesenteric fibrosis (Blazevic et al, 2021); i. prostate cancer (Castillo T et al, 2019); j. low grade glioma (van der Voort et al, 2019b); k. Alzheimer's disease (Bron et al, 2021); and l. head and neck cancer (Aerts et al, 2014).…”

“…Figure2: Examples of the 2D slices from the 3D imaging data from the twelve datasets used in this study to evaluate our WORC framework For each dataset, for one patient of each of the two classes, the 2D slice in the primary scan direction (e.g., axial) with the largest area of the segmentation is depicted; the boundary of the segmentation is projected in color on the image. The datasets included were from different clinical applications: a. lipomatous tumors(Vos et al, 2019); b. desmoid-type fibromatosis(Timbergen et al, 2020); c. primary solid liver tumors(Starmans et al, 2021b); d. gastrointestinal stromal tumors(Starmans et al, 2020b); e. colorectal liver metastases(Starmans et al, 2021a); f. melanoma(Angus et al, 2021); g. hepatocellular carcinoma(Starmans et al, 2020a); h. mesenteric …”

Differential Diagnosis and Molecular Stratification of Gastrointestinal Stromal Tumors on CT Images Using a Radiomics Approach

“…The dataset included 247 patients (125 GISTs, 122 non-GISTs) (see Table 1) and has been publicly released [36]. The dataset of 247 CT scans originated from 66 different scanners, resulting in variation in the acquisition protocols.…”

The WORC database: MRI and CT scans, segmentations, and clinical labels for 930 patients from six radiomics studies

“…The corresponding segmentation is given in the NIfTI file "segmentation.nii.gz", where a label For each dataset, for one patient of each of the two classes, the 2D slice in the primary scan direction (e.g., axial) with the largest area of the segmentation is depicted; the boundary of the segmentation is projected in color on the image. The datasets included were from different clinical applications: a. lipomatous tumors [9]; b. desmoid-type fibromatosis [10]; c. primary solid liver tumors [11]; d. gastrointestinal stromal tumors [12]; e. colorectal liver metastases [13]; and f. melanoma [14].…”

“…This dataset consists of 186 patients with either a malignant (N = 94) or benign (N = 93) primary solid liver tumor, as described in Starmans et al [11]. For each patient, a T2-weighted MRI scan is provided.…”