Development of a classification method for mild liver fibrosis using non-contrast CT image.

PURPOSE: Detection of early-stage liver fibrosis has direct clinical implications on patient management and treatment. The aim of this paper is to develop a non-invasive, cost-effective method for classifying liver disease between "non-fibrosis" (F0) and "fibrosis" (F1-F4), and to evaluate the classification performance quantitatively. METHODS: Image data from 75 patients who underwent a simultaneous liver biopsy and non-contrast CT examination were used for this study. Non-contrast CT image texture features such as wavelet-based features, standard deviation of variance filter, and mean CT number were calculated in volumes of interest (VOIs) positioned within the liver parenchyma. In addition, a combined feature was calculated using logistic regression with L2-norm regularization to further improve fibrosis detection. Based on the final pathology from the liver biopsy, the patients were labelled either as "non-fibrosis" or "fibrosis". Receiver-operating characteristic (ROC) curve, area under the ROC curve (AUROC), specificity, sensitivity, and accuracy were determined for the algorithm to differentiate between "non-fibrosis" and "fibrosis". RESULTS: The combined feature showed the highest classification performance with an AUROC of 0.86, compared to the wavelet-based feature (AUROC, 0.76), the standard deviation of variance filter (AUROC, 0.65), and mean CT number (AUROC, 0.84). The combined feature's specificity, sensitivity, and accuracy were 0.66, 0.88, and 0.76, respectively, showing the most promising results. CONCLUSION: A new non-invasive and cost-effective method was developed to classify liver diseases between "non-fibrosis" (F0) and "fibrosis" (F1-F4). The proposed method makes it possible to detect liver fibrosis in asymptomatic patients using non-contrast CT images for better patient management and treatment.

Combined Use of Texture Features and Morphological Classification Based on Dynamic Contrast-enhanced MR Imaging: Differentiating Benign and Malignant Breast Masses with High Negative Predictive Value.

PURPOSE: We evaluated the diagnostic performance of the texture features of dynamic contrast-enhanced (DCE) MRI for breast cancer diagnosis in which the discriminator was optimized, so that the specificity was maximized via the restriction of the negative predictive value (NPV) to greater than 98%. METHODS: Histologically proven benign and malignant mass lesions of DCE MRI were enrolled retrospectively. Training and testing sets consist of 166 masses (49 benign, 117 malignant) and 50 masses (15 benign, 35 malignant), respectively. Lesions were classified via MRI review by a radiologist into 4 shape types: smooth (S-type, 34 masses in training set and 8 masses in testing set), irregular without rim-enhancement (I-type, 60 in training and 14 in testing), irregular with rim-enhancement (R-type, 56 in training and 22 in testing), and spicula (16 in training and 6 in testing). Spicula were immediately classified as malignant. For the remaining masses, 298 texture features were calculated using a parametric map of DCE MRI in 3D mass regions. Masses were classified into malignant or benign using two thresholds on a feature pair. On the training set, several feature pairs and their thresholds were selected and optimized for each mass shape type to maximize specificity with the restriction of NPV > 98%. NPV and specificity were computed using the testing set by comparison with histopathologic results and averaged on the selected feature pairs. RESULTS: In the training set, 27, 12, and 15 texture feature pairs are selected for S-type, I-type, and R-type masses, respectively, and thresholds are determined. In the testing set, average NPV and specificity using the selected texture features were 99.0% and 45.2%, respectively, compared to the NPV (85.7%) and specificity (40.0%) in visually assessed MRI category-based diagnosis. CONCLUSION: We, therefore, suggest that the NPV of our texture-based features method described performs similarly to or greater than the NPV of the MRI category-based diagnosis.

Dynamic Contrast-enhanced Area-detector CT vs Dynamic Contrast-enhanced Perfusion MRI vs FDG-PET/CT: Comparison of Utility for Quantitative Therapeutic Outcome Prediction for NSCLC Patients Undergoing Chemoradiotherapy.

PURPOSE: To directly compare the utility for therapeutic outcome prediction of dynamic first-pass contrast-enhanced (CE)-perfusion area-detector computed tomography (ADCT), MR imaging assessed with the same mathematical method and 2-[fluorine-18]-fluoro-2-deoxy-d-glucose-positron emission tomography combined with CT (PET/CT) for non-small cell lung cancer (NSCLC) patients treated with chemoradiotherapy. MATERIALS AND METHODS: Forty-three consecutive stage IIIB NSCLC patients, consisting of 25 males (mean age ± standard deviation: 66.6 ± 8.7 years) and 18 females (66.4 ± 8.2 years) underwent PET/CT, dynamic CE-perfusion ADCT and MR imaging, chemoradiotherapy, and follow-up examination. In each patient, total, pulmonary arterial, and systemic arterial perfusions were calculated from both perfusion data and SUVmax on PET/CT, assessed for each targeted lesion, and averaged to determine final values. Receiver operating characteristics analyses were performed to compare the utility for distinguishing responders from non-responders using Response Evaluation Criteria in Solid Tumor (RECIST) 1.1 criteria. Overall survival (OS) assessed with each index were compared between two groups by means of the Kaplan-Meier method followed by the log-rank test. RESULTS: Area under the curve (Az) for total perfusion on ADCT was significantly larger than that of pulmonary arterial perfusion (P < 0.05). Az of total perfusion on MR imaging was significantly larger than that of pulmonary arterial perfusion (P < 0.05). Mean OS of responder and non-responder groups were significantly different for total and systemic arterial (P < 0.05) perfusion. CONCLUSION: Dynamic first-pass CE-perfusion ADCT and MR imaging as well as PET/CT are useful for early prediction of treatment response by NSCLC patients treated with chemoradiotherapy.

Differentiating between Central Nervous System Lymphoma and High-grade Glioma Using Dynamic Susceptibility Contrast and Dynamic Contrast-enhanced MR Imaging with Histogram Analysis.

PURPOSE: We evaluated the diagnostic performance of histogram analysis of data from a combination of dynamic susceptibility contrast (DSC)-MRI and dynamic contrast-enhanced (DCE)-MRI for quantitative differentiation between central nervous system lymphoma (CNSL) and high-grade glioma (HGG), with the aim of identifying useful perfusion parameters as objective radiological markers for differentiating between them. METHODS: Eight lesions with CNSLs and 15 with HGGs who underwent MRI examination, including DCE and DSC-MRI, were enrolled in our retrospective study. DSC-MRI provides a corrected cerebral blood volume (cCBV), and DCE-MRI provides a volume transfer coefficient (Ktrans) for transfer from plasma to the extravascular extracellular space. Ktrans and cCBV were measured from a round region-of-interest in the slice of maximum size on the contrast-enhanced lesion. The differences in t values between CNSL and HGG for determining the most appropriate percentile of Ktrans and cCBV were investigated. The differences in Ktrans, cCBV, and Ktrans/cCBV between CNSL and HGG were investigated using histogram analysis. Receiver operating characteristic (ROC) analysis of Ktrans, cCBV, and Ktrans/cCBV ratio was performed. RESULTS: The 30th percentile (C30) in Ktrans and 80th percentile (C80) in cCBV were the most appropriate percentiles for distinguishing between CNSL and HGG from the differences in t values. CNSL showed significantly lower C80 cCBV, significantly higher C30 Ktrans, and significantly higher C30 Ktrans/C80 cCBV than those of HGG. In ROC analysis, C30 Ktrans/C80 cCBV had the best discriminative value for differentiating between CNSL and HGG as compared to C30 Ktrans or C80 cCBV. CONCLUSION: The combination of Ktrans by DCE-MRI and cCBV by DSC-MRI was found to reveal the characteristics of vascularity and permeability of a lesion more precisely than either Ktrans or cCBV alone. Histogram analysis of these vascular microenvironments enabled quantitative differentiation between CNSL and HGG.